Calcium/calmodulin-dependent protein kinase type IV is a target gene of the Wnt/β-catenin signaling pathway

MOLECULAR AND CELLULAR BIOLOGY, Sept. 1994, p. 6107-6116 Vol. 14, No. 90270-7306/94/$04.00+0Copyright C) 1994, American Society for Microbiology

Calcium/Calmodulin-Dependent Protein Kinase Types II and IVDifferentially Regulate CREB-Dependent Gene ExpressionRANDOLPH P. MATTHEWS,1 CHRIS R. GUTHRIE,1 LAUREN M. WAILES,1 XINYU ZHAO,I

ANTHONY R. MEANS,2 AND G. STANLEY McKNIGHTI*Department of Pharmacology, University of Washington, Seattle, Washington 98195,' and Department of

Pharmacology, Duke University Medical Center, Durham, North Carolina 277102

Received 28 February 1994/Returned for modification 18 April 1994/Accepted 20 June 1994

Phosphorylation of CREB (cyclic AMP [cAMP]- response element [CRE]-binding protein) by cAMP-dependent protein kinase (PKA) leads to the activation of many promoters containing CREs. In neurons andother cell types, CREB phosphorylation and activation of CRE-containing promoters can occur in response toelevated intracellular Ca2 . In cultured cells that normally lack this Ca2+ responsiveness, we conferCa2+-mediated activation of a CRE-containing promoter by introducing an expression vector for Ca2+/calmodulin-dependent protein kinase type IV (CaMKIV). Activation could also be mediated directly by aconstitutively active form of CaMKIV which is Ca2+ independent. The CaMKIV-mediated gene inductionrequires the activity of CREB/ATF family members but is independent of PKA activity. In contrast, transientexpression of either a constitutively active or wild-type Ca2+/calmodulin-dependent protein kinase type II(CaMKII) fails to mediate the transactivation of the same CRE-containing reporter gene. Examination of thesubcellular distribution of transiently expressed CaMKIV and CaMKII reveals that only CaMKIV enters thenucleus. Our results demonstrate that CaMKIV, which is expressed in neuronal, reproductive, and lymphoidtissues, may act as a mediator of Ca2+-dependent gene induction.

A variety of hormonal and electrophysiological stimuli caninduce changes in the intracellular levels of the second mes-sengers, Ca2+ and cyclic AMP (cAMP). Many of the down-stream effects of these second messengers depend on theactivation of specific protein kinases and the subsequentphosphorylation of regulatory proteins such as transcriptionfactors. cAMP directly leads to the dissociation of proteinkinase A (PKA) holoenzyme and release of active catalytic (C)subunit. Elevation of Ca2+ may be caused by influx of Ca2+through voltage-sensitive Ca2+ channels or by release of Ca2+from intracellular stores in response to elevated inositoltrisphosphate. Some of the effects of Ca2+ are mediatedthrough a family of protein kinases stimulated by the Ca2+/calmodulin complex. These include dedicated protein kinasessuch as myosin light-chain kinase and phosphorylase kinaseand the multifunctional Ca2+/calmodulin-dependent proteinkinases (CaM kinases) (43).The type II CaM kinase (CaMKII) isoforms are highly

expressed in the central nervous system but are also distributedin other tissues (20). CaMKIIcx self-associates into an 8- to10-mer (25), and in vitro it phosphorylates a wide variety ofsubstrates, including the transcription factors CREB (cAMPresponse element [CRE]-binding protein) (10, 47) andC/EBPI (CAAT enhancer-binding protein 1) (55) and a widevariety of proteins important in neuronal function (20).CaMKII has been implicated in many physiological phenom-ena, including regulation of the cell cycle, smooth musclecontraction, secretion, and growth factor desensitization (43).Hippocampal slices from mice made deficient in CaMKIlox bygene-targeted homologous recombination show a defect inlong-term potentiation, and the mutant mice are also deficientin certain tasks designed to test spatial memory (48, 49).

* Corresponding author. Mailing address: Department of Pharma-cology SJ-30, University of Washington, Seattle, WA 98195. Phone:(206) 543-7485. Fax: (206) 685-3822.

The type IV CaM kinase (CaMKIV) is a more recentlydiscovered member of the CaM kinases. CaMKIV exists as amonomer and is not distributed as widely as is the type IIisoform; CaMKIV mRNA levels are highest in the cerebellum,forebrain, testis, spleen, and thymus (14, 22, 29, 33). Thededuced primary sequence of CaMKIV shares only 32%identity to the CaMKIlo sequence, but the structural organi-zation of CaMKIV is similar to that of CaMKII, including anamino-terminal catalytic domain and a central calmodulinbinding regulatory domain. The carboxy-terminal domain inCaMKIV is rich in acidic residues and differs from thecorresponding carboxy-terminal association domain ofCaMKII (14, 22, 29). CaMKIV protein can be localized in thenucleus of cerebellar granule cells (21), which would allow thekinase ready access to transcription factors when activated byCa2+. The substrate specificity of CaMKIV appears to differfrom that of CaMKII; sequences from ribosomal protein S6(9), CaMKII-y (31), and Rap-lb protein (41) are phosphory-lated more effectively by CaMKIV than by CaMKIlox.CaMKIV phosphorylates the transcription factors CREB andserum response factor in vitro (9). Phosphorylation of sub-strates in vitro by CaMKIV, however, occurs at a muchreduced rate compared with phosphorylation by CaMKIIot (9,13). This may be due to the requirement for a proteinactivator; CaMKIV appears to require phosphorylation foractivity, but the site of phosphorylation and the identity of theactivating kinase remain controversial (9, 15, 24, 27, 28, 34).CaMKIV is also phosphorylated by the C subunit of PKA invitro, which leads to decreased activity (23).

There are several sites of potential interaction between theCa2' and cAMP signaling systems, including the Ca2+/cal-modulin-stimulated adenylyl cyclases (4), calmodulin-depen-dent phosphodiesterases (2), and Ca2+-activated phosphatases(6). The activity of voltage-sensitive Ca2+ channels can bedirectly regulated by PKA phosphorylation (45). One otherpossible point of convergence between the Ca2' and cAMPsignaling systems is at the phosphorylation and subsequent

6107

on Novem

ber 9, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from

6108 MATTHEWS ET AL.

activation of CREB. This bZIP transcription factor has beenshown to mediate transactivation of CRE-containing promot-ers after phosphorylation of serine 133 by PKA (17). Investi-gators have also shown that a CRE in the fos promoter canmediate the Ca2 -dependent induction offos after depolariza-tion of PC12 cells (46). Depolarization activates CaMKII (18),suggesting the possibility that CaMKII mediates the inductionof fos expression in response to depolarization of PC12 cells(46). The subsequent finding that CREB is phosphorylated invitro by CaMKII (10, 47) supports a role for CaMKII in geneinduction. In addition to being phosphorylated by CaMKII andPKA, CREB is also phosphorylated in vitro by type I (47) andtype IV (9) CaM kinase, suggesting that there are otherpossible Ca"-responsive kinases that could mediate transcrip-tional regulation of CRE-driven promoters.

Here we demonstrate the transactivation of a CRE-contain-ing reporter gene by CaMKIV in transient expression assaysand show that CaMKIV acts through CREB/ATF familymembers to activate transcription. We also demonstrate thatCaMKII is not capable of activating this CRE-containingreporter, although CaMKII does stimulate transcription fromthefos promoter. Our results suggest a possible mechanism toexplain the convergent activation of CRE-containing promot-ers by both cAMP and Ca2+ in cells that express CaMKIV.

MATERIALS AND METHODS

CaM kinase expression vectors. A 1.2-kb NcoI-ApaI frag-ment was prepared from a 15-cycle PCR using rat CaMKIVcDNA as the template, a 5' oligonucleotide primer (CGGCGACCATGGTCAAAGTCACGGTGC) that introduces anNcoI site at the initiator net and changes the next amino acidfrom Leu to Val, and a 3' oligonucleotide primer (GGCCTGGGGCCCTAAAGGAAG). CaMKIV cDNA templates in-cluded both a full-length form (CaMKIVwt) and a mutatedform (CaMKIV313) in which the codon for Gln-314 has beenmutated to a stop codon, thus terminating the protein atLeu-313 (9) and introducing an XbaI site. The 1.2-kb PCRproduct was substituted into the CEV expression vector orig-inally constructed to produce the Cao subunit of PKA (35).Expression vectors consisting of full-length and truncatedconstitutive CaMKIIal (9) driven by the metallothionein pro-moter (CaMKIIwt and CaMKII290) were constructed bysimilar methods, using CaMKIIal cDNA as the template, a 5'oligonucleotide primer (GTGCCACCATGGCTACCATCACCTGC) that introduces an NcoI site at the initiator Met andconserves the Ala at position 2, and a 3' oligonucleotide primer(GAATTCGGGCCCTCAATGGGGCAGGACGGAG).The CaMKIIa-IHG expression vector was constructed by

using a 15-cycle PCR of CaMKIIcx cDNA, the 5' primerdescribed above, and a 3' oligonucleotide primer (GAATTCGGGCCCTCAGGCGTAATCAGGGACGTCGTAAGGGTAATGGGCAGGACGGAGGGCGC) that encodes a shortpeptide epitope (Tyr-Pro-Tyr-Asp-Val-Pro-Asp-Tyr-Ala) fromhemagglutinin (IHG) recognized by monoclonal antibody12CA5 (Berkeley Antibody); the 1.4-kb NcoI-ApaI fragmentwas then substituted into the CEV expression vector (35). Forthe Ca.-TAG construct, a 1.2-kb NcoI-ApaI fragment wasprepared by using PCR of the Cat expression vector as atemplate, a 5' primer previously described (35), and a 3'oligonucleotide primer (GAATTCGGGCCCCTAGTTCAGATCCTCCTCAGAAATAAGCTTCTGCTCCATAAACTCAGTAAACTCCTTGCCACACTT) that contains a sequenceencoding a peptide epitope recognized by monoclonal anti-body AB-1 (Oncogene Science). The PCR product was thensubstituted into the CEV expression vector (35). The epitope-

tagged C subunit displays the same ability to activate geneexpression in transfected cells as the wild-type C subunit. All ofthese expression vectors are driven by the Zn2+-dependentmouse metallothionein-1 promoter and terminated with apolyadenylation signal from the human growth hormone gene.

Cell culture and transient transfections. JEG-3 cells weregrown in Dulbecco modified Eagle medium plus 10% fetalbovine serum (Life Technologies) as described previously (7)in 24-well plates until the cells were -60% confluent, at whichpoint the amount of medium was reduced to 250 pIl per well.Three to six hours later, a mixture of calcium phosphate andDNA, including 2.5 ng of a.168-luciferase (30), 50 ng of theinternal control plasmid RSV-lacZ (12), 30 ng of CaMKexpression vector, or 1 ng of Cao expression vector, plus enoughpBluescript KS(+) (Stratagene) to produce a final DNAamount of 250 ng per well, was added to the medium on thecells as previously described (36). For those experimentsexamining fos-lacZ expression, 75 ng offos-lacZ (42) and 5 ngof promoter driven Rous sarcoma virus (RSV)-luciferase (11)were added in the mixture in place of at168-luciferase andRSV-lacZ. RSV-KCREB (54), and MT-RAB (7) expressionvectors were added at various concentrations as described inthe legends to Fig. 3 and 4. To control for promoter competi-tion, we kept the amount of metallothionein promoter con-stant by adding the appropriate amount of ZEM3 (26), anempty vector containing the metallothionein promoter, tothose conditions not including the CaMK, Ca, or RAB expres-sion vector. We also controlled for promoter competition inthe RSV-KCREB experiments by adding the appropriateamounts of RSV-neo. DNA was prepared by the Qiagenprocedure (Diagen) or standard CsCl methods. After 24 h ofDNA exposure in 3% C02, the medium was replaced withmedium containing 2.5% fetal bovine serum and 80 FiMZnSO4 in Dulbecco modified Eagle medium, and the cellswere returned to incubation at 10% CO2 for 12 to18 h.For those cells to be treated with ionomycin or thapsigargin,

medium containing 2.5% fetal bovine serum and 80 FMZnSO4 remained on the cells for only 10 to 14 h, at which pointionomycin (CalBiochem) or thapsigargin (Sigma) was addeddirectly to the medium on the cells for 5 to 6 h. Theconcentrations of ionomycin and thapsigargin are indicated inFig. 2 to 4. Stock solutions of ionomycin (6 mM) and thapsi-gargin (1.54 mM) were prepared in dimethyl sulfoxide, and thefinal concentration of dimethyl sulfoxide remained constant forall drug treatments. The final concentration of CaCl2 on thecells was 1.5 mM: 1.0 mM from Dulbecco modified Eaglemedium and 0.5 mM added with the drug treatment.

Following incubation with drugs or ZnSO4-containingmedium, cells were washed once with ice-cold phosphate-buffered saline, harvested, and assayed for luciferase activityas described previously (36). 3-Galactosidase activity wasdetermined by using a modification of the Galacto-Lightsystem (Tropix). Substrate buffer consisting of 100 ,ul of 35,uM AMPGD {3-(4-methoxyspiro[1,2-dioxetane-3,2'-tricyclo[3.3.1.13'7]decan]-yl) phenyl ,B-D-galactopyranoside}, 10 mMMgCl2, and 100 mM sodium phosphate (pH 8.0), was added tocell lysates (5 ,ul) from each well in 10-s intervals; thesereactions proceeded at room temperature for 60 min. Emeraldchemiluminescence amplifier (100 p.1) was injected into thetubes containing the cell lysates by a Berthold luminometer,which determined the amount of luminescence producedduring a 5-s period. Luciferase activity was divided by this,-galactosidase activity to normalize for differences in trans-fection efficiency, except for those experiments examiningfos-lacZ expression, in which case the reciprocal ratio was used.

MOL. CELL. BIOL.

on Novem

ber 9, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from

CaM KINASE IV ACTIVATES A CREB-DEPENDENT REPORTER 6109

All transfection experiments presented were performed atleast three times with consistently reproducible results.

In vitro phosphorylation of CREB. We generated CREBS133A by using a CREB expression vector as the template andstandard in vitro mutagenesis techniques. Phosphorylationreactions with CREB (17) or CREB S133A were performedwith purified recombinant CaMKIV313 described previously(9). Each reaction mixture contained 1.75 ,uM CREB orCREB S133A in a final volume of 50 ,ul of 50 mM HEPES(N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid; pH 8.0)-0.5 mM dithiothreitol-10 mM magnesium acetate-100 ,uMATP-1 mM CaCl2 or 5 mM EGTA-60 ng of CaMKIV313. Thespecific activity of [y-32P]ATP was 4,200 cpmlpmol. The reactionswere initiated by addition of kinase, the mixtures were incubatedat 30°C for the time specified in the legend to Fig. 5, and thereactions were terminated by the addition of 4x sample bufferand analyzed on 12.5% polyacrylamide gels.

Immunocytochemistry. Cells for immunocytochemical stain-ing were maintained and transfected as described above. TheRIIo expression vector has been described previously (37).Following incubation with 80 ,uM Zn2+, cells were washedonce with cold methanol and then fixed for 15 min at -20°C inmethanol. Following fixation, cells were washed in phosphate-buffered saline and incubated overnight at 4°C with blockingbuffer (5% bovine serum albumin, 10% horse or goat serum).After removal of blocking buffer, diluted primary antibody wasincubated with the cells for 1 h at room temperature inblocking buffer. Monoclonal antibody AB-1 (Oncogene Sci-ence) was incubated on the cells at a dilution of 1:100, 12CA5ascites fluid (Berkeley Antibody) was used at a dilution of1:400, or anti-CaMKIV antiserum (28) was used at a dilutionof 1:25. Following primary antibody incubation, cells werewashed with phosphate-buffered saline. For cells treated withantibody AB-1 or 12CA5, biotinylated horse anti-mouse affin-ity-purified immunoglobulin G (Vector Laboratories) was thenadded at a dilution of 1:100 in blocking buffer for 1 h. For cellstreated with anti-CaMKIV antiserum, biotinylated goat anti-rabbit affinity-purified immunoglobulin G (Vector Laborato-ries) was added at a dilution of 1:100 in blocking buffer.Following another set of washes, cells were incubated withAvidin D-FITC (Vector Laboratories) diluted 1:4,000 inHEPES-buffered saline (pH 8.2) for 30 min at room temper-ature. Cells were washed three times as before, then stainedwith 5 ,uM propidium iodide (Molecular Probes) in deionizedwater for 20 min, and washed once with deionized water for 1min. Cells were then mounted under coverslips with Vecta-Shield (Vector Laboratories), wells were punched out with a1-cm diameter cork borer, and the cells were mounted onslides. Stained cells were then examined in a Bio-Rad MRC600 confocal microscope with a focal plane thickness of 1 ,um.Images were prepared and labeled with Adobe Photoshopversion 2.5.

RESULTS

CaMKIV transactivates a CRE-containing reporter gene.We designed expression vectors containing CaMKII or CaMKIV cDNA, downstream of the metallothionein promoter, asdescribed in Materials and Methods. These constructs containcDNA clones that produce either a full-length (CaMKIVwt) ora truncated (CaMKIV313) version of CaMKIV (Fig. 1A); theCaMKII constructs that produce full-length (CaMKIIwt) andtruncated (CaMKII290) forms of CaMKII are similar (notshown). Truncation. of CaMKIV at leucine 313 (9) and ofCaMKII at leucine 290 (8) has been shown to producecalcium-independent activity of these enzymes. Figure 1B

A0(4

4Z..

I IIMT promoter I aM Kinase IV Codinc Re ionI hGH [olvA

CL,IT

B

.->c04.-C):0C

V-ac

4-a

04-J*5

0.CL

aL)ol

12

10

8-

4-

2

* Reporter alone3I CaMKllwtI CaMKII290l CaMKIVwt

CaMKIV313

al 68-luciferase

I

- = 250 bp

T

fos-lacZ

ReporterFIG. 1. Abilities of various CaM kinases to induce a168-luciferase

and fos-lacZ in JEG-3 cells. (A) Coding regions for CaMKIVwt andCaMKIV313 are downstream of the metallothionein (MT) promoterand are followed by the human growth hormone (hGH) polyadenyla-tion signal. The termination codon and XbaI site unique to theCaMKIV313 construct are noted in parentheses. (B) JEG-3 cells weretransiently transfected with, per well, either (i) 30 ng of CaMKIIwt(vertically hatched columns), CaMKII290 (speckled columns), CaMKIVwt (diagonally hatched columns), or CaMKIV313 (open columns)expression vector and 2.5 ng of a168-luciferase reporter or (ii) 30 ng ofCaMK expression vector and 75 ng of fos-lacZ reporter. After expo-sure to DNA precipitates for 24 h, medium was replaced and cells weretreated with medium containing 80 ,uM Zn2+ for 16 h before harvest-ing. Extracts were analyzed for luciferase and P-galactosidase activityas described in Materials and Methods. The activity of the inducibleconstruct was divided by the activity from the constitutive construct,and the activity ratio for reporter alone (closed columns) was normal-ized to 1. The transfections were performed in triplicate, and the mean± standard deviation is shown for each datum point.

displays the transactivation of two different reporter con-structs, al168-luciferase and fos-lacZ, by the CaMK expressionvectors in transient transfections of JEG-3 cells. The al168-luciferase construct consists of the 168-bp promoter of theglycoprotein hormone a subunit, containing two tandem con-sensus CREs upstream of firefly luciferase cDNA (30). Thefos-lacZ reporter construct contains the 611-bp fos promoter,including the Ca/CRE site, serum response element, andsis-inducible element, upstream of a fusion gene that encodesthe 315 amino-terminal amino acids from Fos and 1,015carboxy-terminal amino acids from ,B-galactosidase (42).CaMKIV313 is the only CaMK expression vector that trans-activates at168-luciferase, while fos-lacZ is transactivated byeither CaMKII290 or, to a lesser extent, CaMKIV313 (Fig.1B). Expression of the full-length form of CaMKII or

VOL. 14, 1994

on Novem

ber 9, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from

6110 MATFHEWS ET AL.

A5001

:.20)

0)

C-

.H

:i.2'-

CY)

[lonomycin], pM

B

[Thapsigargin], nM

FIG. 2. lonomycin and thapsigargin induce a CRE-containing re-

porter in CaMKIVwt-transfected JEG-3 cells. Cells were transfectedwith, per well, 30 ng of CaMKIVwt expression vector and 2.5 ngaL168-luciferase (triangles), 30 ng of CaMKIIwt expression vector and2.5 ng of cL168-luciferase (squares in panel A), or 2.5 ng of (x168-luciferase alone (circles) and then treated with medium containing 80p.M Zn2+ for 14 h. (A) Following Zn2+ treatment, cells were incubatedwith the indicated concentrations of ionomycin and 1.5 mM CaCl2 for5 h. (B) Following Zn2+ treatment, cells were incubated with theindicated concentrations of thapsigargin and 1.5 mM CaCl2 for 5 h,when cells were harvested and assayed as for Fig. 1. Both sets oftransfections were performed in triplicate, and the mean + standarddeviation is shown for each datum point. Datum points are presentedas a ratio of luciferase (luc) activity to ,B-galactosidase (lgal) activity,resulting from expression of the internal control plasmid, RSV-lacZ.

CaMKIV does not transactivate expression of oU68-luciferasein the absence of agents that increase intracellular calcium,although we see a small increase in fos-lacZ expression inresponse to the wild-type CaMKs (Fig. 1B).CaMKIV confers Ca2+ sensitivity to JEG-3 cells. When

JEG-3 cells are transiently transfected with CaMKIVwt, sub-sequent treatment with the calcium ionophore ionomycinresults in transactivation of ot168-luciferase expression (Fig.2A). At 3 ,uM ionomycin, CaMKIVwt stimulation approachesthe level of stimulation seen with the constitutive CaMKIV313(Fig. 2A). lonomycin has very little effect on aL168-luciferaseactivity in control cells not transfected with a CaMK expressionvector. There is also no change in oa168-luciferase expressionwhen cells are cotransfected with CaMKIIwt and treated withionomycin (Fig. 2A).

Thapsigargin inhibits reuptake of Ca2+ into intracellularstores and leads to an elevation of intracellular calcium (52). InJEG-3 cells that have been transiently transfected withCaMKIVwt, thapsigargin (3 nM) produces transactivation ofoL168-luciferase activity to approximately the same fold induc-tion as seen with the constitutive CaMKIV313 (Fig. 2B).

Thapsigargin did not induce ot168-luciferase expression in theabsence of CaMKIVwt (Fig. 2B). Analogous cotransfectionexperiments with the CaMKIIwt expression vector produce noincrease in ot168-luciferase expression when cells are treatedwith thapsigargin (data not shown).CaMKIV-induced CRE transactivation does not require

PKA. Since the effects of CaMKIV on gene expression couldbe directly or indirectly influenced by PKA, we used a domi-nant negative PKA mutant (RAB) in cotransfection experi-ments to examine the role of PKA in the CaMKIV response.Incorporation of this mutant form of the PKA Rlh subunit intothe PKA tetramer inactivates the C subunit and preventsthe dissociation of C and R subunits in the presence of cAMP(5, 7). Increasing amounts of RAB decrease the basal expres-sion of oc168-luciferase but do not reduce the induction byCaMKIV313 (Fig. 3A). In contrast, Fig. 3B demonstrates thattransactivation of cx168-luciferase expression by cotransfectedPKA Cot subunit is completely inhibited as the amount of RABexpression vector increases. In Fig. 3C, the data from Fig. 3Aand B are depicted after elimination of background ot168-luciferase expression so that the otl68-luciferase inductionproduced by CoL and CaMKIV313 may be directly compared.Although we observe a small decrease in CaMKIV313-medi-ated cx168-luciferase expression by RAB, the same amount ofRAB results in the complete inhibition of Cot-mediated (x168-luciferase expression. These results indicate that the inductionof aL168-luciferase by CaMKIV313 does not require PKAactivity.To test whether the Ca2+ response observed in cells cotrans-

fected with CaMKIVwt is also independent of PKA, weperformed experiments similar to those described above, usingan amount of RAB vector (25 ng) sufficient to completelyinactivate endogenous PKA activity. JEG-3 cells that havebeen cotransfected with RAB and CaMKIVwt respond toionomycin treatment with a fivefold stimulation of ox168-luciferase expression (Fig. 3D). There is no stimulation ofot168-luciferase expression by ionomycin in cells that have notbeen transfected with the CaMKIV expression vector. Thus,we continue to see a Ca2+-stimulated response in CaMKIV-transfected cells in which PKA has been inactivated.

Inhibition of CaMKIV-induced transactivation by a domi-nant negative CREB mutant. We cotransfected JEG-3 cellswith an expression vector for the dominant negative CREBmutant KCREB (3, 54) to examine the dependence of thetransactivation of cx168-luciferase by CaMKIV on endogenousCREB. KCREB contains a mutation (Arg-287->Leu) in theDNA-binding domain of CREB, which allows KCREB toheterodimerize with CREB and other members of the CREB/ATF family that heterodimerize with wild-type CREB (such asATF-1 or CREM) and reduces the ability of these transcrip-tion factors to interact with DNA (54). CaMKIV313-mediatedot168-luciferase expression is completely inhibited by as little as1 ng of KCREB expression vector per well (Fig. 4A). Asexpected, PKA-mediated aL168-luciferase expression is alsoinhibited by KCREB expression (Fig. 4B), which supportsprevious studies demonstrating the importance of CREB in theactivation of gene transcription by PKA (17). The stimulationof oc168-luciferase expression produced by 1 F.M ionomycintreatment of CaMKIVwt-transfected JEG-3 cells is attenuatedby KCREB coexpression (Fig. 4C) to a similar degree as isCaMKIV313-mediated transactivation (Fig. 4A). The effect ofKCREB on CaMKIV- and Cot-stimulated activity may bedirectly compared in Fig. 4D, which displays the data from Fig.4A to C as a percentage of maximal induction. As shown in Fig.4D, both CaMKIV- and Cot-mediated inductions of ox168-luciferase are inhibited in parallel by KCREB. Neither

MOL. CELL. BIOL.

on Novem

ber 9, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from

CaM KINASE IV ACTIVATES A CREB-DEPENDENT REPORTER

zooo BA^ CaMKIV313 0 Ca

a 1000

0- -~~~~~S- TS; -An.o 2.5 5.0 7.5 25.1

MT-RAB, ng

ACa

A

0

A2-)u

0)'a

,

O.o 2.5 5.0 7.5 25.0

MT-RAB, ng

0 0.1 1 10

MT-RAB, ng [lonomycin], pM

FIG. 3. CaMKIV-induced a168-luciferase expression does not require active PKA. (A) JEG-3 cells were transfected with the indicated amountsof RAB plus, per well, 30 ng of CaMKIV313 and 2.5 ng of a168-luciferase (triangles) or 2.5 ng of a168-luciferase alone (filled circles). After 24h of DNA exposure, cells were treated with medium containing 80 ,uM Zn2+ for 16 h before assay. (B) JEG-3 cells were transfected with theindicated amounts of RAB plus, per well, 1 ng of Ca and 2.5 ng of a168-luciferase (open squares) or 2.5 ng of a168-luciferase alone (filled circles).After 24 h of DNA exposure, cells were treated with medium containing 80 ,uM Zn2+ for 16 h, lysed, and assayed. (C) The background (DKgd)control activity was subtracted from each point in panels A (triangles) and B (open squares), and the absolute activity at 0 ng of RAB per well wasset at 100%. (D) JEG-3 cells were transfected with, per well, 25 ng of RAB plus 30 ng of CaMKIVwt and 2.5 ng of al68-luciferase (triangles) or

2.5 ng of a168-luciferase alone (circles). After 24 h of DNA exposure, cells were treated with medium containing 80 p.M Zn2+ for 14 h, at whichpoint cells were incubated with medium containing the indicated concentrations of ionomycin and 1.5 mM CaCl2 for 5 h. Cells were lysed andassayed for luciferase and P-galactosidase activity. The transfections were performed in triplicate, and the mean + standard deviation is shown foreach datum point. Datum points in panels A, B, and D are presented as a ratio of luciferase (luc) activity to ,B-galactosidase (Pgal) activity, resultingfrom expression of the internal control plasmid, RSV-lacZ. MT, metallothionein-1 promoter.

KCREB nor RAB coexpression significantly decreased theactivity of the constitutively active construct, RSV-lacZ, whichis cotransfected with a168-luciferase in all of our studies tocontrol for transfection efficiency.

In all of these experiments, the basal expression of a168-luciferase is also decreased by KCREB (Fig. 4), suggesting thatthis basal activity is CREB dependent. Given the similarreduction seen with RAB cotransfection (Fig. 3), we concludethat endogenous PKA activity accounts for much of the basaltranscription. The difference in the effects of RAB and KCREBinhibition is most clearly demonstrated by comparing Fig. 3Cand 4D, which display the data so that the effect of backgroundexpression is eliminated. KCREB and RAB overexpressioneffectively inhibit Ca-mediated transactivation (Figs. 3C and4D), while only KCREB overexpression inhibits CaMKIV-mediated transactivation of a168-luciferase (Fig. 4D). Thisfinding suggests that the activity of CREB/ATF family mem-bers, but not PKA activity, is necessary for the CaMKIV-mediated stimulation of a168-luciferase.CaMKIV phosphorylates CREB exclusively at serine 133 in

vitro. The site of CREB phosphorylation by CaMKIV in vitrowas examined by using recombinant CaMKIV313 to phosphor-ylate wild-type CREB and CREB S133A. As depicted in Fig. 5,purified CaMKIV313 phosphorylates wild-type CREB to amaximum level of one incorporated phosphate per CREBmolecule, reaching this maximum after 6 to 8 min. CREB

S133A is not appreciably phosphorylated by CaMKIV (Fig. 5).The inset displays an autoradiograph depicting the phosphor-ylation of wild-type CREB by recombinant CaMKIV313 afteran 8-min incubation at 30°C with or without Ca2", demonstrat-ing that the constitutive CaMKIV is active in the absence ofCa2+ and that CaMKIV313 does not transfer detectablephosphate to CREB S133A. We conclude from these data thatCaMKIV stoichiometrically phosphorylates Ser-133 on CREBand does not recognize additional sites.

Subcellular localization of transiently expressed CaMKII,CaMKIV, and Ca. We constructed expression vectors encod-ing CaMKIIa-IHG and Ca-TAG, which contain sequencesencoding peptide epitopes, as described in Materials andMethods. We investigated the subcellular localization ofCaMKIIa-IHG, CaMKIV, and Ca-TAG in transient transfec-tions of JEG-3 cells to determine whether the transientlyexpressed kinases were able to gain access to the nucleus. Thelocalization of transiently expressed CaMKIV was determinedby using an antibody that recognizes the central regulatorydomain (29); nuclei were visualized by using propidium iodideas described in Materials and Methods. Confocal images thatoptically section through the upper half of CaMKIV-trans-fected cells detect CaMKIV in the nucleus, although appar-ently excluded from nucleoli (Fig. 6A and B), while thosethrough the bottom of the same cells display staining forCaMKIV that appears to be either primarily cytoplasmic (Fig.

A

0)U

0)

.0-. E*t E0oCuE~0a)0

CIA r2MKIV313

.0

120

iool80'

60

40*

20

VOL. 14, 1994 6111

on Novem

ber 9, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from

6112 MATTHEWS ET AL.

CaMKIV31o3|-Control

) 2 4 6RSV-KCREB, ng

C

8000>.

*6000:-)

D) 4000

'2)-2000.

8 10

A CaMKIVwt + lono0 CaMKIVwt + DMSO

0 2 4 6

RSV-KCREB, ng

8 10

BCao

| Control|

C0

2 4 6RSV-KCREB, ng

8 10

A CaMKIV3130 CaA CaMKIVwt + lono

4 6

RSV-KCREB, ng

FIG. 4. CaMKIV-induced ox168-luciferase expression is inhibited by a dominant negative CREB mutant. (A) JEG-3 cells were transfected withthe indicated amounts of KCREB plus, per well, 30 ng of CaMKIV313 and 2.5 ng of ot168-luciferase (filled triangles) or 2.5 ng of at168-luciferasealone (filled circles). After 24 h of DNA exposure, cells were treated with medium containing 80 FM Zn2+ for 16 h, at which point cells were lysedand assayed. (B) JEG-3 cells were transfected with the indicated amounts of KCREB plus, per well, 1 ng of Co and 2.5 ng of ao168-luciferase (opensquares) or 2.5 ng of ao168-luciferase alone (filled circles) and were then treated and assayed, as for panel A. (C) JEG-3 cells were transfected withthe indicated amounts of KCREB plus 30 ng of CaMKIVwt and 2.5 ng of ox168-luciferase per well. After 24 h of DNA exposure, cells were treatedwith medium containing 80 p.M Zn2+ for 14 h, at which point cells were incubated with medium containing 1 ,uM ionomycin (lono) and 1.5 mMCaCl2 for 5 h (open triangles) or with the same amount of dimethyl sulfoxide (DMSO) (open circles). The transfections were performed intriplicate, and the mean ± standard deviation is shown for each datum point. In all experiments, the amount of RSV promoter was kept constantby the addition of RSV-neo so that all conditions included 10 ng of RSV per well in addition to 50 ng of RSV-lacZ per well. Datum points are

presented as a ratio of luciferase (luc) activity to 1-galactosidase (Igal) activity, resulting from expression of the internal control plasmid,RSV-lacZ. (D) The background (bkgd) control activity was subtracted from each point in panels A (filled triangles), B (open squares), and C (opentriangles), and the absolute activity at 0 ng of KCREB per well was set at 100%.

6C) or in both nuclear and cytoplasmic compartments (Fig.6D). Figures 6E and F depict JEG-3 cells transiently trans-fected with the CaMKIIax-IHG expression vector and stainedwith anti-IHG antibody and propidium iodide as described inMaterials and Methods. In contrast to the cells depicted inFigs. 6A to D, the CaMKIIcx-IHG-transfected cells displayCaMKIIot-IHG staining exclusively in the cytoplasm (Fig. 6Eand F), throughout all cellular focal planes (not shown). Bycomparison, Ca-TAG is localized entirely in the cytoplasmwhen Rlla is coexpressed (Fig. 6G), forming the inactiveholoenzyme. When Ca.-TAG is expressed without RIloa, thestaining for Cat-TAG is found in both the nucleus and thecytoplasm (Fig. 6H). These results indicate that transientlyexpressed CaMKIV and Cat-TAG may be localized to thenucleus in conditions that correlate with the activation ofot168-luciferase expression described above. Transiently ex-

pressed full-length CaMKIIa-IHG is excluded from the nu-cleus.

DISCUSSION

The transient expression of CaMKIV in JEG-3 cells conferson these cells the ability to induce a CRE-containing reporter,

a168-luciferase. This transactivation is elicited directly by a

truncated, constitutively active form of CaMKIV or by wild-type CaMKIV if cells have also been treated with ionomycin or

thapsigargin to elevate intracellular calcium. The stimulationof a.168-luciferase by CaMKIV appears to be independent ofendogenous PKA activity, since we have shown that transientcoexpression of a dominant inhibitory R subunit (RAB) doesnot block transactivation by CaMKIV. The induction of o168-luciferase by CaMKIV does appear to be dependent on CREBor other members of the CREB/ATF family such as ATF-1 or

CREM, however, since coexpression of a dominant negativeCREB inhibits CaMKIV stimulation of ac168-luciferase. Fur-thermore, using purified proteins, we have shown thatCaMKIV stoichiometrically phosphorylates a single residue onCREB, serine 133, which corresponds to the residue previouslyshown to be necessary for CREB activation by PKA in vivo(17). These results demonstrate that CaMKIV is capable ofmediating a Ca2"-dependent activation of CRE-containingpromoters.

In contrast to CaMKIV, transient expression of CaMKIIcxdoes not mediate the activation of oa168-luciferase in JEG-3cells. We have shown that neither the constitutively active formof CaMKIIlo nor the wild-type form, in the presence of agents

A5000

4000'

3000'

2000

10004

C-)

c,)U-FCY

.5

-)

I(

MOL. CELL. BIOL.

on Novem

ber 9, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from

CaM KINASE IV ACTIVATES A CREB-DEPENDENT REPORTER 6113

2.0

LU

u.5

a

0

EC4

caL

r_rn

1.5-

0.O

0

-Ca2+ +Ca2+< <

Ln c

D an m anLU LU LU CU

_ _ - 43 kDa

CREB (wt)

D- 0

CREB SI 33A

_~~~~~~ 01~~~~.e

n1) -2 4 6 8 10

Time in minutes

FIG. 5. CaMKIV phosphorylates CREB at serine 133 in vitro.Recombinant CaMKIV313 was incubated with wild-type (wt) CREBor CREB S133A at 30°C for the times indicated as described inMaterials and Methods. Data are depicted as a stoichiometric ratio ofphosphate incorporated per CREB molecule. The inset displays an

autoradiogram of the polyacrylamide gel electrophoresis analysis ofCaMKIV313 phosphorylation reactions with CREB or CREB S133Awith or without Ca2+ after an 8-min incubation at 30°C.

that increase intracellular calcium, activates al68-luciferase.The cells are expressing CaMKIIa protein, as demonstrated byimmunocytochemistry, and the constitutive CaMKII290 is bi-ologically active and capable of activating a fos-lacZ reportergene, presumably through a nuclear event. This activation offos-lacZ by CaMKII290 is not dependent on either PKA or

CREB activity (19), implying that CaMKII targets a transcrip-tion factor other than CREB in the activation of the fospromoter. Other reports have suggested that CaMKII may beresponsible for the calcium-mediated activation of CRE-mod-ulated genes in PC12 cells and in hippocampal neurons (10,47). Our results suggest that CaMKIV may be a more likelymediator of Ca2+-dependent gene activation involving CRE-containing promoters.The activation of CRE-containing promoters by phosphor-

ylation of CREB requires that the kinase have access to

CREB, which is localized in the nucleus (53). CaMKIV hasbeen shown previously to be localized to the nucleus byelectron microscopy (21); our results indicate that in JEG-3cells, expressed CaMKIV is detectable not only in the nucleusbut in the cytoplasm as well. This distribution favoring thecytoplasm may be cell type specific or influenced by therelatively high expression levels seen in transient transfections.The nonuniform distribution of CaMKIV within the nucleus as

visualized at different focal planes may indicate the presence ofspecific nuclear binding sites. Our results clearly indicate,however, that CaMKIV is able to reach the nuclear compart-

ment where phosphorylation of CREB is thought to occur. Acomparison of the apparent ratio of nuclear to cytosolicCaMKIV with that of the relative distribution of the PKA Casubunit between nucleus and cytosol indicates that the relativeamount of CaMKIV in the nucleus is less than that of the Csubunit, suggesting that most of the transfected CaMKIVremains in the cytosol. This difference in localization may limitaccess of CaMKIV to CREB and therefore may partially

explain the reduced effectiveness of CaMKIV in gene induc-tion relative to PKA.

In contrast to CaMKIV, staining of cells transfected withIHG-tagged full-length CaMKIIa indicates that this protein isfound only in the cytoplasm, in agreement with previousstudies examining the localization of CaMKII in neurons (38).Recently, however, alternatively spliced isoforms of theCaMKII family that contain nuclear translocation signals havebeen discovered (44). In our transactivation studies, we usedthe truncated CaMKII290, because this protein is both consti-tutively active and unlikely to be excluded from the nucleus.Overexpression of CaMKII290 has been shown previouslyto cause arrest of the cell cycle in G2 in C127 cells (40). Slight-ly higher molecular weight truncations (positions 1 to 326)of CaMKII are able to enter the nucleus, as determined bykinase assays of nuclear preparations (44). Our results forCaMKII290 suggest that even if the full-length CaMKIIlo wereable to translocate into the nucleus, it would not inducetranscription of CRE-driven promoters when activated byCa2+/calmodulin.The inability of the constitutively active CaMKII290 to

transactivate a CREB-dependent reporter such as ax168-lucif-erase was surprising, especially since CaMKIV313 is an activeinducer. Both kinases phosphorylate CREB on Ser-133 invitro, but CaMKII has also been shown to phosphorylate otherunidentified sites on CREB as well (10, 47). Recent work bySun et al. (51) demonstrates that CaMKII phosphorylatesCREB on Ser-142 as well as Ser-133 and that phosphoryla-tion of CREB at Ser-142 inhibits the ability of CREB toactivate transcription. These investigators and others alsodemonstrate that CaMKIV is an effective inducer of CREB-dependent transcription in GH3 pituitary cells, as determinedby CaMKIV-mediated stimulation of GAL4 constructs byGAL4-CREB (13, 51), which supports our results for JEG-3cells.The magnitude of a168-luciferase stimulation by PKA is

much higher than that with CaMKIV under similar conditions.The induction of a168-luciferase by CaMKIV ranges from 3-to 5-fold, while the maximal C-induced stimulation is as high as50-fold. The expression vectors used are identical except forthe coding region, but we cannot directly compare the levels ofexpressed protein in our immunocytochemistry studies sincedifferent antibodies were used for detection. Recently, in vitroanalysis of CREB phosphorylation by CaMKIV and PKA hasrevealed that although the Kms of CaMKIV and PKA forCREB are similar, the VmS, of CREB phosphorylation by PKAis 30-fold greater than the CREB phosphorylation Vmax forCaMKIV (13). Certainly such a difference in the kinetics of invivo phosphorylation in our experiments on JEG-3 cells may beresponsible for the different degrees of stimulation byCaMKIV and PKA. Partial exclusion of CaMKIV from thenucleus, due to limited diffusion of CaMKIV into the nucleusor a rate-limiting step in nuclear transport, may also explainthe reduced activity of CaMKIV as a transcriptional activator.Others have suggested that CaMKIV must be phosphorylatedfor maximal activity and that this phosphorylation requiresCa2+/calmodulin and a protein activator that may also be aprotein kinase (34). This phosphorylation appears to be onsites near the amino terminus (27) and would therefore bepresent in both CaMKIVwt and CaMKIV313. If JEG-3 cellsdo not contain the putative CaMKIV activator, this couldexplain the lower apparent activity of CaMKIV relative to Ccxin our transfection assays.Our results help clarify several issues regarding the ability of

Ca2+ to stimulate gene transcription. We have demonstrated aremarkable specificity in the ability of the type II and type IV

VOL. 14, 1994

n r

on Novem

ber 9, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from

6114 MATTHEWS ET AL.

CaMK-IV

CaMK-II_ __I . _

._

._

._* _. ____,_

_E_| ___I ____I _

___-__

PKA

FIG. 6. Subcellular localization of expressed CaMKIV, CaMKII, and the PKA Cx subunit. JEG-3 cells were transiently transfected with, perwell, 30 ng of CaMKIVwt expression vector (A to D), 30 ng of CaMKIIcx-IHG expression vector (E and F), 2.5 ng of Ca-TAG and 25 ng of RIlaoexpression vectors (G), or 2.5 ng of Ca-TAG expression vector alone (H) and then subjected to Zn2+ treatment and fluorescent staining asdescribed in Materials and Methods. Cells were stained with anti-CaMKIV antiserum (A to D, left panels), anti-IHG antibody 12CA5 (E and F,left panels), anti-TAG antibody ABI (G and H, left panels), and propidium iodide (A to H, right panels). Panels A and B display 1-,um confocaloptical slices though the upper half of cells, while panel C and D represent 1-p.m optical slices through the lower half of the same cells. PanelsE to H are 1-p.m optical slices taken through the lower half of cells.

MOL. CELL. BIOL.

on Novem

ber 9, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from

CaM KINASE IV ACTIVATES A CREB-DEPENDENT REPORTER 6115

CaM kinases to induce CRE-driven gene expression in tran-sient transfection assays and conclude that only the type IVkinase has this ability. Although we have yet to determinewhether CaMKIV stimulates Ca2"-mediated CRE-drivengene expression in actual physiological processes, we believethat the ability of CaMKIV to elicit Ca2"-mediated geneactivation in our transfection studies strongly suggests thatCaMKIV is a likely player in Ca2"-mediated gene activation invivo. Other potential multifunctional Ca2+/calmodulin-depen-dent protein kinases, such as CaMKI (32, 39) and otherisoforms of CaMKII (43), may activate CRE-mediated tran-scription through Ca2+, although their ability to do so remainsto be examined. The overall extent of CRE-dependent geneinduction in response to elevated Ca2+ may depend on whichtypes of CaM kinases are expressed in specific cell types.We have also shown that the more complex fos promoter

responds to both CaMKII290 and CaMKIV313. Preliminaryevidence suggests that the CaMKII-mediated induction of thefos promoter does not depend on CREB but does require theupstream serum response element (19). The response of thefos promoter to CaMKIV may be regulated through the severalCREs that are contained within the promoter, including theone also termed the calcium response element at nucleotide-60 (46). fos promoter induction has been studied in hip-pocampal neurons in response to Ca2' entry through eitherglutamate receptors or L-type Ca2+ channels. Constructscontaining the CRE were induced by Ca2' entry throughL-type Ca2+ channels but not by glutamate receptor activation,whereas similar constructs containing the serum responseelement regained the ability to respond to glutamate receptoractivation (1). The hippocampal neurons used in these studiesmay express isoforms of CaMKII as well as CaMKIV; ourresults would suggest that they could play distinct roles in theinduction pathway.We must also consider the potential role of PKA in the

response to intracellular Ca2 , since many neurons expressCa2+/calmodulin-sensitive adenylate cyclases (4) which havebeen shown to become activated in neurons after depolariza-tion (50). Interestingly, the induction of fos in response todepolarization is strongly inhibited if PC12 cells are deficient inPKA activity (16). Thus, the PKA system may also becomeactivated by elevated Ca2+ in certain cell types, which wouldthen lead to an increase in CRE-driven gene expression similarto the increase produced by CaMKIV activation. Although our

experiments demonstrate that Ca2+, acting through CaMKIV,can induce a CRE-driven promoter in the absence of func-tional PKA, it seems very likely that both kinase systemsinteract with each other in the regulation of gene expressionunder physiological conditions.

ACKNOWLEDGMENTS

We thank Thong Su and Mona Belyamani for expert technicalassistance and Rejean Idzerda for helpful discussion and comments on

the manuscript. The fos-lacZ construct was kindly provided by TomCurran. We thank Richard Goodman for the RSV-KCREB andRSV-CREB vectors and Marc Montminy for recombinant CREB. Weespecially thank Richard Maurer and Thomas Soderling and themembers of their laboratories for sharing their data with us prior topublication.

This work was supported by grants GM32875 (G.S.M.), HD12629(G.S.M.), and HD07503 (A.R.M.) from the National Institutes ofHealth. Training grant support for R.P.M. (T32 GM07266), C.R.G.(HL07312-15), L.M.W. (T32 GM07270), and X.Z. (Merck fellowship)is also acknowledged.

REFERENCES1. Bading, H., D. D. Ginty, and M. E. Greenberg. 1993. Regulation of

gene expression in hippocampal neurons by distinct calciumsignaling pathways. Science 260:181-186.

2. Bentley, J. K., and J. A. Beavo. 1992. Regulation and function ofcyclic nucleotides. Curr. Opin. Cell Biol. 4:233-240.

3. Boutillier, A.-L., F. Barthel, J. L. Roberts, and J.-P. Loeffler. 1992.,-Adrenergic stimulation of cFOS by protein kinase A is mediatedby cAMP regulatory element binding protein (CREB)-dependentand tissue-specific CREB-independent mechanisms in cortico-trope cells. J. Biol. Chem. 267:23520-23526.

4. Choi, E. J., Z. Xia, E. C. Villacres, and D. R. Storm. 1993. Theregulatory diversity of mammalian adenylyl cyclases. Curr. Opin.Cell Biol. 5:269-273.

5. Clegg, C. H., L. A. Correll, G. C. Cadd, and G. S. McKnight. 1987.Inhibition of intracellular cAMP-dependent protein kinase usingmutant genes of the regulatory type I subunit. J. Biol. Chem.262:13111-13119.

6. Cohen, P. 1989. The structure and regulation of protein phos-phatases. Annu. Rev. Biochem. 58:453-508.

7. Correll, L. A., T. A. Woodford, J. D. Corbin, P. L. Mellon, andG. S. McKnight. 1989. Functional characterization of cAMP-binding mutations in type I protein kinase. J. Biol. Chem. 264:16672-16678.

8. Cruzalegui, F. H., M. S. Kapiloff, J. P. Morfin, B. E. Kemp, M. G.Rosenfeld, and A. R Means. 1992. Regulation of intrastericinhibition of the multifunctional calcium/calmodulin-dependentprotein kinase. Proc. Natl. Acad. Sci. USA 89.12127-12131.

9. Cruzalegui, F. H., and A. R Means. 1993. Biochemical character-ization of the multifunctional Ca2+/CaM-dependent protein ki-nase type IV expressed in insect cells. J. Biol. Chem. 268:26171-26178.

10. Dash, P. K., K. A. Karl, M. A. Colicos, R. Prywes, and E. R.Kandel. 1991. cAMP response element-binding protein is acti-vated by Ca2+/calmodulin- as well as cAMP-dependent proteinkinase. Proc. Natl. Acad. Sci. USA 88:5061-5065.

11. de Wet, J. R., K. V. Wood, M. deLuca, D. R. Helinski, and S.Subramani. 1987. Firefly luciferase gene: structure and expressionin mammalian cells. Mol. Cell. Biol. 7:725-737.

12. Edlund, T., M. D. Walker, P. J. Barr, and W. J. Rutter. 1985.Cell-specific expression of the rat insulin gene: evidence for role oftwo distinct 5' flanking elements. Science 230:912-916.

13. Enslen, H., P. Sun, D. Brickey, S. H. Soderling, E. Klamo, andT. R. Soderling. 1994. Characterization of Ca2+/calmodulin-de-pendent protein kinase IV: role in transcriptional regulation. J.Biol. Chem. 269:15520-15527.

14. Frangakis, M. V., T. Chatila, E. R. Wood, and N. Sahyoun. 1991.Expression of a neuronal Ca2+/calmodulin-dependent proteinkinase, CaM kinase-Gr, in rat thymus. J. Biol. Chem. 266:17592-17596.

15. Frangakis, M. V., C.-A. Ohmstede, and N. Sahyoun. 1991. Abrain-specific Ca2+/calmodulin-dependent protein kinase (CaMkinase-Gr) is regulated by autophosphorylation. J. Biol. Chem.266:11309-11316.

16. Ginty, D. D., D. Glowacka, D. S. Bader, H. Hidaka, and J. A.Wagner. 1991. Induction of immediate early genes by Ca2+ influxrequires cAMP-dependent protein kinase in PC12 cells. J. Biol.Chem. 266:17454-17458.

17. Gonzalez, G. A., and M. R. Montminy. 1989. Cyclic AMP stimu-lates somatostatin gene transcription by phosphorylation of CREBat serine 133. Cell 59:675-680.

18. Griffith, L. C., and H. Schulman. 1988. The multi-functionalcalcium/calmodulin-dependent protein kinase mediates calcium-dependent phosphorylation of tyrosine hydroxylase. J. Biol. Chem.263:9542-9549.

19. Guthrie, C. R., and G. S. McKnight. Unpublished data.20. Hanson, P. I., and H. Schulman. 1992. Neuronal Ca2+/calmodulin-

dependent protein kinases. Annu. Rev. Biochem. 61:559-601.21. Jensen, K. F., C.-A. Ohmstede, R S. Fisher, and N. Sahyoun. 1991.

Nuclear and axonal localization of Ca2+/calmodulin-dependentprotein kinase type Gr in rat cerebellar cortex. Proc. Natl. Acad.Sci. USA 88:2850-2853.

22. Jones, D. A., J. Glod, D. Wilson-Shaw, W. E. Hahn, and J. M.

VOL. 14, 1994

on Novem

ber 9, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from

6116 MATTHEWS ET AL.

Sikela. 1991. cDNA sequence and differential expression of themouse Ca2+/calmodulin-dependent protein kinase IV gene. FEBSLett. 289:105-109.

23. Kameshita, I., and H. Fujisawa. 1991. Phosphorylation and func-tional modification of calmodulin-dependent protein kinase IV bycAMP-dependent protein kinase. Biochem. Biophys. Res. Com-mun. 180:191-196.

24. Kameshita, I., and H. Fujisawa. 1993. Autophosphorylation ofcalmodulin-dependent protein kinase IV from rat cerebral cortex.J. Biochem. (Tokyo) 113:583-590.

25. Kaneseki, T., Y. Ikeuchi, H. Sugiura, and T. Yamauchi. 1991.Structural features of Ca2+/calmodulin-dependent protein kinaseII revealed by electron microscopy. J. Cell Biol. 115:1049-1060.

26. Low, M. J., R. E. Hammer, R. H. Goodman, J. F. Habener, R. D.Palmiter, and R. L. Brinster. 1985. Tissue-specific posttransla-tional processing pre-prosomatostatin encoded by a metallothio-nein-somatostatin fusion gene in transgenic mice. Cell 41:211-219.

27. McDonald, 0. B., B. M. Merrill, M. M. Bland, L. C. E. Taylor, andN. Sahyoun. 1993. Site and consequences of the autophosphory-lation of Ca2+/calmodulin-dependent protein kinase type "Gr." J.Biol. Chem. 268:10054-10059.

28. Means, A. R. Unpublished data.29. Means, A. R., F. Cruzalegui, B. LeMagueresse, D. S. Needleman,

G. R. Slaughter, and T. Ono. 1991. A novel Ca2+/calmodulin-dependent protein kinase and a male germ cell-specific calmodu-lin-binding protein are derived from the same gene. Mol. Cell.Biol. 11:3960-3971.

30. Mellon, P., C. H. Clegg, L. A. Correll, and G. S. McKnight. 1989.Regulation of transcription by cyclic AMP-dependent proteinkinase. Proc. Natl. Acad. Sci. USA 86:4887-4891.

31. Miyano, O., I. Kameshita, and H. Fujisawa. 1992. Purification andcharacterization of a brain-specific multifunctional calmodulin-dependent protein kinase from rat cerebellum. J. Biol. Chem.267:1198-1203.

32. Mochizuki, H., T. Ito, and H. Hidaka. 1993. Purification andcharacterization of Ca2+/calmodulin-dependent protein kinase Vfrom rat cerebrum. J. Biol. Chem. 268:9143-9147.

33. Ohmstede, C.-A., K. F. Jensen, and N. Sahyoun. 1989. Ca2+/calmodulin-dependent protein kinase enriched in cerebellar gran-ule cells. Identification of a novel neuronal calmodulin-dependentprotein kinase. J. Biol. Chem. 264:5866-5875.

34. Okuno, S., and H. Fujisawa. 1993. Requirement of brain extractfor the activity of brain calmodulin-dependent protein kinase IVexpressed in Escherichia coli. J. Biochem. (Tokyo) 114:167-170.

35. Orellana, S. A., and G. S. McKnight. 1990. The S49 kin- cell linetranscribes and translates a functional mRNA coding for thecatalytic subunit of cAMP-dependent protein kinase. J. Biol.Chem. 265:3048-3053.

36. Orellana, S. A., and G. S. McKnight. 1992. Mutations in thecatalytic subunit of cAMP-dependent protein kinase result inunregulated biological activity. Proc. Natl. Acad. Sci. USA 89:4726-4730.

37. Otten, A. D., and G. S. McKnight. 1989. Overexpression of thetype II regulatory subunit of the cAMP-dependent protein kinaseeliminates the type I holoenzyme in mouse cells. J. Biol. Chem.264:20255-20260.

38. Ouimet, C. C., T. L. McGuinness, and P. Greengard. 1984.Immunocytochemical localization of calcium/calmodulin-depen-

dent protein kinase II in rat brain. Proc. Natl. Acad. Sci. USA 81:5604-5608.

39. Picciotto, M. R., A. J. Czernik, and A. C. Nairn. 1993. Calcium/calmodulin-dependent protein kinase I. J. Biol. Chem. 268:26512-26521.

40. Planas-Silva, M. D., and A. R. Means. 1992. Expression of aconstitutive form of calcium/calmodulin dependent protein kinaseII leads to arrest of the cell cycle in G2. EMBO J. 11:507-517.

41. Sahyoun, N., 0. B. McDonald, F. Farrell, and E. G. Lapetina.1991. Phosphorylation of a ras-related GTP-binding protein, rap-lb, by a neuronal Ca2+/calmodulin-dependent protein kinase,CaM kinase Gr. Proc. Natl. Acad. Sci. USA 88:2643-2647.

42. Schilling, K., D. Luk, J. I. Morgan, and T. Curran. 1991. Regu-lation of a fos-lacZ fusion gene: a paradigm for quantitativeanalysis of stimulus-transcription coupling. Proc. Natl. Acad. Sci.USA 88:5665-5669.

43. Schulman, H. 1993. The multifunctional Ca2+/calmodulin-depen-dent protein kinases. Curr. Opin. Cell. Biol. 5:247-253.

44. Schulman, H. (Stanford University). Personal communication.45. Sculptoreanu, A., T. Scheuer, and W. A. Catterall. 1993. Voltage-

dependent potentiation of L-type Ca2+ channels due to phopho-rylation by cAMP-dependent protein kinase. Nature (London)364:240-243.

46. Sheng, M., G. McFadden, and M. E. Greenberg. 1990. Membranedepolarization and calcium induce c-fos transcription via phos-phorylation of transcription factor CREB. Neuron 4:571-582.

47. Sheng, M., M. A. Thompson, and M. E. Greenberg. 1991. CREB:a Ca2+-regulated transcription factor phosphorylated by calmodu-lin-dependent kinases. Science 252:1427-1430.

48. Silva, A. J., R. Paylor, J. M. Wehner, and S. Tonegawa. 1992.Impaired spatial learning in ox-calcium-calmodulin kinase II mu-tant mice. Science 257:206-211.

49. Silva, A. J., C. F. Stevens, S. Tonegawa, and Y. Wang. 1992.Deficient hippocampal long-term potentiation in oC-calcium-calm-odulin kinase II mutant mice. Science 257:201-206.

50. Suidan, H. S., R. D. Murrell, and A. M. Tolkovsky. 1991. Carba-chol and bradykinin elevate cyclic AMP and rapidly deplete ATPin cultured rat sympathetic neurons. Cell Regul. 2:13-25.

51. Sun, P., H. Enslen, P. S. Myung, and R. A. Maurer. Submitted forpublication.

52. Thastrup, O., P. J. Cullen, B. K. Dr0bak, M. R. Hanley, and A. P.Dawson. 1990. Thapsigargin, a tumor promoter, discharges intra-cellular Ca2+ stores by specific inhibition of the endoplasmicreticulum Ca2+-ATPase. Proc. Natl. Acad. Sci. USA 87:2466-2470.

53. Waeber, G., and J. F. Habener. 1991. Nuclear translocation andDNA recognition signals colocalize within the bZIP domain ofcyclic adenosine 3',5'-monophosphate response element-bindingprotein CREB. Mol. Endocrinol. 5:1431-1438.

54. Walton, K. M., R. P. Rehfuss, J. C. Chrivia, J. E. Lochner, andR. H. Goodman. 1992. A dominant repressor of cyclic adenosine3',5'-monophosphate (cAMP)-regulated enhancer binding proteinactivity inhibits the cAMP-mediated induction of the somatostatinpromoter in vivo. Mol. Endocrinol. 6:647-655.

55. Wegner, M., Z. Cao, and M. G. Rosenfeld. 1992. Calcium-regulated phosphorylation within the leucine zipper of C/EBPI3.Science 256:370-373.

MOL. CELL. BIOL.

on Novem

ber 9, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from