Neuron, Vol. 39, 625–639, August 14, 2003, Copyright 2003 by Cell Press

Calcium Channel and NMDA Receptor ActivitiesDifferentially Regulate Nuclear C/EBP� Levelsto Control Neuronal Survival

NMDA receptors and appears to contribute to glutamateexcitotoxicity (Ruiz et al., 2000; Guo et al., 2001a).

Differential calcium effects may arise, in part, fromdifferences in the subcellular distribution of calcium-dependent enzymes. For instance, calcium influx via

John Marshall,1,* Bridget M. Dolan,1

Elizabeth P. Garcia,1 Suvarna Sathe,1

Xiaoli Tang,2 Zixu Mao,2

and Leslie A.C. Blair11Department of Molecular Pharmacology,

extrasynaptic NMDA receptors specifically promotesPhysiology, and Biotechnologydeath, while activation of synaptic receptors has beenBrown Universityassociated with survival (Hardingham et al., 2002).Providence, Rhode Island 02912Among the CaMKs, the cytoplasmic CaMKII has been2 Department of Medicineimplicated in controlling synaptic plasticity (Soderling,Rhode Island Hospital and Brown University2000), while nuclear CaMKIV has been linked to survivalProvidence, Rhode Island 02903in multiple cell types including neurons and neural celllines (McGinnis et al., 1998; Curtis and Finkbeiner, 1999;See et al., 2001).

Summary At the nuclear level, IGF-1 and L channel activity areeffective at activating survival-supporting transcription

Insulin-like growth factor-1 (IGF-1) promotes the sur- factors such as CREB and MEF2 (Misra et al., 1994;vival of cerebellar granule neurons by enhancing cal- Bonni et al., 1999; Mao et al., 1999; Riccio et al., 1999).cium influx through L-type calcium channels, whereas The mechanism requires calmodulin (CaM), which asso-NMDA receptor-mediated calcium influx can lead to ciates with neuronal L channels (Peterson et al., 1999).excitotoxic death. Here we demonstrate that L and Calcium influx leads to CaM activation, which, via theNMDA receptor channel activities differentially regu- MAPK pathway, elevates pCREB levels (Dolmetsch etlate the transcription factor C/EBP� to control neu- al., 2001). Additionally, L channel activity can lead toronal survival. Specifically, we show that L channel- nuclear translocation of CaM and elevated nuclear cal-dependent calcium influx results in increased CaMKIV cium levels that, in turn, stimulate Ca2�/CaM-dependentactivity, which acts to decrease nuclear C/EBP� lev- protein kinase-IV (CaMKIV) to phosphorylate CREB (Deis-els. Conversely, NMDA receptor-mediated influx rap- seroth et al., 1998; Chawla et al., 1998; Hu et al., 1999).idly elevates nuclear C/EBP� and induces excitotoxic Other calcium-dependent transcription factors thatdeath via activation of the calcium-dependent phos- may play significant roles in survival include the CCAATphatase, calcineurin. Moderate levels of AMPA recep- enhancer binding proteins (C/EBPs). To date, six genestor activity stimulate L channels to improve survival, encoding a minimum of eight isoforms have been identi-whereas higher levels stimulate NMDA receptors and fied. C/EBP proteins function as dimers and contain areduce neuronal survival, suggesting differential syn- transactivating domain, a DNA binding region, and aaptic effects. Finally, N-type calcium channel activity leucine zipper (Olive et al., 1996). They are regulatedreduces survival, potentially by increasing glutamate transcriptionally, posttranslationally, and by the abilityrelease. Together, these results show that the L-type to hetero- as well as homodimerize (Niehof et al., 2001);calcium channel-dependent survival and NMDA re- at least some C/EBPs can heterodimerize to form re-ceptor death pathways converge to regulate nuclear pressors (Descombes and Schibler, 1991). C/EBPs areC/EBP� levels, which appears to be pivotal in these also being increasingly associated with survival and dif-mechanisms. ferentiation processes (Sterneck and Johnson, 1998;

Cortes-Canteli et al., 2002; Ramji and Foka, 2002; Zhuet al., 2002). Interestingly, the full-length � subtype isIntroductionregulated by CaMKIV (Yukawa et al., 1998), as well asPI3K and Akt (Guo et al., 2001b; Piwien-Pilipuk et al.,In neurons, calcium influx can serve as both a signal2001), both of which are essential for IGF-1/L channelpromoting survival as well as in some instances promot-potentiation-induced survival (Blair and Marshall, 1997;ing cell death. Survival is promoted by insulin-likeBlair et al., 1999).

growth factor-1 (IGF-1) and L channel-mediated calciumHere, we investigate the downstream signaling re-

influx (Galli et al., 1995; Dudek et al., 1997). L channelsponsible for L channel-mediated neuronal survival and

activity is directly controlled by depolarization, but also potential antagonistic actions of NMDA receptor activ-is enhanced by IGF-1 through a pathway that requires ity. We focused on the potential roles of C/EBP� andPI 3-kinase (PI3K), Akt, and ultimately src-dependent CaMKIV in IGF-1- and L channel-mediated survival, andphosphorylation of the channel (Blair and Marshall, we found a novel mechanism. IGF-1, in an L channel-1997; Blair et al., 1999; Bence-Hanulec et al., 2000). dependent manner, rapidly stimulates CaMKIV activityNonetheless, calcium does not invariably act as a pro- to promote neuronal survival by reducing nuclear levelssurvival factor. Calcineurin, a calcium- and CaM-depen- of C/EBP�. Conversely, loss of growth factor support ordent serine/threonine phosphatase, can be activated by strongly stimulating NMDA receptors rapidly increased

nuclear import of C/EBP� and induced subsequent celldeath.*Correspondence: john_marshall@brown.edu

Neuron626

IGF-1- and L Channel-Dependent SurvivalRequire CaMKIVWe next tested the potential role of CaMKIV in IGF-1-and L-mediated survival. Cultured cerebellar granuleneurons were transiently cotransfected with cDNAs en-coding a catalytically inactive CaMKIV (dn-CaMKIV;Chatila et al., 1996) and, to identify transfected neurons,GFP. After 1 day, survival of neurons overexpressingdn-CaMKIV was compared with that of untransfectedneurons in the same cultures and that in sister culturestransfected with wild-type (wt) CaMKIV. Cells were con-firmed as neuronal by immunolabeling with Tuj1, an anti-body against neuron-specific class III �-tubulin (seeSupplemental Figure S1 at http://www.neuron.org/cgi/content/full/39/4/625/DC1). We found that overexpress-ing dn-CaMKIV prevented survival (Figures 2A, 2B, and2F), suggesting that CaMKIV activity may be an absoluterequirement. In contrast, overexpressing wt-CaMKIVFigure 1. IGF-1 Rapidly Stimulates Both CaMKII and -IV, but Onlyhad only modest effects when IGF-1 was present at aCaMKIV Activity Remains Elevated for Extended Timesrelatively high level (50 ng/ml; Figure 2F, SupplementalTime course of IGF-1 stimulation revealed a small, transient increaseFigures S2A–S2C). To block endogenous and overex-in CaMKII activity (circles), but a large and longer duration increasepressed wt-CaMK activity acutely, neurons were treatedin CaMKIV activity (squares). Although both activities increased

within 2 min, the small change in CaMKII activity rapidly returned with KN62 or its vehicle, then briefly stimulated withto baseline, while CaMKIV activity remained elevated for �2 hr. IGF-1 in the presence or absence of calcium channelSquares with vertical stripes: blocking L channel activity with 5 �M inhibitors. We found that acute inhibition of CaMKs fullynimodipine largely blocked the IGF-1-stimulated increase in CaMKIV

blocked the IGF-1-induced increase in survival and alsoactivity. Square with horizontal stripes: the general CaMK inhibitor,appeared to block basal L channel-dependent survival,KN62 (10 �M), also blocked the increase in CaMKIV activity. Sistersuggesting that L channel and CaMKIV activities uti-cultures of granule neurons were stimulated with IGF-1 (50 ng/ml)

for 2 min–4 hr, then CaMKII and -IV were specifically immunoprecipi- lize the same molecular signaling pathways (Figures 2Gtated and their activities assessed in vitro. Activity levels following and 7).IGF-1 stimulation were normalized to the baseline activities ob- Significantly, constitutively active (ca) CaMKIV stronglyserved in the absence of IGF-1. For CaMKIV, data are means � promoted neuronal survival in the absence of IGF-1SEMs from four experiments; n � 2 experiments for CaMKII.

(Figures 2C–2F), while a cytoplasmically targeted consti-tutively active CaMKII had no effect (Figure 2F). Over-expresssing ca-CaMK kinase (CaMKK), a calcium-dependent enzyme required for CaMKIV activation, also

Resultsincreased survival (Figure 2F). We further tested if ca-or wt-CaMKIV-dependent survival required L channel

IGF-1 Rapidly Stimulates Both CaMKII and -IV, activity: survival of untransfected, ca-CaMKIV-trans-but Only CaMKIV Activity Remains Elevated fected, and wt-CaMKIV-transfected neurons was �95%IGF-1 and CaMK activity have been separately shown in the presence of IGF-1 but was reduced to �65%to improve neuronal survival, but it was unknown if IGF-1 by the L channel blocker nimodipine (Figures 2D–2G,could activate CaMKs. We therefore tested this and Supplemental Figures S2B and S2C). Indistinguishableassessed dependence on L channel activity. Sister cul- results were obtained when the nonspecific blocker cad-tures of cerebellar granule neurons were exposed to mium (Cd2�) was used the during brief (1 min) exposureIGF-1 (50 ng/ml) for up to 4 hr, and CaMKII and -IV to IGF-1 (Figure 3A). Interestingly, extended Cd2�-blockspecifically immunoprecipitated. Kinase activities were improved survival, suggesting that extended block ofthen assayed in vitro by determining 32P-incorporation other calcium channel subtypes might exert prosurvivalinto specific substrates (Yoshida et al., 2000). Both activ- effects and leading us to test N channel inhibition (seeities increased within 2 min, but the peak elevation of Figure 6B).CaMKIV activity was �4-fold greater and of longer dura- We also found that increasing CaMKIV levels shiftedtion than that of CaMKII (Figure 1; n � 4 independent the sensitivity to IGF-1- and L channel-mediated sur-experiments). Maximal CaMKIV stimulation occurred at vival. Physiological levels of IGF-1 in the CNS are typi-�10 min. Comparable results were obtained using cally defined as the levels detected in cerebrospinalwhole-cell extracts and a CaMKIV-specific substrate fluid. Nonetheless, the levels directly experienced by(n � 2 independent experiments). neurons and their fluctuation over time are unknown.

L channel activity was necessary. When IGF-1 was We therefore determined the dose responsiveness ofadded in the presence of nimodipine and CaMKIV activ- survival to IGF-1, if transient stimulation was sufficientity assessed after a 10 min stimulation, the increase was and if responsiveness was affected by CaMKIV levels.largely, although not completely, blocked (Figure 1). As Neurons were transfected with wt-CaMKIV and exposedan internal control, the general CaMK inhibitor, KN62, briefly (1 min) to increasing IGF-1 concentrations. Cul-was also tested; it, but not its vehicle, fully blocked the tures were then maintained for 1 day in unsupplemented

test medium. We found that moderate-high levels ofIGF-1-induced increase (Figure 1).

L Channels and NMDA Receptors Regulate C/EBP�627

Figure 2. Overexpressing Inactive CaMKIV Abolishes Granule Neuron Survival, while Overexpressing Constitutively Active CaMKIV MimicsIGF-1

(A–E) After 1 day in the indicated test media, all neurons were stained with propidium iodide (PI) to reveal DNA (lower), and transfected neuronswere identified by GFP fluorescence (upper). Healthy transfectants having dispersed chromatin are indicated by open arrowheads; apoptotictransfectants having bright, condensed chromatin, closed arrowheads.(A and B) Neurons expressing inactive (dn) CaMKIV are apoptotic, even though in IGF-1-containing media, the surrounding untransfectedneurons are healthy.(C) In the absence of exogenous support factors, neurons transfected with a constitutively active (ca) CaMKIV thrive, while most untransfectedones do not. Immunocytochemical analysis confirmed that the heterologous CaMKIV localized correctly to the nucleus (see SupplementalFigure S2D).(D) In IGF-1, untransfected as well as ca-CaMKIV-transfected neurons survive.(E) When L channel activity is inhibited (nim), survival in both ca-CaMKIV-transfectants and nontransfectants is reduced, indicating a requirementfor calcium influx.(F) CaMKIV is required for neuronal survival. Overexpressing dominant-negative CaMKIV eliminates survival, even in the presence of IGF-1.Conversely, constitutively active CaMKIV and CaMKK induce full survival by themselves (No add), but only if L channels are active; when Lchannels are inhibited (IGF-1�nim, nim), survival drops to levels similar to those seen in untransfected neurons and wild-type CaMKIVtransfectants. Interestingly, constitutively active CaMKII failed to improve survival. IGF-1, 50 ng/ml; nimodipine (nim), 5 �M. Open bars, survivalof transfected neurons. Narrow closed bars, survival of untransfected neurons in the same cultures. Data are means � SEMs from 6 independentexperiments. All detectable transfectants were counted; n/bar � �700 neurons (range, 594–853). For nontransfectants, all nongreen neuronsin 4 randomly chosen fields/dish were counted; n/bar � �3500 neurons (range, 2887–4416). Asterisk indicates when the survival of transfectantswas significantly different from untransfected neurons in the identical test condition (p � 0.005, Student’s t test); double asterisk indicatesthe cases when survival was slightly different between transfectants and nontransfectants in a given test condition (p � 0.025).(G) The general CaMK inhibitor, KN62, eliminates IGF-1-induced survival in neurons expressing endogenous and elevated wt-CaMKIV levels.Open and closed bars, as in (F). Few neurons survived 1 day after transient exposure to KN62 (10 �M for 2 hr); the level was indistinguishablefrom that seen when L channels were blocked in the absence of IGF-1 (see F, 33% � 2%). After KN62 pretreatment, neither IGF-1 (50 ng/ml,1 min) nor L channel inhibition (5 �M nimodipine or 100 �M Cd2�) had any effect. Conversely, IGF-1 in the KN62 vehicle (0.1% DMSO) producednearly complete survival, and vehicle alone produced a level of survival similar to that observed in test media alone (i.e., no IGF-1 but alsono L channel block; see F, 41% � 2%). Data are means � SEMs from 4 experiments.

Neuron628

Figure 3. Overexpressing Wild-Type CaMKIV Reduces the Requirement for IGF-1, but Maintains an L Channel-Dependent Component

Solid symbols, transfected neurons; open symbols, sister nontransfectants. Values are means � SEMs.(A) A brief pulse of IGF-1 promotes neuronal survival: dose-response analysis shows that neurons with elevated wt-CaMKIV levels need lessIGF-1 to survive, but retain a requirement for calcium influx. IGF-1 was applied for 1 min, then cultures were rinsed and placed in 0-serummedium for 24 hr. L channel activity was inhibited during the IGF-1 stimulation with either nimodipine (5 �M) or Cd2� (100 �M). No differencesbetween brief Cd2�- and nimodipine-block were detectable at any IGF-1 concentration (p � 0.4, Student’s t test). n/test condition � �450wt-transfected neurons (range, 433–473) and �2700 nontransfectants (range, 2387–3182) from 3 independent experiments.(B) Overexpressing wt-CaMKIV allows very low concentrations of the L channel agonist, (-)-BayK8644 (1 day treatment), to promote survival.n/test concentration � �500 wt-transfected neurons (range, 373–645) and �3100 nontransfectants (range, 2638–3605) from 3 independentexperiments.

IGF-1 (10–100 ng/ml) increased survival of both transfec- 1% (wt-CaMKIV transfectants and nontransfectants, re-spectively). Interestingly, neurons depolarized for 10 mintants and nontransfectants almost to the levels obtained

with a 24 hr application, but that at very low concentra- survived less well (62% � 2% and 61% � 1%, p �0.001) than those depolarized for 1 min, again implyingtions (0.01–0.1 ng/ml), wt-CaMKIV-overexpressing neu-

rons survived much better (Figure 3A). This suggests that calcium levels need to be optimized. Together, theresults indicate that increasing CaMKIV levels primesthat the requirement for IGF-1 can be reduced or partially

bypassed by increasing CaMKIV. We further found that the survival mechanism, reducing but not eliminating arequirement for moderate calcium influx.at all IGF-1 concentrations, blocking L channel activity

reduced survival (Figure 3A), implying that L channel-mediated calcium influx remains required and that addi- Inhibiting C/EBP Activity Enhances Neuronal

Survival in the Absence of IGF-1tional survival processes are calcium dependent.L channel activity is also neuroprotective in the ab- C/EBPs are calcium- and CaMK-regulated transcription

factors recently implicated in the survival of a variety ofsence of growth factor support. We therefore tested ifincreasing CaMKIV levels would affect survival pro- cell types, but not directly tested in neurons. To establish

if they contribute to the antiapoptotic effects of IGF-1moted by the L channel agonist, (-)-BayK8644 (Figure3B). As previously shown (Blair et al., 1999), optimal or L channel activity in primary neurons, we utilized a

dominant-negative C/EBP (dn-C/EBP) known to inhibitconcentrations of (-)-BayK8644 were strongly but in-completely neuroprotective. In its absence, survival of the DNA binding activity of all C/EBPs in vivo (Olive

et al., 1996). Cultured granule neurons were transientlyboth transfectants and nontransfectants was poor. Inhigh concentrations, survival decreased below that of cotransfected with cDNAs encoding dn-C/EBP plus

GFP. After 1 day, survival of transfectants was assessedthe unsupplemented test medium, consistent with theidea that excessive calcium causes cell death. However, using either propidium iodide labeling or the TUNEL

assay for nicked DNA. Intriguingly, we found that generallow (�1 �M) but nonzero concentrations were muchmore neuroprotective for wt-CaMKIV-overexpressing inhibition of C/EBP activity dramatically elevated sur-

vival (Figures 4A and 4F): in serum-free test medium,neurons than their untransfected sisters.Similarly, depolarization-induced survival was better 84% � 2% of the dn-C/EBP-transfected neurons sur-

vived, as compared to 44% � 2% among the non-in neurons expressing higher levels of wt-CaMKIV andwas fully inhibited by simultaneous treatment with L transfectants in the same culture. Wild-type was also

tested.channel blockers. Neurons were exposed for 1 min to90 mM KCl (expected to drive the membrane potentialto �10 mV), rinsed, and returned to unsupplemented Specific Inhibition of C/EBP� Mimics

IGF-1-Induced Survival and Reduces NMDAtest medium for 24 hr. We found that survival was 73% �2% in wt-CaMKIV-overexpressing neurons, but only Receptor-Mediated Death

The � isoform of C/EBP was of particular interest since66% � 1% in untransfected neurons (p � 0.001, Stu-dent’s t test). In addition, the improved survival required its activity can be controlled via PI3K and Akt (Guo et

al., 2001b; Piwien-Pilipuk et al., 2001), known compo-L channel-mediated influx: in 90 mM KCl plus nimodipine(5 �M), survival was reduced to 43% � 1% and 42% � nents of IGF-1-dependent neuroprotection. To test this,

L Channels and NMDA Receptors Regulate C/EBP�629

C/EBP� levels were manipulated by several approaches. independent experiments). In contrast, nuclear levelsremained low in neurons maintained in serum-free me-First, we increased them by overexpressing wild-typedia supplemented with IGF-1; at 1 hr, the relativeC/EBP� and found that survival significantly decreasedchange, 1.0 � 0.1-fold, was not detectably different from(30% � 2%; Figures 4A and 4F).1.0 (Figure 5A, panel a1; n � 5 independent experi-To confirm the involvement of C/EBP�, its levels werements). After 2 hr in IGF-1, levels even appeared tospecifically reduced using both antisense and siRNAdecrease (0.6 � 0.3). Moreover, IGF-1 appears to reverseapproaches (Taubenfeld et al., 2001a; Krichevsky andthe effects of serum withdrawal, slightly decreasing nu-Kosik, 2002). As controls for any nonspecific effects ofclear C/EBP� levels as soon as 15 min after additiontransfection, sister cultures were treated with either aand strongly decreasing them by 1 hr (0.7 � 0.1; Figuresequence-scrambled antisense C/EBP� oligodeoxy-5B). Dose dependence was also observed (Figure 5A,nucleotide (ODN) or an siRNA that did not affect nuclearpanel a2; n � 3 independent experiments): althoughC/EBP� levels. Cell counts were performed after 1 daynuclear C/EBP� levels were high after 2 hr in unsupple-in serum-free unsupplemented test medium, treating allmented medium, levels in sister cultures exposed toneurons as probable transfectants (see Experimentalminimal IGF-1 concentrations were low (0.5 ng/ml,Procedures). Survival in both C/EBP� antisense and39% � 2% relative to no IGF-1), and increasing IGF-1siRNA was high: 89% � 1% and 90% � 3%, respectivelyfurther reduced nuclear levels (5 ng/ml, 24% � 4%; 50(Figures 4B, 4D, and 4G), values indistinguishable fromng/ml, 18% � 4%; 500 ng/ml, 11% � 5%). Experimentalthose obtained upon IGF-1 treatment of nontransfectedcontrols included assessing CREB; as expected, IGF-1neurons (91% � 2%, Figure 4F). Conversely, survival inincreased its phosphorylation (data not shown). As con-the presence of the sequence-scrambled ODN or non-trols for the subcellular fractionation procedure, bothblocking siRNA was 44% � 3% and 37% � 3% (Figuresnuclear and cytoplasmic fractions were assayed for the4C, 4E, and 4G), similar to untreated neurons in basicpresence of unique markers: the transcription factor,test medium (43% � 2%, Figure 4F) and showing thatc-jun, was found only in the nuclear fraction, while thethe improved survival was directly dependent on inhib-nonnuclear marker, the mitochondrial enzyme GAPDH,iting the C/EBP� subtype.was excluded (Supplemental Figure S3B).Both antisense and siRNA inhibition of C/EBP� antag-

The rapid effects did not appear to involve degrada-onized NMDA receptor-mediated death. When neuronstion or de novo synthesis. With or without IGF-1, levelswere continuously exposed to high levels of NMDA (300were, respectively, 0.8 � 0.2 or 1.0 � 0.1 at 30 min and�M � 1 �M glycine), a substantial fraction of the anti-0.9 � 0.2 or 1.1 � 0.1 at 2 hr (Supplemental Figure S3Csense and siRNA transfectants survived (69% � 5%at http://www.neuron.org/cgi/content/full/39/4/625/and 67% � 3%, respectively; Figures 4B, 4D, and 4G).DC1; n � 2 independent experiments). When assayedHowever, almost no neurons treated with the controlimmunocytochemically and biochemically using transla-ODN or nonblocking siRNA lived (3% � 1%, 4% � 2%;tional or transcriptional inhibitors (cycloheximide, acti-Figures 4C, 4E, and 4G). Significantly, antisense treat-nomycin D), levels were not significantly higher than inment may be able to rescue severely traumatized neu-IGF-1 alone (Supplemental Figure S3D), suggesting thatrons. When pretreated with NMDA prior to transfectionC/EBP� was being translocated between subcellularand then maintained in NMDA, essentially all controlcompartments. In addition, treatment with either anti-neurons died (3% � 2%), but 22% � 6% of the anti-sense ODN or siRNA was able to keep nuclear C/EBP�sense-treated neurons survived.levels low in neurons maintained 1 day in 0-serum (Fig-ure 5C; n � 3 independent experiments, ODN; n � 4

Antagonistic Regulation of Nuclear C/EBP� independent experiments, siRNA).Levels and Survival The IGF-1 effect was linked through L channel andThe survival data implies that increasing C/EBP� ex- CaMKIV activities. Inhibiting L channels during IGF-1pression reduces survival. To establish if this was due treatment led to a �2-fold increase in nuclear C/EBP�to increased availability of the transcription factor, nu- levels (Figure 5A, panel a1; n � 10 independent experi-clear C/EBP� levels were assayed biochemically and ments). Inhibiting L channel activity also increased nu-immunocytochemically. Briefly, we found that nuclear clear levels in neurons overexpressing wild-typelevels were high in the absence of growth factors but C/EBP�. However, any IGF-1 effects mediated by Llow in the presence of IGF-1, although blocking either channel potentiation rely on spontaneous electrical ac-CaMK or L channel activities increased them (Figures tivity to open the channels. Consequently, we tested if5A and 5B; Supplemental Figure S3A at http://www. directly activating L channels with moderate depolariza-neuron.org/cgi/content/full/39/4/625/DC1). Similarly, tion (25 mM KCl, predicted to drive the membrane poten-NMDA receptor activity also increased them, an effect tial to �45 mV) would also maintain low nuclear C/EBP�that antisense C/EBP� blocked (Figures 5A and 5D; Sup- levels. Neurons depolarized in the absence of growthplemental Figures S4 and S5). factor support did have low C/EBP� levels, but not if L

The increase in nuclear C/EBP� levels observed upon channel activity was blocked with nimodipine (Figuregrowth factor deprivation was very rapid (Figure 5B). 5A, panel c; n � 3 independent experiments). Similarly,Effects were detectable within 1 min and increased with the general CaMK inhibitor, KN62 (10 �M), blocked thetime: the fold-changes at 1 min, 5 min, and 1 hr were, ability of IGF-1 to maintain low nuclear levels: levelsrespectively, 1.7 � 0.4, 2.8 � 0.9, and 4.0 � 0.9 (mean � increased 3.0 � 0.3-fold within 1 hr (Figure 5A, panelSEM, calculated relative to the 1.0 level observed in full a1; n � 4 independent experiments). Assessed immuno-serum media, quantified by Western blot analyses and cytochemically, overexpressing a dominant-negative

CaMKIV and treatment with KN62 both increased nu-scanning densitometry of subcellular fractions; n � 7

Neuron630

Figure 4. Role of C/EBP� in Neuronal Survival/Death Pathways

(A) General inhibition of C/EBPs enhances neuronal survival while overexpressing C/EBP� reduces it. Granule neurons were cotransfectedwith GFP and either a dominant-negative C/EBP (dn-C/EBP) that inhibits DNA binding activity of all known C/EBPs or full-length wild-typeC/EBP�. Survival was assessed after 1 day in 0-serum test medium. Top, transfectants identified by GFP fluorescence. Bottom, all neuronsin the same field labeled with propidium iodide (PI) or TUNEL to reveal DNA; closed arrowheads, apoptotic transfectants; open arrowheads,healthy cells. Left panels: in the absence of growth factor support, dn-C/EBP-expressing neurons survived much better than untransfectedneurons in the same culture when assessed by PI staining. When assessed using the TUNEL assay for damaged DNA, dn-C/EBP-expressingneurons were TUNEL negative, even though many surrounding untransfected neurons were TUNEL positive. Center panels: L channel inhibitionreduced, but did not eliminate, survival of dn-C/EBP-expressing neurons. Right panels: in sister cultures, neurons overexpressing wild-typeC/EBP� survived poorly, having condensed chromatin (PI) and being TUNEL positive.(B and D) Specific inhibition of C/EBP� with an antisense oligodeoxynucleotide (ODN, B) or siRNA (D) enhances neuronal survival in theabsence of growth factor support and opposes NMDA-induced excitotoxity. Left, antisense- and siRNA-treated neurons survived well. Right,continuous NMDA treatment induced only moderate damage.

L Channels and NMDA Receptors Regulate C/EBP�631

clear C/EBP� levels (Figure 5A, panels d2 and d3), while (p � 0.001; n � 2 independent experiments), implyingthat translocation is an important mechanism by whichlevels fell in neurons overexpressing a constitutively ac-

tive CaMKIV (Figure 5A, panel d1; n � 6 independent IGF-1 regulates C/EBP� to control survival (Figure 7).experiments). Analysis of confocal Z-sections verifiedthat, in the presence of IGF-1, L channel activity, and AMPA Receptors Can Activate L Channels

and Promote L Channel-Mediated SurvivalC/EBP�-antisense, C/EBP� was largely excluded fromnuclei, while serum deprivation and L channel block IGF-1 potentiates but does not directly activate voltage-

sensitive L channels. We therefore attempted to deter-increased nuclear levels (Figure 5A, panel b, Supple-mental Figures S3A and S5B). mine the means by which L channel activity is stimulated

and found that AMPA receptors appear to play a pivotalWe also found that NMDA treatment increased nu-clear levels and that the NMDA-induced nuclear localiza- role, increasing L channel activity to promote survival

and NMDA receptor activity to increase excitotoxiction required the calcium-dependent phosphatase cal-cineurin (Figure 5A, panel a1, Supplemental Figures S4A death.

Low concentrations of the AMPA receptor agonist,and S4B at http://www.neuron.org/cgi/content/full/39/4/625/DC1). Deltamethrin (1 �M), a specific inhibitor, kainate, significantly elevated L channel-mediated sur-

vival (Figure 6A). Increasing kainate concentrationsblocked the NMDA-dependent increase in nuclearC/EBP� (n � 6 independent experiments) and simulta- (0–100 �M) were tested in combination with 0.8 �M

(-)-BayK8644, a concentration of the L channel modula-neously increased survival (72% � 4%, p � 0.001 whencompared with 15% � 2% for NMDA alone; n � 3 inde- tor that produces submaximal survival (�60%, Blair et

al., 1999). We found that the lower kainate doses im-pendent experiments). Deltamethrin also reduced nu-clear levels in neurons maintained in 0-serum test me- proved survival, but that the improvement was entirely

reversed by L channel blockade. Moreover, when thedium (Supplemental Figure S4C). Moreover, IGF-1partially reversed the NMDA-induced rise in nuclear NMDA receptor inhibitor APV was included with

BayK8644�10 �M kainate, nearly all neurons survived.C/EBP� levels (Supplemental Figure S4D; n � 6 inde-pendent experiments) and, in parallel, raised survival Consistent results were obtained with kainate alone or

direct NMDA receptor stimulation (Figure 6A). Com-(66% � 3%, p � 0.001; n � 3 independent experiments).Significantly, IGF-1 could reverse the nuclear localiza- pared to “No add,” 10 �M kainate elevated survival in

an L channel-dependent manner, but 100 �M kainatetion induced by serum deprivation and NMDA treatment(Figures 5B and 5D). L channel activity was required. reduced it. When NMDA receptor activity was directly

stimulated (300 �M NMDA � 1 �M glycine for 30 min),When inhibited, nuclear C/EBP� levels rose (Supple-mental Figure S4E) and survival fell (40% � 3%; n � 4 survival after 1 day was poor and reduced further by

coaddition of 100 �M kainate. However, coaddition ofindependent experiments). This suggests that in termsof survival, NMDA receptor and L channel activities act 10 �M kainate significantly enhanced survival, again in

an L channel-dependent manner. This suggests that theantagonistically. Since neurotransmission occurs spon-taneously in these cultures (Mellor et al., 1998), we overriding effect of strongly activating AMPA receptors

is to stimulate the NMDA receptor/prodeath pathways,tested blockade of endogeneous NMDA receptor activ-ity. The NMDA receptor inhibitor, APV, which alone pro- while moderate activation stimulates L channels to pro-

mote survival.motes survival (see Figure 6B), was able to keep nuclearlevels low (Figure 5A, panel a1, Supplemental Figure Testing combinations of specific antagonists con-

firmed the dual role of AMPA receptor activity (FigureS4F), as did N channel inhibition (Supplemental FigureS4G), suggesting that the rise observed in unsupple- 6B; “No IGF-1”). After 1 day without IGF-1, survival was

poor (“No add”), and blocking basal L channel activitymented test medium was due, at least in part, to excit-atory neurotransmission. Strikingly, the ability to reverse (“nim”) produced a �10% further decrease. In contrast,

inhibiting NMDA receptors (“APV”) radically improvedthe effects of growth factor withdrawal and NMDA alsorequired the nuclear transport protein, CRM1, also survival, but simultaneous L channel block (“APV�nim”)

again produced a �10% decrease, implying thatcalled exportin-1. Inhibition by leptomycin B (Hendersonand Eleftheriou, 2000) blocked the reversal (Figure 5D; blocking NMDA receptor activity did not affect L channel

activity. No effect of kainate receptor activity was de-n � 3 independent experiments). Blocking C/EBP� ex-port also kept survival low: 22% � 6% after 1 day in tected. Blocking both kainate and AMPA receptors with

CNQX or high concentrations of NBQX (100 �M) wasleptomycin�IGF-1 versus 90% � 4% for IGF-1 alone

(C and E) As controls, sister cultures were treated with an ODN in which the antisense sequence had been scrambled (C) or with an siRNAthat failed to reduce nuclear C/EBP� levels (E). Left, without growth factor, support survival was modest. Right, NMDA treatment inducednearly global death.(F) In the absence of growth factor support, survival of neurons overexpressing a dominant-negative C/EBP is almost as high as in IGF-1-treated untransfected neurons. Conversely, overexpressing wild-type C/EBP� reduced survival. Open and closed bars, transfected anduntransfected neurons, respectively. Data are means � SEMs from 6 experiments. For transfectants, n/bar � �500 neurons (range, 446–561).For nontransfectants, n/bar � �3100 neurons (range, 3008–3250). Asterisk, in nearly all cases, the survival of transfectants was significantlydifferent from untransfected neurons in the identical test condition (p � 0.005).(G) Specific block of C/EBP� with either antisense ODN or siRNA strongly promotes survival and moderates NMDA-induced death. Whenadded, NMDA (300 �M � 1 �M glycine) was present throughout the 1 day test period. Data are means � SEMs from 8 experiments. n/bar ��3700 neurons (range, 2834–4830). In all cases, the survival of antisense and siRNA transfectants was significantly higher than in the pairedcontrol (p � 0.005).

Neuron632

Figure 5. Nuclear C/EBP� Levels Rapidly Rise in the Absence of Growth Factor Support, while IGF-1 Keeps Them Low in an L Channel- andCaMKIV-Dependent Manner

(A) IGF-1 keeps nuclear C/EBP� levels low (a and c), requiring L channel (a1, b, and c) and CaMK (a1 and d) activities. (a) Cultured cerebellargranule neurons were serum starved or treated with IGF-1 (50 ng/ml) for 1 hr, then lysed and the nuclear fractions were Western blotted forC/EBP�. Note that two bands of �40 kDa are often observed, although the lower band predominates (Pohnke et al., 1999; Piwien-Pilipuk etal., 2001). (a1) IGF-1 stimulation maintains low nuclear levels of C/EBP� (lane 2), but levels increase when growth factor support is removed(lane 1) or when L channel or CaMK activities are inhibited (lanes 3 and 4); nimodipine, 5 �M; KN62, 10 �M. Conversely, NMDA receptoractivity increases the levels in a calcineurin-dependent manner: blocking NMDA receptors with 100 �M APV reduces nuclear C/EBP� (lane5), stimulation with 300 �M NMDA � 1 �M glycine increases it (lane 6), but when 1 �M deltamethrin is coapplied, nuclear levels decline (lane7). (a2) Concentration dependence: effectiveness increases dramatically as IGF-1 levels are increased (0 versus 0.5 to 500 ng/ml, 2 hr exposure).(b) L channel dependence. Compared to neurons maintained in standard full serum medium or 4 hr in IGF-1 (left, center), blocking L channelactivity with 5 �M nimodipine raises nuclear C/EBP� levels (right). (c) Direct activation of L channels via depolarization (4 hr) in 25 mM KClalso keeps nuclear levels low (upper), but concurrent L channel inhibition raises them (lower). (d) Assessed immunocytochemically, CaMKIVactivity leads to low nuclear C/EBP� levels. (d1) Overexpressing a constitutively active CaMKIV reduces nuclear levels in serum-starvedneurons, although neighboring untransfected neurons display high levels. Paired panels show the same field, identifying transfectants (anti-GFP) and C/EBP� levels (anti-C/EBP�, transfectants indicated by solid arrowheads). All immunocytochemistry experiments were performed

L Channels and NMDA Receptors Regulate C/EBP�633

indistinguishable from blocking AMPA receptors alone. activity and that the activity remains elevated for hours,providing a potential explanation for how a brief stimulusHowever, blocking AMPA receptors with GYKI 52466 orcan exert long-lasting consequences. Consistent withlow concentrations of NBQX (5 �M; Bureau et al., 1999)studies showing that neuronal death induced by L chan-produced an intermediate level of survival that was nonel block specifically involves caspase-3-dependentlonger sensitive to L channel blockade.cleavage of CaMKIV (See et al., 2001), the effects weNeurons were also briefly exposed to IGF-1 in theobserve appear to be specific to CaMKIV. Survival waspresence of inhibitors (Figure 6B; “�IGF-1”). As pre-not maintained by constitutively active CaMKII, andviously shown (Blair et al., 1999), survival was high fol-IGF-1 effects on CaMKII activity were relatively minor.lowing a 5 min IGF-1 stimulation (“IGF-1 alone”) butInterestingly, however, we find that elevating the levelswere much lower when L channels were blocked byof wild-type CaMKIV had little effect on survival at mod-nimodipine or Cd2� during the IGF-1 stimulation. N chan-erate-high levels of IGF-1, but a large effect at lowernel and sodium channel block, both of which shouldIGF-1 concentrations or lower levels of L channel activ-reduce neurotransmission, exerted prosurvival effectsity. Although altering CaMKIV levels cannot be consid-(“-CgTx-GVIA,” “TTX”). Here, however, blocking NMDAered a natural situation, this result and the findings thatreceptors produced no significant increase in survivalantisense and siRNA to C/EBP� can rescue neurons(“APV”), suggesting that their prodeath actions might befrom excitotoxic death suggest possible biomedical ap-compensated by growth factor support. Nonetheless,plications.an L channel-dependent component was maintained

Multiple steps appear to be calcium dependent.(“APV�nim”). Finally, AMPA receptor block again pro-CaMKIV is indirectly calcium sensitive via CaM, but theduced intermediate levels of survival and occluded theconstitutively active CaMKIV that promoted survival waseffects of L channel block, implying that AMPA receptorstruncated to remove the autoinhibitory and CaM bindingare involved in both the prosurvival and prodeath path-domains (Chatila et al., 1996), thereby eliminating theways. Moreover, our data indicate that L channel andproximal requirement for calcium. Nevertheless, blockingNMDA receptor activities antagonistically compete toL channel activity while expressing ca-CaMKIV de-control survival and support the thesis that the mode ofcreased survival, indicating that basal levels of L channelcalcium entry is essential to determining the net effect:activity may be sufficient for other calcium-requiringsurvival or death (Figure 7).steps. Potential targets include CaMKK, which requiresCaM binding and phosphorylates CaMKIV to increase

Discussion its activity, and calcium-dependent transcription factorssuch as CREB and MEF2, both of which are essential

IGF-1 is known to promote neuronal survival via CREB- for depolarization- and neurotrophin-mediated neuronalmediated transcription (Brunet et al., 2001). Our data survival (Tao et al., 1998; Bonni et al., 1999; Hu et al.,demonstrate that IGF-1 also acts through elevated L 1999; Mao et al., 1999; Rajadhyaksha et al., 1999;channel-dependent calcium influx, resulting first in in- Blaeser et al., 2000; Mao and Wiedmann, 2000; Gongcreased CaMKIV activity and subsequently in reduced et al., 2003). Recent work further demonstrates that CaMnuclear levels of the transcriptional activator C/EBP�. binding to L channels is required for CREB activationConversely, NMDA receptor-mediated calcium influx in- (Dolmetsch et al., 2001), and that MEF2 is sensitive toduces excitotoxic death, at least in part via the calcium- L channel-mediated calcium influx (Mao and Wiedmann,dependent phosphatase calcineurin and strongly ele- 2000), with calcium and CaM relieving repression (Younvated nuclear C/EBP� levels. et al., 2000).

Previously, we showed that IGF-1 improves neuronal Another family of transcription factors, the CCAATsurvival by potentiating L channel activity and that this enhancer binding proteins (C/EBPs), can be regulatedwas surprisingly rapid: a 1 min exposure to IGF-1 pro- by CaMKIV activity (Yukawa et al., 1998). Typically,moted survival in an L channel-dependent manner (Blair C/EBPs have been associated with survival in nonneu-and Marshall, 1997; Blair et al., 1999). We now find that ronal cells (Buck et al., 2001; Zhu et al., 2002). In theCaMKIV is an essential component of this pathway. We PC12 line, C/EBP� activity is increased by NGF, a pro-

survival factor (Sterneck and Johnson, 1998), althoughalso determined that IGF-1 rapidly increases CaMKIV

a minimum of 6 times. (d2) In contrast, even though nuclear C/EBP� levels are low in untransfected neurons in IGF-1, neurons overexpressingdn-CaMKIV maintain high levels. (d3) Similarly, the general CaMK inhibitor, KN62 (10 �M), fully blocks the ability of IGF-1 to maintain lownuclear C/EBP� levels.(B) Nuclear C/EBP� levels rise within 1 min of withdrawing growth factor support (lanes 1–4), but subsequent IGF-1 addition reduces them(lanes 5–6).(C) After 1 day in 0-serum, both antisense ODN and siRNA keep nuclear C/EBP� levels low. Conversely, levels are high when neurons aretreated with either the scrambled antisense ODN or nonblocking siRNA, or maintained in the 0-serum vehicle.(D) IGF-1 reverses serum deprivation- and NMDA-induced nuclear localization of C/EBP� in a CRM1-dependent manner. Without growthfactor support (first panel, 3 h no IGF-1), nuclear C/EBP� levels rise, but subsequent addition of IGF-1 reduces them (second panel, after 3hr in 0-serum, 50 ng/ml IGF-1 was added for another 4 hr). When the nuclear transport protein CRM1 was blocked by treatment with leptomycinB, nuclear levels remained high in IGF-1 (third panel, same as the second panel except that 20 ng/ml leptomycin B was added after thesecond hour in 0-serum and was present during the subsequent IGF-1 exposure). Similarly, IGF-1 can at least partially reverse NMDA-inducedrises in nuclear C/EBP� levels (fourth panel, see also Supplemental Figure S4) in a CRM1-dependent manner (fifth panel). Here, 300 �MNMDA � 1 �M glycine was added to 0-serum medium at the beginning of the experiment; otherwise, neurons were treated identically tothose in the second and third panels.

Neuron634

Figure 6. NMDA and AMPA Receptors Exert Differential Effects on Neuronal Survival: NMDA Receptor Activity Promotes Death while AMPAReceptors Appear to Play a Pivotal Role, Promoting Both L Channel-Mediated Survival and NMDA Receptor-Mediated Death

(A) L channel plus moderate AMPA activity strongly promote survival, and L channel agonists oppose NMDA receptor-induced death. First,sister cultures of neurons were exposed to the L channel agonist, (-)-BayK8644 (0.8 �M), and increasing concentrations of the AMPA receptoragonist, kainate (0–100 �M), for 24 hr. BayK8644 alone (“0 KA”) elevated survival over that in unsupplemented test medium (“No add”), butcoaddition of moderate concentrations of kainate (10 �M, 30 �M) strongly improved it; asterisk indicates survival significantly different fromBayK8644 alone (p � 0.005, Student’s t test). In some cases, APV (100 �M) or nimodipine (5 �M) was also included to inhibit, respectively,endogenous NMDA receptor or L channel activity: NMDA receptor block also improved survival relative to the same kainate concentrationswithout APV (***p � 0.005) and L channel block consistently reduced it (p � 0.005). When neither BayK8644 nor APV were added, 10 �Mkainate improved survival compared to “No add,” but coaddition of nimodipine decreased it, as did high (100 �M) kainate. Alternatively, whenNMDA receptor activity was stimulated for 30 min with 300 �M NMDA � 1 �M glycine, few neurons were alive after 1 day. However,simultaneous exposure to 10 �M kainate significantly ameliorated survival (**p � 0.005, 15% versus 32%) in an L channel-dependent manner(**p � 0.005, 32% versus 14% when nimodipine was present). Again, coadding 100 �M kainate with NMDA reduced survival to near zero,but coadding 0.8 �M BayK8644 improved it (�, p � 0.005 versus NMDA alone); this was partially diminished by inhibiting endogenous AMPAreceptor activity with 40 �M CNQX (p � 0.025). Values are means � SEMs. n/test combination � �4500 neurons (range, 3040–6126) from 5independent experiments.(B) Pharmacological isolation of specific glutamate receptor subtypes. Sister cultures of neurons were maintained for 1 day in test media �specific inhibitors (No IGF-1). Alternatively, neurons were first briefly exposed to IGF-1 (5 min, 50 ng/ml) in the presence of the inhibitors,washed to remove IGF-1, and then maintained in the inhibitors for 1 day (�IGF-1). Inhibitors, paired �L channel block with 5 �M nimodipine,were: TTX, 1 �M; APV, 100 �M; GYKI 52466, 50 �M; CNQX, 40 �M; NBQX, 5 �M, 100 �M; and -conotoxin-GVIA, 400 nM (“-CgTx-GVIA”).(One exception: 100 �M Cd2� was present only during the IGF-1-stimulation because extended N channel block was found to alter survival.)Asterisk indicates when L channel block resulted in significantly lower survival (p � 0.005); no difference was detectable between L channelblock by nimodipine or brief exposure to Cd2� (p � 0.6). Summarized, NMDA receptor block (“APV,” “APV�nim”) improved survival, whichstill contained an L channel-dependent component, indicating that NMDA receptors do not stimulate L channel activity. In contrast, blockingAMPA receptors (“GYKI,” “GYKI�nim,” and “5 �M NBQX,” “5 �M NBQX�nim”) produced smaller overall effects on survival, but also eliminatedL channel dependence. This suggests that AMPA receptor activity stimulates both voltage-sensitive L channels (prosurvival) and NMDAreceptors (prodeath). Kainate receptors played no detectable role: blocking both AMPA and kainate receptors was indistinguishable from

L Channels and NMDA Receptors Regulate C/EBP�635

Figure 7. Glutamate Receptor Activity and IGF-1 Potentiation of Calcium Channel Subtypes May Perform Dual and Antagonistic Functions

Schema based on: (1) IGF-1 acts through PI3K, Akt, and src to potentiate neuronal L channel activity, and L channel potentiation promotessurvival (Blair and Marshall, 1997; Blair et al., 1999; Bence-Hanulec et al., 2000); (2) IGF-1 rapidly stimulates CaMKIV activity in an L channel-dependent manner; (3) CaMKIV promotes survival; (4) IGF-1, L channel, and CaMKIV activities reduce nuclear C/EBP� levels; (5) NMDAreceptor activity increases nuclear C/EBP� levels and reduces survival, both being calcineurin dependent; (6) AMPA receptor activity stimulatesL channels to improve survival; and (7) AMPA receptors relieve Mg2� block of NMDA receptors (Nowak et al., 1984), thereby also contributingto the NMDA receptor prodeath pathways. Narrow arrows indicate movement of signaling components between intracellular compartments.

it has been suggested that C/EBP� may be proapoptotic In cerebellar neurons, we find that nuclear C/EBP�levels are reduced by IGF-1 and L channel activity by ain Neuro2A cells (Cortes-Canteli et al., 2002). In neurons,

functions unrelated to survival were also recently de- process dependent upon CaMKIV and the nuclear ex-port protein, CRM1. These results raise the possibilityscribed, with C/EBP� being associated developmentally

with biasing cortical precursors to neurogenesis (Men- that CaMKIV may directly phosphorylate C/EBP� to in-duce its translocation. A potentially similar mechanismard et al., 2002) and, in mature animals, with long-term

memory (Taubenfeld et al., 2001a, 2001b). We predicted, regulates the prodeath Forkhead transcription factorsFOXO and AFX: Akt, which promotes survival by multiplebased on the prosurvival effect of C/EBPs on a variety

of cell types, that C/EBP� would act similarly in neurons. means including L channel potentiation (Blair et al.,1999), negatively regulates FOXO by eliciting transloca-We found, however, that overexpressing dominant-neg-

ative C/EBP strongly elevated survival in the absence tion out of the nucleus (Brunet et al., 2001) and AFX bypreventing nuclear import (Brownawell et al., 2001). Weof IGF-1. To establish if the effect was specifically due

to C/EBP�, we additionally tested if wild-type C/EBP� also find that when nuclear levels decrease, C/EBP�is localized in large perinuclear puncta. Although thewas proapoptotic and if antisense oligonucleotides and

siRNA specific to C/EBP� were prosurvival. Finding that function of this distribution remains to be determined,possibilities include that the protein is being held forall three approaches indicated that C/EBP� might be

proapoptotic, we analyzed potential mechanisms. relocation back into the nucleus or is being targeted fordegradation. The first possibility is particularly attractiveRegulation of C/EBPs is complex. It is known that

they can act as either transcriptional enhancers or re- in light of our observation that, in a calcineurin-depen-dent manner, nuclear C/EBP� levels rapidly rise underpressors, that individual subtypes can heterodimerize

with other members, potentially altering DNA binding conditions that induce excitotoxic death.Survival is a homeostatic process, but how neuronsspecificities or transactivating capabilities, and that for

C/EBP�, dominant-negative isoforms occur naturally balance stimuli is not well understood. Our data suggestthat L channels and glutamate receptors perform antag-(Xiong et al., 2001). It is also known that C/EBPs contain

nuclear localization signals (NLS). In nonneural cells, onistic functions. IGF-1, direct L channel stimulation,and blocking NMDA receptor activity all improve survivalTNF can induce rapid nuclear export of C/EBP� via

phosphorylation of a specific serine residue within the and reduce nuclear C/EBP� levels, while NMDA receptorstimulation or blocking L channel activity decrease sur-NLS (Yin et al., 1996; Williams et al., 1997; Buck et al.,

2001). vival but increase nuclear C/EBP�. We also find that N

blocking AMPA receptors alone. When sodium-dependent action potential activity was blocked (“TTX”), results roughly similar to NMDAreceptor block were obtained; note that any calcium-dependent action potentials (D’Angelo et al., 1997) would have been unaffected in TTXalone. Likewise, blocking N channel activity (“-CgTx-GVIA”), which would be expected to reduce neurotransmission, increased survival, butL channel activity was still required to maximize it (“-CgTx-GVIA�nim”). Values are means � SEMs. n/test combination � �3000 neurons(range, 1846–4509) from 3 independent experiments.

Neuron636

channel blockade, which would be predicted to reduce lating IRS-1, the docking protein mediating IGF-1 recep-tor responses (Hallak et al., 2001).neurotransmission, reduces nuclear levels and pro-

motes survival, implying that IGF-1, which potentiates C/EBP� appears to be another pivot point. We showhere that both the excitotoxic and L channel pathwaysN as well as L channel activity (Blair and Marshall, 1997),

may serve a more complicated role than simply acting regulate its subcellular localization. IGF-1 potentiationof L channels acts through CaMKIV to promote survivalas a survival factor. Moreover, given that NMDA recep-

tors also mediate calcium influx, it is especially striking by reducing nuclear C/EBP� levels, and NMDA receptoractivity acts antagonistically through calcineurin to in-that the mode of calcium entry differentially regulates

survival processes. L channel activity and IGF-1 stimu- crease them.late CREB phosphorylation and activity (Rajadhyaksha

Experimental Procedureset al., 1999), while extrasynaptic NMDA receptor activityleads to CREB dephosphorylation (Sala et al., 2000;

Cell CultureHardingham et al., 2002). Our results also imply differen- Cerebellar granule neurons were cultured from P5–P6 rats (Blair ettial synaptic effects: moderate levels of AMPA receptor al., 2001). Standard medium was DMEM with 10% fetal bovine serumactivity appear to stimulate L channels to improve sur- (GIBCO-BRL), penicillin/streptomycin, 20 �M fluorodeoxyuridine to

arrest division of nonneuronal cells, 25 mM KCl, and 6 g/l glucose.vival, whereas higher levels predominantly act to stimu-Test media were without serum and contained 5 mM KCl.late NMDA receptor-mediated death. The concept that

the conditions of activation are important is underscoredExpression Vectors and Neuronal Transfection

by the multiple functions of NMDA receptors, L chan- Granule neurons were transfected with full-length wild-type C/EBP�;nels, and C/EBP�: all are involved in learning and mem- dominant-negative C/EBP (Olive et al., 1996); wild-type, catalyticallyory as well as cellular survival. inactive, or constitutively active CaMKIV (Chatila et al., 1996); consti-

tutively active CaMKK; or constitutively active CaMKII (Sun et al.,Mechanisms underlying differential calcium signaling1994). Note that the dn-CaMKIV is unlikely to act nonspecifically: itare likely to involve subcellular compartmentalization ofwas rendered catalytically inactive by a single-point mutationintermediates. Studies have demonstrated that L chan-(K75E), making it almost identical to wt-CaMKIV. All cDNAs were in

nel activity need not increase bulk cytosolic calcium to pSG5 expression vectors (Stratagene). Neurons, 2–3 days in culture,induce transcription and imply that the initial step may were cotransfected in 2 ml media with 0.5–1 �g of the cDNA ofbe rapid coupling of incoming calcium to calcium bind- interest or plus 0.5 �g SuperGlo-GFP cDNA (Quantum Biotechnolo-

gies) to identify transfectants (Blair et al., 1999, 2001).ing proteins closely associated with the channel (Deis-seroth et al., 1996, 1998). CaM is known to bind to L

Oligodeoxynucleotide, siRNA, and Transfectionchannels (Peterson et al., 1999) and to be capable ofAn antisense C/EBP� ODN [CCA GCA GGC GGT GCA TGA AC]nuclear translocation (Deisseroth et al., 1998; Wang et directed against the extreme 5� end and the first in-frame AUG of

al., 2000; Mermelstein et al., 2001). Interestingly, L chan- the rat C/EBP-� transcript (Taubenfeld et al., 2001a) was synthe-nel potentiation is greatest at moderately hyperpolarized sized (Life Technologies). Scrambled antisense C/EBP� [TCG GAG

ACT AAG CGC GGC AC] served as control and was of identicalpotentials (e.g., �40 mV; Blair and Marshall, 1997;length, containing in scrambled form the same A/T C/G composition;Bence-Hanulec et al., 2000), and activating L channelsit also showed no homology to any sequence in the GenBank data-via excitatory postsynaptic potentials (i.e., moderatelybase. ODNs had phosphorothioate linkage on the three terminalhyperpolarized potentials) is more effective than action bases of the 5� and 3� ends and phosphodiester internal bonds

potentials for promoting CREB phosphorylation (Mer- because such ODNs retain biochemical specificity, are more stablemelstein et al., 2000). than the unmodified phosphodiester ODNs in vivo, and are less

toxic than the full phosphorothioate ODNs (Ogawa and Pfaff, 1998).Although compartmentalization offers a means forODN transfection was performed according to manufacturer’s in-separating calcium-dependent survival and deathstructions using Oligofectamine reagents (Life Technologies); finalmechanisms, extensive possibilities for crosstalk alsoconcentrations were 300–400 nM. Efficiency was assessed after 1exist. Our data, summarized in Figure 7, suggest that day in serum-free medium (Figure 5C, Supplemental Figure S5 at

CaMKIV may be a shared target of the pro- and antiapo- http://www.neuron.org/cgi/content/full/39/4/625/DC1).ptotic pathways. Calcineurin can deactivate CaMKIV C/EBP� siRNA [AAG CTG AGC GAC GAG TAC AAG] and non-

blocking sequence [AAG TGG CCA ACT TCT ACT ACG] were de-(Enslen et al., 1994) and calpain cleaves it (McGinnis etsigned according to manufacturer’s recommendations (Quiagen-al., 1998). Interestingly, the role(s) of calcineurin are notXeragon). The blocking sequence has no homology to any other inentirely clear. We and others find that inhibition of cal-the GenBank database. Transfections were performed per manufac-

cineurin promotes neuronal survival and opposes exci- turer’s instructions (Quiagen-Xeragon) and efficiency assessed aftertotoxicity (Dawson et al., 1993; Ankarcrona et al., 1996; 1 day (Figure 5C).McDonald et al., 1996; Wang et al., 1999; Ruiz et al.,

Survival Assays2000). However, under some conditions, NMDA receptorNeurons were cultured in standard medium for 2 days, then rinsed,activity promotes survival, a disparity that may be duesubjected to manipulation, and grown in test media for 1 day (Blairto activating extrasynaptic versus synaptic receptorset al., 1999). Primary media were: test medium (0-serum with 5 mM

(Sinor et al., 2000; Hardingham et al., 2002). Calcineurin KCl), or test medium with 50 ng/ml IGF-1 with or without the Lhas also been reported both to reduce and enhance the channel inhibitor, nimodipine. For each experiment, test media wereactivity of neuronal L channels (Hernandez-Lopez et al., assayed in quadruplicate on sister cultures.

For transient pharmacological inhibition of CaMKs, the general2000; Norris et al., 2002), and it can control prosurvivalCaMK inhibitor KN62 was employed at a concentration (10 �M)transcription factors, increasing the activity of MEF2where it is known to block CaMKIV (Enslen et al., 1994). Prior to(Mao and Wiedmann, 2000) and inducing nuclear importexposure to IGF-1, neurons were pretreated for 15 min with KN62

of NFAT (Crabtree and Olson, 2002). Conversely, cal- or its vehicle (0.1% DMSO), then briefly stimulated with IGF-1 (50cineurin activated by NMDA receptors may negatively ng/ml for 1 min) in the presence of inhibitor or vehicle. IGF-1 was

then removed by washing 3� in basic test medium (complete bathregulate the IGF-1 survival pathways by dephosphory-

L Channels and NMDA Receptors Regulate C/EBP�637

exchange), and KN62 or vehicle was readded for an additional 2 hr. observed in full serum as 1.0 and expressing the experimental valuesas fold-change. When the reference was serum-starved neurons,Neurons were washed again to remove inhibitor and replaced in

basic test medium for another 22 hr. values were calculated as percentage-change relative to that ob-served in 0-serum (100%).To determine apoptosis, the primary assays were: propidium io-

dide (PI) labeling to reveal condensation state of the chromatin andthe TUNEL assay to reveal chromatin cleavage; in some cases, the Immunocytochemistrynuclear stain DAPI (500 nM) was employed. For PI labeling, cells Cells were fixed in 1.5% PFA, permeabilized in 0.1% Triton X-100were fixed (1.5% PFA) and permeabilized with 0.1% Triton X-100, and, following block of nonspecific sites in 4% goat serum (Geminiand the chromatin was stained with 2 �g/ml PI. Neurons were Bio-Products), labeled for indirect immunofluorescence. Primary an-counted “blind” on a Zeiss Axioskop equipped for epifluorescence tibodies were anti-C/EBP� (rabbit polyclonal sc-150, Santa Cruzusing rhodamine optics to visualize the PI and GFP filter settings Biotechnology, 1:1000 dilution), anti-FLAG epitope (M2 mouseto identify transfectants. All transfected neurons in a culture were monoclonal anti-FLAG, Sigma, 2 �g/ml), anti-pCREB (rabbit poly-counted; for nontransfectants, all neurons in four randomly chosen clonal #06-519, Upstate Biotechnology, 1:500), Tuj1 (Covance Re-fields were assessed. The survival of transfectants was then com- search Products #MMS-435P, 1:500), and anti-GFP (rabbit poly-pared to nontransfectants within each culture; comparison within clonal “living color peptide Ab,” Clontech, 1:50, or mousecultures ensures that all neurons will be handled identically and that monoclonal #AFP5002, Q-Biogene, 1:100). Secondary antibodiesthe sole difference will be whether a neuron is overexpressing the were anti-mouse IgG-FITC (#115-095-071, Jackson ImmunoRe-exogenous gene products. For the TUNEL assay, fixed cultures were search Lab, 1:600), anti-mouse IgG-rhodamine (#115-295-116, Jack-exposed to a 1:2 solution of ethanol:acetic acid to quench GFP son ImmunoResearch Lab, 1:100), and anti-rabbit IgG-rhodaminefluorescence and processed as per manufacturer’s instructions us- Red-X (#111-295-144, Jackson ImmunoResearch Lab, 1:100). Neu-ing the In Situ Cell Death Detection kit (Boehringer Mannheim). GFP- rons were visualized on a Zeiss Axiovert 100 laser scanning confocalcontaining neurons were then identified by indirect immunolabeling microscope or a Zeiss Axioskop equipped for epifluorescence.using a rhodamine-conjugated secondary. Labeled (i.e., cleaved)DNA of individual cells was visualized under fluorescein optics. Con- Acknowledgmentstrols included TUNEL-positive tissue (Boehringer Mannheim), a neg-ative control (omission of the reaction enzyme), and omission of the For all her generous help and support, we particularly wish to ac-anti-GFP immunolabeling to ensure that GFP fluorescence was fully knowledge Dr. Cristina Alberini. Additionally, for their generous gifts,quenched. For ODN and siRNA experiments, all neurons were con- we thank Drs. C. Kane and A. Means and Dr. J. Yin (CaMKIV andsidered to be possible transfectants because the efficiencies of CaMKII constructs, respectively), Drs. A. Nairn and M. Picciottotransfection are very high (Lakkaraju et al., 2001; Krichevsky and (constitutively active CaMKK construct), and Dr. C. Vinson (domi-Kosik, 2002). nant-negative C/EBP construct). We also thank Dr. V. Kumaresan for

generously taking time from her experiments to provide additionalIGF-1 Regulation of CaMKII and CaMKIV Activity Assays granule neuron cultures, Dr. G. Pei for help with the TUNEL assays,Immunoprecipitated CaMKII and -IV were assayed in vitro using and Dr. G. Moss for kindly reviewing the manuscript. This work wasspecific substrates as per Yoshida et al. (2000): cultures were supported by NIH RO1 NS37676 and COBRE P20 RR 15578 toswitched to test medium for 16 hr, then stimulated with 50 ng/ml L.A.C.B., AHA 9940131N and R01 NS39063-01A2 to J.M., and NIHIGF-1 or an equal volume of test medium alone, then washed, lysed HD39446 to Z.M.on ice for 30 min in the presence of protease and phosphataseinhibitors, passed through a syringe to disrupt nuclear membranes, Received: May 28, 2003and centrifuged to remove DNA. CaMKII and -IV were specifically Revised: June 4, 2003immunoprecipitated (anti-CaMKIV: mAb clone 26, Transduction Accepted: June 30, 2003Laboratories, and rabbit pAb M-176, Santa Cruz Biotechnology; Published: August 13, 2003anti-CaMKII goat pAb M-20, Santa Cruz Biotechnology), collectedon protein G-sepharose beads, and washed 2� with assay buffer References(25 mM �-glycerophosphate, 1 mM Na-orthovanadate, 1 mM DTT,1 mM CaCl2, 20 mM MOPS [pH 7.2]). Activities were assayed in vitro Ankarcrona, M., Dypbukt, J.M., Orrenius, S., and Nicotera, P. (1996).(20 min, 30C in the presence of 1 �M PKC inhibitor peptide and 1 �M

Calcineurin and mitochondrial function in glutamate-induced neu-PKA inhibitor peptide) using [�-32P]ATP and peptide � as a CaMKIV-

ronal cell death. FEBS Lett. 394, 321–324.specific substrate and autocamtide II as a substrate for CaMKII.

Bence-Hanulec, K.K., Marshall, J., and Blair, L.A.C. (2000). Potentia-After incubation, 10 �l aliquots were spotted on phosphocellulosetion of neuronal L calcium channels by IGF-1 requires phosphoryla-discs, which were washed, dried, and subjected to scintillationtion of the 1 subunit on a specific tyrosine residue. Neuron 27,counting. Background was estimated by performing identical proce-121–131.dures in the absence of substrate.Blaeser, F., Ho, N., Prywes, R., and Chatila, T.A. (2000). Ca2�-depen-dent gene expression mediated by MEF2 transcription factors. J.Western Blot Analysis of Subcellular FractionsBiol. Chem. 275, 197–209.Neurons were exposed to 0-serum or IGF-1-containing test medium,Blair, L.A.C., and Marshall, J. (1997). IGF-1 modulates N and L cal-then lysed, and cytoplasmic and nuclear fractions were isolated percium channels in a PI 3-kinase-dependent manner. Neuron 19,manufacturer’s instructions using NE-PER Nuclear and Cytoplasmic421–429.Extraction Reagents (Pierce). Protein content was estimated using

BioRad Protein Assay (BioRad Laboratories). Equal amounts of total Blair, L.A.C., Bence-Hanulec, K.K., Mehta, S., Franke, T.F., Kaplan,protein, typically 10–20 �g/lane, were resolved on denaturing 10% D., and Marshall, J. (1999). Akt-dependent potentiation of L channelsSDS-PAGE gels, then transferred to Hybond ECL nitrocellulose by insulin-like growth factor-1 is required for neuronal survival. J.membranes pretreated with 5% BLOTTO buffer (Amersham Phar- Neurosci. 19, 1940–1951.macia Biotechnology). Membranes were incubated with an anti-C/

Blair, L.A.C., Bence-Hanulec, K.K., and Marshall, J. (2001). GFP inEBP� rabbit polyclonal antibody (sc-150, Santa Cruz Biotechnology,

the study of neuronal signaling pathways. Curr. Protocols Neurosci.,1:200 dilution) for 6 hr, washed, treated with secondary HRP-labeled

vol. 5.16.goat anti-rabbit antibody (R14745, Transduction Laboratories,

Bonni, A., Brunet, A., West, A.E., Datta, S.R., Takasu, M.A., and1:1000) for 1 hr, washed again, incubated with HRP-streptavidinGreenberg, M.E. (1999). Cell survival promoted by the Ras-MAPKcomplex and ECL detection reagents (SuperSignal West Pico,signaling pathway by transcription-dependent and -independentPierce), and exposed to BioMax MS film (Eastman Kodak). Reactionmechanisms. Science 286, 1358–1362.product levels were quantified by scanning densitometry (NIH Image

1.62). When the reference was neurons in full serum medium, any Brownawell, A.M., Kops, G.J., Macara, I.G., and Burgering, B.M.(2001). Inhibition of nuclear import by protein kinase B (Akt) regulateschanges were calculated within each experiment, defining the level

Neuron638

the subcellular distribution and activity of the forkhead transcription Johnson, P.F., and Unterman, T.G. (2001b). Insulin suppressestransactivation by C/EBP�: signaling to p300/CBP by protein kinasefactor AFX. Mol. Cell. Biol. 21, 3534–3546.B disrupts interaction with the major activation domain of C/EBP�.Brunet, A., Datta, S.R., and Greenberg, M.E. (2001). Transcription-J. Biol. Chem. 276, 8516–8523.dependent and -independent control of neuronal survival by the

PI3K-Akt signaling pathway. Curr. Opin. Neurobiol. 11, 297–305. Hallak, H., Ramadan, B., and Rubin, R. (2001). Tyrosine phosphoryla-tion of insulin receptor substrate-1 (IRS-1) by oxidant stress in cere-Buck, M., Zhang, L., Halasz, N.A., Hunter, T., and Chojkier, M. (2001).bellar granule neurons: modulation by N-methyl-D-aspartateNuclear export of phosphorylated C/EBP� mediates the inhibitionthrough calcineurin activity. J. Neurochem. 77, 63–70.of albumin expression by TNF-. EMBO J. 20, 6712–6723.

Hardingham, G.E., Fukunaga, Y., and Bading, H. (2002). Extrasynap-Bureau, I., Bischoff, S., Heinemann, S.F., and Mulle, C. (1999). Kai-tic NMDARs oppose synaptic NMDARs by triggering CREB shut-nate receptor-mediated responses in the CA1 field of wild-type andoff and cell death pathways. Nat. Neurosci. 5, 405–414.GluR6-deficient mice. J. Neurosci. 19, 653–663.

Chatila, T., Anderson, K.A., Ho, N., and Means, A.R. (1996). A unique Henderson, B.R., and Eleftheriou, A. (2000). A comparison of theactivity, sequence specificity, and CRM1-dependence of differentphosphorylation-dependent mechanism for the activation of Ca2�/

calmodulin-dependent protein kinase type IV/GR. J. Biol. Chem. nuclear export signals. Exp. Cell Res. 256, 213–224.271, 21542–21548. Hernandez-Lopez, S., Tkatch, T., Perez-Garci, E., Galarraga, E., Bar-Chawla, S., Hardingham, G.E., Quinn, D.R., and Bading, H. (1998). gas, J., Hamm, H., and Surmeier, D.J. (2000). D2 dopamine receptorsCBP: a signal-regulated transcriptional coactivator controlled by in striatal medium spiny neurons reduce L-type Ca2� currents andnuclear calcium and CaM kinase IV. Science 281, 1505–1509. excitability via a novel PLC�1-IP3-calcineurin-signaling cascade. J.

Neurosci. 20, 8987–8995.Cortes-Canteli, M., Pignatelli, M., Santos, A., and Perez-Castillo, A.(2002). CCAAT/enhancer-binding protein � plays a regulatory role Hu, S.-C., Chrivia, J., and Ghosh, A. (1999). Regulation of CBP-in differentiation and apoptosis of neuroblastoma cells. J. Biol. mediated transcription by neuronal calcium signaling. Neuron 22,Chem. 277, 5460–5467. 799–808.Crabtree, G.R., and Olson, E.N. (2002). NFAT signaling: choreo- Krichevsky, A.M., and Kosik, K.S. (2002). RNAi functions in culturedgraphing the social lives of cells. Cell 109, S67–S79. mammalian neurons. Proc. Natl. Acad. Sci. USA 99, 11926–11929.Curtis, J., and Finkbeiner, S. (1999). Sending signals from the syn- Lakkaraju, A., Dubinsky, J.M., Low, W.C., and Rahman, Y.E. (2001).apse to the nucleus: possible roles for CaMK, Ras/ERK, and SAPK Neurons are protected from excitotoxic death by p53 antisensepathways in the regulation of synaptic plasticity and neuronal oligonucleotides delivered in anionic liposomes. J. Biol. Chem. 276,growth. J. Neurosci. Res. 58, 88–95. 32000–32007.D’Angelo, E., De Filippi, G., Rossi, P., and Taglietti, V. (1997). Synap- Mao, Z., and Wiedmann, M. (2000). Calcineurin enhances MEF2 DNAtic activation of Ca2� action potentials in immature rat cerebellar binding activity in calcium-dependent survival of cerebellar granulegranule cells in situ. J. Neurophysiol. 78, 1631–1642. neurons. J. Biol. Chem. 274, 31102–31107.Dawson, T.M., Steiner, J.P., Dawson, V.L., Dinerman, J.L., Uhl, G.R.,

Mao, Z., Bonni, A., Xia, F., Nadal-Vicens, M., and Greenberg, M.E.and Snyder, S.H. (1993). Immunosuppressant FK506 enhances

(1999). Neuronal activity-dependent cell survival mediated by tran-phosphorylation of nitric oxide synthase and protects against gluta-

scription factor MEF2. Science 286, 785–790.mate neurotoxicity. Proc. Natl. Acad. Sci. USA 90, 9808–9812.

McDonald, J.W., Goldberg, M.P., Gwag, B.J., Chi, S.I., and Choi,Deisseroth, K., Bito, H., and Tsien, R.W. (1996). Signaling from syn-D.W. (1996). Cyclosporine induces neuronal apoptosis and selectiveapse to nucleus: postsynaptic CREB phosphorylation during multi-oligodendrocyte death in cortical cultures. Ann. Neurol. 40, 750–758.ple forms of hippocampal synaptic plasticity. Neuron 16, 89–101.McGinnis, K.M., Whitton, M.M., Gnegy, M.E., and Wang, K.K.W.Deisseroth, K., Heist, E.K., and Tsien, R.W. (1998). Translocation(1998). Calcium/calmodulin-dependent protein kinase IV is cleavedof calmodulin to the nucleus supports CREB phosphorylation inby caspase-3 and calpain in SH-SY5Y human neuroblastoma cellshippocampal neurons. Nature 392, 198–202.undergoing apoptosis. J. Biol. Chem. 273, 19993–20000.

Descombes, P., and Schibler, U. (1991). A liver-enriched transcrip-Mellor, J.R., Merlo, D., Jones, A., Wisden, W., and Randall, A.D.tional activator protein, LAP, and a transcriptional inhibitory protein,(1998). Mouse cerebellar granule cell differentiation: electrical activ-LIP, are translated from the same mRNA. Cell 67, 569–579.ity regulates the GABAA receptor 6 subunit gene. J. Neurosci. 18,Dolmetsch, R.E., Pajvani, U., Fife, K., Spotts, J.M., and Greenberg,2822–2833.

M.E. (2001). Signaling to the nucleus by an L-type calcium channel-Menard, C., Hein, P., Paquin, A., Savelson, A., Yang, X.M., Lederfein,calmodulin complex through the MAP kinase pathway. Science 294,D., Barnabe-Heider, F., Mir, A.A., Sterneck, E., Peterson, A.C., et al.333–339.(2002). An essential role for a MEK-C/EBP pathway during growthDudek, H., Datta, S.R., Franke, T.F., Birnbaum, M.J., Yao, R., Cooper,factor-regulated cortical neurogenesis. Neuron 36, 597–610.G.M., Segal, R.A., Kaplan, D.R., and Greenberg, M.E. (1997). Regula-Mermelstein, P.G., Bito, H., Deisseroth, K., and Tsien, R.W. (2000).tion of neuronal survival by the serine-threonine protein kinase Akt.Critical dependence of cAMP response element-binding proteinScience 275, 661–665.phosphorylation on L-type calcium channels supports a selectiveEnslen, H., Sun, P., Brickey, D., Soderling, S.H., Klamo, E., andresponse to EPSPs in preference to action potentials. J. Neurosci.Soderling, T.R. (1994). Characterization of Ca2�/calmodulin-depen-20, 266–273.dent protein kinase IV. Role in transcriptional regulation. J. Biol.Mermelstein, P.G., Deisseroth, K., Dasgupta, N., Isaksen, A.L., andChem. 269, 15520–15527.Tsien, R.W. (2001). Calmodulin priming: nuclear translocation of aGalli, C., Meucci, O., Scorziello, A., Werge, T.M., Calissano, P., andcalmodulin complex and the memory of prior neuronal activity. Proc.Schettini, G. (1995). Apoptosis in cerebellar granule cells is blockedNatl. Acad. Sci. USA 98, 15342–15347.by high KCl, forskolin, and IGF-1 through distinct mechanisms of

action: the involvement of intracellular calcium and RNA synthesis. Misra, R.P., Bonni, A., Miranti, C.K., Rivera, V.M., Sheng, M., andJ. Neurosci. 15, 1172–1179. Greenberg, M.E. (1994). L-type voltage-sensitive calcium channel

activation stimulates gene expression by a serum response factor-Gong, X., Tang, X., Wiedmann, M., Wang, X., Peng, J., Zheng, D.,dependent pathway. J. Biol. Chem. 269, 25483–25493.Blair, L.A.C., Marshall, J., and Mao, Z. (2003). Cdk-5-mediated inhibi-

tion of the protective effects of transcription factor MEF2 in neuro- Niehof, M., Kubicka, S., Zender, L., Manns, M.P., and Trautwein, C.toxicity-induced apoptosis. Neuron 38, 33–46. (2001). Autoregulation enables different pathways to control CCAAT/

enhancer binding protein � (C/EBP�) transcription. J. Mol. Biol. 309,Guo, X., Dillman, J.F., Dawson, V.L., and Dawson, T.M. (2001a).855–868.Neuroimmunophilins: novel neuroprotective and neuroregenerative

targets. Ann. Neurol. 50, 6–16. Norris, C.M., Blalock, E.M., Chen, K.C., Porter, N.M., and Landfield,P.W. (2002). Calcineurin enhances L-type Ca2� channel activity inGuo, S., Cichy, S.B., He, X., Yang, Q., Ragland, M., Ghosh, A.K.,

L Channels and NMDA Receptors Regulate C/EBP�639

hippocampal neurons: increased effect with age in culture. Neuro- Wang, H.G., Pathan, N., Ethell, I.M., Krajewski, S., Yamaguchi, Y.,Shibasaki, F., McKeon, F., Bobo, T., Franke, T.F., and Reed, J.C.science 110, 213–225.(1999). Ca2�-induced apoptosis through calcineurin dephosphoryla-Nowak, L., Bregestovski, P., Ascher, P., Herbet, A., and Prochiantz,tion of BAD. Science 284, 339–343.A. (1984). Magnesium gates glutamate-activated channels in mouseWang, D., Tolbert, L.M., Carlson, K.W., and Sadee, W. (2000). Nu-central neurones. Nature 307, 462–465.clear Ca2�/calmodulin translocation activated by the �-opiod (OP3)Ogawa, S., and Pfaff, D.W. (1998). Current status of antisense DNAreceptor. J. Neurochem. 74, 1418–1425.methods in behavioral studies. Chem. Senses 23, 249–255.Williams, S.C., Angerer, N.D., and Johnson, P.F. (1997). C/EBP pro-Olive, M., Williams, S.C., Dezan, C., Johnson, P.F., and Vinson, C.teins contain nuclear localization signals imbedded in their basic(1996). Design of a C/EBP-specific, dominant-negative bZIP proteinregions. Gene Expr. 6, 371–385.with both inhibitory and gain-of-function properties. J. Biol. Chem.Xiong, W., Hsieh, C.-C., Kurtz, A.J., Rabek, J.P., and Papaconstanti-271, 2040–2047.nou, J. (2001). Regulation of CCAAT/enhancer-binding protein-�Peterson, B.Z., DeMaria, C.D., Adelman, J.P., and Yue, D.T. (1999).isoform synthesis by alternative translational initiation at multipleCalmodulin is the Ca2� sensor for Ca2�-dependent inactivation ofAUG sites. Nucleic Acids Res. 29, 3087–3098.L-type calcium channels. Neuron 22, 549–558.Yin, M., Yang, S.Q., Lin, H.Z., Lane, M.D., Chatterjee, S., and Diehl,Piwien-Pilipuk, G., Van Mater, D., Ross, S.E., MacDougald, O.A.,A.M. (1996). Tumor necrosis factor- promotes nuclear localizationand Schwartz, J. (2001). Growth hormone regulates phosphorylationof cytokine-inducible CCAAT/enhancer binding protein isoforms inand function of CCAAT/enhancer-binding protein � by modulatinghepatocytes. J. Biol. Chem. 271, 17974–17978.Akt and glycogen synthase kinase-3. J. Biol. Chem. 276, 19664–Yoshida, H., Nozu, F., Lankisch, T.O., Mitamura, K., Owyang, C., and19671.Tsunoda, Y. (2000). A possible role for Ca2�/calmodulin-dependentPohnke, Y., Kempf, R., and Gellersen, B. (1999). CCAAT/enhancer-protein kinase IV during pancreatic acinar stimulus-secretion cou-binding proteins are mediators in the protein kinase A-dependentpling. Biochim. Biophys. Acta 1497, 155–167.activation of the decidual prolactin promoter. J. Biol. Chem. 274,Youn, H.D., Grozinger, C.M., and Liu, J.O. (2000). Calcium regulates24808–24818.transcriptional repression of myocyte enhancer factor 2 by histoneRajadhyaksha, A., Barczak, A., Macias, W., Leveque, J.C., Lewis,deacetylase 4. J. Biol. Chem. 275, 22563–22567.S.E., and Konradi, C. (1999). L-type Ca2� channels are essential forYukawa, K., Tanaka, T., Tsuji, S., and Akira, S. (1998). Expressionsglutamate-mediated CREB phosphorylation and c-fos gene expres-of CCAAT/Enhancer-binding proteins � and � and their activitiession in striatal neurons. J. Neurosci. 19, 6348–6359.are intensified by cAMP signaling as well as Ca2�/calmodulin kinasesRamji, D.P., and Foka, P. (2002). CCAAT/enhancer-binding proteins:activation in hippocampal neurons. J. Biol. Chem. 273, 31345–structure, function and regulation. Biochem. J. 365, 561–575.31351.

Riccio, A., Ahn, S., Davenport, C.M., Blendy, J.A., and Ginty, D.D.Zhu, S., Yoon, K., Sterneck, E., Johnson, P.F., and Smart, R.C. (2002).(1999). Mediation by a CREB family transcription factor of NGF-CCAAT/enhancer binding protein-� is a mediator of keratinocytedependent survival of sympathetic neurons. Science 286, 2358–survival and skin tumorigenesis involving oncogenic Ras signaling.2361.Proc. Natl. Acad. Sci. USA 99, 207–212.

Ruiz, F., Alvarez, G., Ramos, M., Hernandez, M., Bogonez, E., andSatrustegui, J. (2000). Cyclosporin A targets involved in protection

Note Added in Proofagainst glutamate excitotoxicity. Eur. J. Pharmacol. 404, 29–39.

Sala, C., Rudolph-Correia, S., and Sheng, M. (2000). Developmen- A recent paper by Thiagarajan et al. (Neuron 36, 1103–1114) empha-tally regulated NMDA receptor-dependent dephosphorylation of sizes the reciprocal importance of AMPA receptor and L channelcAMP response element-binding protein (CREB) in hippocampal activity.neurons. J. Neurosci. 20, 3529–3536.

See, V., Boutillier, A.-L., Bito, H., and Loeffler, J.-P. (2001). Calcium/calmodulin-dependent protein kinase type IV (CaMKIV) inhibits apo-ptosis induced by potassium deprivation in cerebellar granule neu-rons. FASEB J. 15, 134–144.

Sinor, J.D., Du, S., Venneti, S., Blitzblau, R.C., Leszkiewicz, D.N.,Rosenberg, P.A., and Aizenman, E. (2000). NMDA and glutamateevoke excitotoxicity at distinct cellular locations in rat cortical neu-rons in vitro. J. Neurosci. 20, 8831–8837.

Soderling, T.R. (2000). CaM-kinases: modulators of synaptic plastic-ity. Curr. Opin. Neurobiol. 10, 375–380.

Sterneck, E., and Johnson, P.F. (1998). CCAAT/enhancer bindingprotein � is a neuronal transcriptional regulator activated by nervegrowth factor signaling. J. Neurochem. 70, 2424–2433.

Sun, P., Enslen, H., Myung, P.S., and Maurer, R.A. (1994). Differentialactivation of CREB by Ca2�/calmodulin-dependent protein kinasestype II and type IV involves phosphorylation of a site that negativelyregulates activity. Genes Dev. 8, 2527–2539.

Tao, X., Finkbeiner, S., Arnold, D.B., Shaywitz, A.J., and Greenberg,M.E. (1998). Ca2� influx regulates BDNF transcription by a CREBfamily transcription factor-dependent mechanism. Neuron 20,709–726.

Taubenfeld, S.M., Milekic, M.H., Monti, B., and Alberini, C.M. (2001a).The consolidation of new but not reactivated memory requires hip-pocampal C/EBP�. Nat. Neurosci. 4, 813–818.

Taubenfeld, S.M., Wiig, K.A., Monti, B., Dolan, B., Pollonini, G., andAlberini, C.M. (2001b). Fornix-dependent induction of hippocampalCCAAT enhancer-binding protein � and � co-localizes with phos-phorylated cAMP response element-binding protein and accompan-ies long-term memory consolidation. J. Neurosci. 21, 84–91.