Spontaneous Calcium Oscillations Control c-fosTranscription via the Serum Response Element in Neuroendocrine Cells*

In excitable cells the localization of Ca2+ signals plays a central role in the cellular response, especially in the control of gene transcription. To study the effect of localized Ca2+signals on the transcriptional activation of the c-fosoncogene, we stably expressed various c-fos β-lactamase reporter constructs in pituitary AtT20 cells. A significant, but heterogenous expression of c-fos β-lactamase was observed in unstimulated cells, and a further increase was observed using KCl depolarization, epidermal growth factor (EGF), pituitary adenylate cyclase-activating polypeptide (PACAP), and serum. The KCl response was almost abolished by a nuclear Ca2+ clamp, indicating that a rise in nuclear Ca2+ is required. In contrast, the basal expression was not affected by the nuclear Ca2+ clamp, but it was strongly reduced by nifedipine, a specific antagonist of l-type Ca2+ channels. Spontaneous Ca2+ oscillations, blocked by nifedipine, were observed in the cytosol but did not propagate to the nucleus, suggesting that a rise in cytosolic Ca2+ is sufficient for basal c-fos expression. Inactivation of the c-fos promoter cAMP/Ca2+ response element (CRE) had no effect on basal or stimulated expression, whereas inactivation of the serum response element (SRE) had the same marked inhibitory effect as nifedipine. These experiments suggest that in AtT20 cells spontaneous Ca2+ oscillations maintain a basal c-fos transcription through the serum response element. Further induction of c-fos expression by depolarization requires a nuclear Ca2+ increase.

How short-lived intracellular signals cause marked alterations in gene expression leading to long-term cellular responses is still a puzzling aspect of cell biology. Immediate early genes (IEGs), 1 such as c-fos, function as a relay in this process; the transcription and protein levels of IEGs can be changed rapidly as a direct function of cellular signals. IEGs, which are mostly transcription factors, in turn control the expression of "late" responsive genes, which then modify the cellular functions (1).
The control by intracellular signals of the transcriptional activation of the c-fos gene has been widely studied (2,3,4). The c-fos gene includes in its promoter two major response elements that are targets of phosphorylation cascades, the cAMP/Ca 2ϩ response element (CRE) and the serum response element (SRE) (2). These two consensus sequences bind different transcription factors, the activities of which are modulated by Ca 2ϩ -dependent (11,12) as well as Ca 2ϩ -independent phosphorylation (13).
The CRE, which is located ϳ60 nucleotides upstream of the transcription initiation site of the c-fos gene, is bound by a leucine zipper transcription factor, CREB (CRE binding protein) (14). Following stimulation, CREB is phosphorylated on a critical residue, the serine 133. This phosphorylation allows CREB to recruit the transcriptional adapter CBP (CREB binding protein) and activates transcription of c-fos. Activation of various signaling pathways results in CREB phosphorylation, most notably the cAMP/PKA cascade, Ca 2ϩ signaling acting via Ca 2ϩ -calmodulin kinases (CaMkinases), as well as the mitogen-activated protein (MAP) kinases.
A second Ca 2ϩ responsive element in the c-fos promoter is the serum response element, SRE, which is located ϳ310 nucleotides upstream of the transcriptional initiation site of the c-fos gene (15). SRE binds the serum response factor (SRF) and its accessory factor TCF, the ternary complex factor. TCFs are encoded by a family of Ets proteins that includes Elk-1, SAP-1a, and SAP-2. Phosphorylation of SRF on serine residue 103 is crucial for it's transcriptional activity. This residue can be phosphorylated in vitro by MAP kinases and CaMkinases. Therefore, as for CREB, the phosphorylation of SRF may implicate various signaling pathways including Ca 2ϩ signaling cascades mediated by CaMkinase (16,17). Physiological activation of transcription via the SRE is thought to occur predominantly following growth factor stimulation of MAP kinase cascades.
Control of gene expression by changes in intracellular calcium (Ca 2ϩ ) concentration nearly always involves changes in protein phosphorylation. Ca 2ϩ signals can cause such changes directly via Ca 2ϩ -calmodulin-activated protein kinases, CaMkinases, or calcineurin, a Ca 2ϩ -dependent protein phosphatase (6,7).
A study suggests that finely tuned Ca 2ϩ signals triggered in restricted cellular domains may be able to activate processes that are selectively affecting c-fos gene transcription (5). An increase of intracellular Ca 2ϩ concentration can occur inside or outside of the nucleus. A cytosolic Ca 2ϩ increase can activate CaMkinase II and/or calcineurin, which then translocate from their site of activation to the nucleus (8,7). Transcriptional activation by nuclear Ca 2ϩ can occur by a direct binding on DREAM (downstream regulatory element antagonist modulator), a Ca 2ϩ binding transcriptional repressor (9), or by the translocation of the Ca 2ϩ /calmodulin complex from the cytosol to the nucleus. This translocated complex will then activate a nuclear CaMkinase, such as CaMkinase IV (10).
Based on the stimulus-secretion coupling concept proposed by W. W. Douglas, pioneering studies in the 1960s demonstrated the electrical excitability of endocrine cells of the pituitary, the pancreas, and the adrenal medulla, now termed neuroendocrine cells (18,19,20). These early studies showed that voltage-gated Ca 2ϩ channels were instrumental in the generation of action potentials in a neuroendocrine cell; depolarizing currents were indeed sodium (Na ϩ ) and Ca 2ϩ influxes, whereas repolarization was achieved by potassium (K ϩ ). Later we demonstrated in neuroendocrine cells that a single action potential may cause a well defined intracellular Ca 2ϩ signal, that those action potentials occur spontaneously and that they are controlled by somatostatin (21). A wealth of subsequent literature illustrates that the modulation of Ca 2ϩ action potentials in neuroendocrine cells is a versatile signaling option utilized by most releasing factors and releasing modulators, but also by growth factors and other extracellular signals (34). Activation or inhibition of neuroendocrine cell activity nearly always involves such modulation, which results in alterations of action potential frequency and/or rhythm. In addition, amplitude, duration, and intracellular propagation of action potential-linked Ca 2ϩ signals are modulated as well. In summary, Ca 2ϩ action potentials and the resulting intracellular Ca 2ϩ transients are fundamental signaling units in neuroendocrine cells.
Transcriptional control by Ca 2ϩ has been investigated within the context of Ca 2ϩ oscillations generated by intracellular mechanisms originally described in hepatocytes (22). Such Ca 2ϩ oscillations are produced following receptor-mediated phospholipase C (PLC) activation and the generation of the second messenger inositol 1,4,5-trisphosphate (IP 3 ). IP 3 binds to the IP 3 receptor at the endoplasmic reticulum (ER) and induces the release of Ca 2ϩ from the ER. This Ca 2ϩ release from intracellular stores occur in many cell types including pituitary cells (23). Ca 2ϩ oscillations may activate gene transcription in a manner dependent on their pattern, as demonstrated for the transcription factor nuclear factor AT (NF-AT), which is activated by a mechanism involving the Ca 2ϩ -dependent phosphatase, calcineurin (24,25).
Action potential-linked Ca 2ϩ transients most likely function also for the activation of gene transcription by mechanisms sensitive to frequency and rhythm of oscillatory Ca 2ϩ changes. To test this hypothesis, we used a single cell reporter gene approach (26) in order to link the observations on Ca 2ϩ transients in individual cells to gene transcription. Furthermore, at the single cell level Ca 2ϩ signals can be selectively manipulated by microinjection of high molecular weight Ca 2ϩ chelators into specific compartments such as the nucleus.
Here we show that Ca 2ϩ transients driven by spontaneous action potentials in the pituitary cell line AtT20 sustain a basal transcriptional activity of the IEG c-fos through a mechanism that involves the SRE and does not require changes in nuclear Ca 2ϩ concentration.
Plasmid Constructions-We used the bacterial enzyme ␤-lactamase as a reporter gene. We already dispose of luciferase reporter constructs containing the c-fos promoter (27,28). The luciferase coding sequence from our reporter constructs was then excised by NocI and XbaI digestion and replaced by the ␤-lactamase coding sequence.
For our experiments four different constructs were used. The first contains c-fos proto-oncogene for the region of the mouse, extending from Ϫ379 to ϩ1073. This construct is called c-fos-␤L and corresponds to the wild type c-fos promoter plus part of the gene (first exon and first intron). The second and third constructions contained one each a mutation inactivating the CRE (c-fos-⌬CRE-␤L) or the SRE (c-fos-⌬SRE-␤L) response elements. The last construction contained two mutations inactivating both the SRE and CRE response elements (c-fos-⌬SRE/ CRE-␤L). All these constructions are schematically presented in Fig. 5. Mutations were performed by site-directed mutagenesis using the Quick-Change mutagenesis kit from Stratagene (Basel, Switzerland). The strategy of mutagenesis has already been described (5,27,28).
Cell Culture-Rat corticotrope AtT20 cells were cultured in a Dulbecco's modified Eagle's medium-F12 supplemented with 10% fetal calf serum (FCS) at 37°C in a humidified atmosphere of 95% air and 5% CO 2 . Stably transfected AtT20 clones (see below) were selectively maintained with 100 g/ml G418 (an antibiotic analog to neomycin) added to the culture medium. For all measurements cells were plated on 25 mm coverslips, and serum was removed from the culture medium 24 h before experiments.
Establishment of Stable ␤-lactamase Expressing AtT20 Pituitary Cell Clones-Five stable AtT20 cell lines were established, four cell lines containing a reporter construct plasmid (c-fos-␤L, c-fos-⌬CRE-␤L, c-fos-⌬SRE-␤L, c-fos-⌬SRE/CRE-␤L) and one cell line with viral promoter SV40 (SV40-␤L). A pcDNA3 vector containing the neomycin resistance gene co-transfected in a mass ratio of 1:10 was used for selection. The DOSPER liposomal transfection reagent (Roche) was used following the manufacturer's recommendations. The cotransfection was performed as follows. AtT20 cells were grown to ϳ60% confluence in 35 mm diameter Petri dishes. After removal of the culture medium and one wash with HBS (Hepes buffer saline, pH 7.4), cells were incubated with 2 ml of HBS containing the plasmids (0.5 g of pcDNA3 and 5 g of reporter plasmid) dissolved previously in 40 l of DOSPER. After 6 h, cells were washed with HBS, and culture was continued in Dulbecco's modified Eagle's medium-F12 (10% fetal calf serum) containing 400 g/ml G418.
After 3 weeks of culture in G418 medium, several neomycin-resistant clones were selected and tested for their ability to induce fluorescence changes in CCF-2 in response to stimulation by 20 mM KCl and 3 M cpt-cAMP (chlorophenylthio-cyclic AMP).
Microinjection-AtT20 cells were injected using an Eppendorf transjector 5246 mounted on a Zeiss Axiovert S100TV microscope. For the measurements of nuclear Ca 2ϩ , the injection solution was 50 M Fura 2-D70, half-strength phosphate-buffered saline, 1 mM MgCl 2 , pH 7.2. For the nuclear-Ca 2ϩ clamp, 2.5 mM Bapta-D70 and 1 mM Ca 2ϩ were added (free Ca 2ϩ concentration was calculated to be 83 nM). To measure the ␤-lactamase activity, Fura 2-D70 was replaced by 5% TR-D70 as injection marker. After injection, cells were kept in the incubator at 37°C for 4 to 5 h before experiments.
Calcium Measurements-Cytosolic calcium concentration variations were measured using the calcium probe Fura-2. Cells were loaded for 30 min at room temperature with the membrane permeant Fura-2 AM in a medium containing NaCl 140 mM, KCl 5 mM, Ca 2ϩ 1.2 mM, MgCl 2 1 mM, glucose 10 mM, Hepes 20 mM, pH 7.4.
Fura-2 fluorescence (excitation 340/380 nm, emission 510 nm) was monitored with an imaging system. Loaded cells plated on coverslips were mounted on an inverted Zeiss Axiovert S100TV microscope coupled to a Princeton Instruments cooled back-illuminated frame-transfer charge-coupled device camera. The light source for illumination came from a Xenon XBO 75-W lamp. Excitation wavelengths were selected with a PTI (Photon Technology International) monochromator. Images were acquired with the Metamorph 4.1 software (Universal Imaging Corp.). Measurements were performed at 37°C in an open perfusion micro-incubator (Harward Apparatus). The cytosolic and nuclear calcium (Ca 2ϩ ) concentrations were determined as described (29).
Monitoring ␤-Lactamase Reporter Activity-AtT20 cells cultured on 25 mm coverslips were loaded with the fluorescent substrate CCF-2AM as described (26). Briefly, cells were incubated for 1 h at room temper-ature in culture medium with 1 M CCF-2AM and 1% pluronic acid. Cells were then washed in the perfusion chamber of the imaging system (as described before for Ca 2ϩ measurements). Measurements were made at 37°C. CCF-2 fluorescence (excitation 405 nm) was imaged alternatively at emission wavelengths 450/530 nm, and data were stored at 3-min intervals. From these raw data in the form of image series at the two wavelengths, the fluorescence emission ratios F(450 nm)/F(530 nm) integrated over the cell area ratio were calculated (

Activation of c-fos Transcription in AtT20
Cells-To visualize the activation of gene transcription at the single cell level, we used ␤-lactamase as a reporter gene (26). Expression of this bacterial enzyme can easily be detected by monitoring the fluorescence of a cell-permeant substrate, CCF-2, composed of two fluorophores linked by a ␤-lactame ring. The uncleaved CCF-2 emits green light (530 nm) because of fluorescence resonance energy transfer (FRET) between the two fluorophores. Upon ␤-lactamase expression, the substrate is cleaved and emits blue light (450 nm) because of the loss of FRET. ␤-lactamase activity is then measured as the change in green to blue CCF-2 fluorescence. To assess c-fos transcription, the ␤-lactamase coding sequence was placed under control of the promoter and first intron of the c-fos gene, yielding the reporter construct c-fos-␤L that is drawn schematically in Fig. 5. Our earlier studies had shown that the first intron of the c-fos gene contains important regulatory elements without which transcription control by Ca 2ϩ is incomplete (27,28).
We generated AtT20 cell clones stably expressing c-fos-␤L, which were loaded with CCF-2 AM. As seen in Fig. 1 (left  panels), stimulation of the CCF-2-loaded c-fos-␤L cells with agonists induced a change in the color of CCF-2 fluorescence from green to blue within 15-20 min. The degree of ␤-lactamase expression was quantified as the ratio of blue (450 nM) to green (530 nm) fluorescence intensities, an increase in CCF-2 ratio (F450/F530) thus reporting c-fos-␤L activation. Following cell stimulation, CCF-2 fluorescence images were acquired every 3 min, and CCF-2 ratios were calculated to obtain a time course of c-fos-␤L transcriptional activation (Fig. 1, right panels). In the absence of any stimulation, no change in CCF-2 ratio could be detected (top right panel), indicating that the degree of basal c-fos-␤L transcriptional activation was stable in AtT20 cells over the 2-h measurement period. In contrast, c-fos-␤L reporter expression was markedly enhanced by stimuli known to activate transcription of the c-fos gene in AtT20 cells (Fig. 1, middle and lower panels), such as EGF (20 nM To study the effect of Ca 2ϩ on c-fos transcription, cells were depolarized with 20 mM KCl in order to open voltage-gated Ca 2ϩ channels (VOCs) and trigger Ca 2ϩ influx. As shown in Fig. 2, KCl depolarization also powerfully activated c-fos-␤L transcription, causing a near doubling in the CCF-2 ratio after 2 h of stimulation. This increase was almost completely abolished by preloading the cells with the Ca 2ϩ -chelator BAPTA-AM (Fig. 2, bottom panel), confirming that the KCl response was mediated by an increase in cellular Ca 2ϩ . Thus, in AtT20 cells a sustained increase in intracellular Ca 2ϩ strongly activates the transcription of the c-fos gene.
Spontaneous Ca 2ϩ Transients Regulate Basal c-fos Transcription-Although the basal CCF-2 ratio remained stable for up to 2 h in the absence of exogenous stimulus (Fig. 1), there was considerable cell-to-cell heterogeneity among non-stimulated cells (cf. the time 0Ј images in Figs. 1 and 2). This suggested that c-fos-␤L might already be activated in a fraction of the resting AtT20 cells. Accordingly, the histogram distribution of the CCF-2 ratios could be separated into three subpopulations centered at ratio values of 0.7, 1.8, and 3.5, and comprising 29%, 59%, and 12% of resting c-fos-␤L AtT20 cells, respectively (Fig. 3A). This suggested that in the absence of stimulation transcription was strongly repressed in only approximately one-third of the cell population, whereas c-fos-␤L was transcribed at intermediate levels in 59% of cells and at high levels in 12% of cells. Interestingly, the ratio value of the "active" population (3.5) was similar to the average CCF-2 ratio measured upon KCl stimulation (Fig. 3C), suggesting that c-fos transcription was fully activated in 12% of unstimulated cells. To assess whether this heterogeneity could reflect spontaneous Ca 2ϩ transients caused by the basal electrical activity of AtT20 cells, we used the L-type Ca 2ϩ channel inhibitor, nifedipine. Exposure of cells to 1 M nifedipine for 4 h, a concentration that completely blocks L-type Ca 2ϩ channels, shifted the CCF-2 ratio to lower values, leaving nearly no transcriptionally fully active cells (Fig. 3B). Most cells (61%) now displayed a CCF-2 ratio of 0.8, and the proportion of cells with intermediate c-fos-␤L activity (1.6) was reduced to 34%. An opposite effect was observed with a 2-h KCl stimulation (Fig. 3C); the proportion of cells with fully activated c-fos-␤L increased from 12 to 69% (ratio 3.64), whereas intermediate c-fos-␤L activity was observed in the remaining 31% of cells (ratio 2.26). Thus, in AtT20 cells, basal c-fos transcription is relatively high as a result of electrical activity. Ca 2ϩ channel blockers strongly down-regulate this endogenous activity, whereas a long-lasting depolarization powerfully activates c-fos transcription.
Basal c-fos Transcription Is Mediated by the SRE Element-The convergent effect of nifedipine on basal Ca 2ϩ transients and c-fos-␤L activity suggested an important role for the spontaneous Ca 2ϩ transients in the regulation of gene expression. To assess which element within the c-fos promoter mediated this Ca 2ϩ response, we separately mutated two Ca 2ϩ response elements within the c-fos-␤L promoter (SRE and CRE, Fig. 5A). The CRE and SRE elements were inactivated by site-directed mutagenesis (see Experimental Procedures), and AtT20 clones stably expressing the mutated constructs were established. These two clones (c-fos-⌬CRE-␤L and c-fos-⌬SRE-␤L) also displayed spontaneous Ca 2ϩ transients, with amplitude and frequency similar to those observed in c-fos-␤L or wild type AtT20 cells (data not shown). As shown in Fig. 5B (hatched bars), deletion of the CRE element only marginally increased the basal CCF-2 ratio and did not affect the ability of cells to respond to nifedipine, but precluded their activation by KCl (2.69 Ϯ 0.37 versus 2.15 Ϯ 0.24, p ϭ 0.07, unpaired t test). In contrast, deletion of the SRE element had a profound effect on endogenous c-fos-␤L activity, decreasing the basal CCF-2 ratio from 1.80 Ϯ 0.06 to 0.83 Ϯ 0.03 (filled bars, p Ͻ 10 Ϫ8 , unpaired t test). Basal c-fos-⌬SRE-␤L transcription could not be further reduced by nifedipine (0.80 Ϯ 0.04), but the cells were still able to respond to a depolarizing stimulus. The ratio increased from , to values only slightly higher than the minimal values measured in non-transfected cells (dotted line). These cells were insensitive to nifedipine and to KCl stimulation. These data suggest that the CRE element is not required for the basal activation of c-fos. Instead, the SRE element appears to mediate the basal c-fos transcription driven by spontaneous action potentials. In contrast, the SRE element is less essential for the Ca 2ϩ -mediated transcriptional activation induced by a large and long-lasting depolarization.
Ca 2ϩ Transients Caused by Spontaneous Action Potentials Propagate Poorly to the Nucleus-The differential regulation of the spontaneous and KCl-induced c-fos transcription suggested that the two types of Ca 2ϩ signals might propagate differently to the nucleus. In particular, because of their relatively low amplitude (118 nM, Fig. 4), we suspected that the spontaneous Ca 2ϩ transients might fail to reach the nuclear compartment because of space restriction and buffering effects (32,33). To test this hypothesis, the Ca 2ϩ concentration was measured in the nuclear compartment by nuclear microinjection of Fura-2 Dextran (70 kDa). The large dextran molecule allowed a stable confinement of the Ca 2ϩ probe within the nuclear compartment for more than 4 h (Fig. 6A). In the absence of stimulation, spontaneous nuclear Ca 2ϩ transients were observed only in 5 of 19 microinjected cells (26%), a proportion much lower than the corresponding cytosolic Ca 2ϩ transients (87%, Fig. 4). Furthermore, although the two compartments had similar basal Ca 2ϩ levels (75 Ϯ 22 versus 70 Ϯ 20 nM), the amplitude of the measured Ca 2ϩ transients was lower in the nucleus than in the cytosol (80 Ϯ 26 versus 118 Ϯ 87 nM). The failure to detect higher spontaneous nuclear activity was not caused by a lack of sensitivity of the microinjected Fura-2-dextran probe, because subsequent KCl stimulation caused a large increase in nuclear Ca 2ϩ , regardless of the basal activity of the cells (Fig. 6A). Thus, spontaneous Ca 2ϩ transients propagate poorly to the nuclear compartment, whereas the KCl-induced Ca 2ϩ transients produce large Ca 2ϩ signals within the nucleus.
To assess whether Ca 2ϩ propagation to the nucleus was required for the KCl response, we then attempted to prevent the nuclear Ca 2ϩ signals by co-injecting a dextran-coupled Ca 2ϩ chelator (BAPTA-D70) together with the nuclear Fura-2 probe. As shown in Fig. 6B, no Ca 2ϩ transients were observed under these conditions (n ϭ 9), although the nuclear Ca 2ϩ clamp did not affect the cytosolic Ca 2ϩ transients (data not shown). More importantly, although the cytosolic KCl response was preserved (data not shown), the nuclear Ca 2ϩ signal was strongly reduced (Fig. 6B, mean amplitude ϭ 72 Ϯ 23 nM), indicating that the nuclear Ca 2ϩ clamp achieved with BAPTA-D70 injection was highly effective.
A Rise in Nuclear Ca 2ϩ Is Required for KCl-Induced, but Not for Basal, c-fos Expression-To assess the effects of a nuclear Ca 2ϩ clamp on basal-and KCl-induced transcription, c-fos-␤L activity was measured in cells injected with BAPTA-D70 or with the injection marker Texas-Red-D70 (TR-D70) alone. As shown in Fig. 7B, the nuclear Ca 2ϩ clamp did not affect basal c-fos-␤L activity, the CCF-ratio averaging 1.79 Ϯ 0.12 and 2.05 Ϯ 0.17 in BAPTA-D70-and control-injected cells, respectively. Furthermore, basal c-fos-␤L activity could still be inhibited by nifedipine in BAPTA-D70-injected cells, the CCF-2 ratio decreasing to 1.37 Ϯ 0.1 versus 1.41 Ϯ 0.1 in control-injected cells. In sharp contrast however, the nuclear Ca 2ϩ clamp completely abolished the KCl-induced c-fos activation (Fig. 7B). clamp had no effect on the basal c-fos-␤L activity, indicating that endogenous c-fos activity, driven by the spontaneous activity of AtT-20 cells, does not require a rise in nuclear Ca 2ϩ . In contrast, large and sustained increases in Ca 2ϩ , such as those elicited by the KCl-induced depolarization, cause enhanced c-fos transcription in a manner stringently dependent on a nuclear Ca 2ϩ increase.
A Nuclear Ca 2ϩ Rise Is Required for CRE, but Not for SRE, Mediated c-fos Transcription-To determine whether a rise in nuclear Ca 2ϩ activates the SRE or the CRE element, we performed nuclear Ca 2ϩ clamp experiments with the mutated constructs. As shown in Fig. 8A, nuclear microinjection of BAPTA-D70 had no effect on the construct bearing an inactivated CRE element (c-fos-⌬CRE-␤L). This construct displayed a strong basal expression sensitive to nifedipine and failed to respond to KCl, regardless of the nuclear clamp. In contrast, nuclear Ca 2ϩ chelation nearly abolished the KCl-induced activation of the construct bearing an inactivated SRE element (c-fos-⌬SRE-␤L). This construct had a reduced basal expression insensitive to nifedipine and a strong KCl response that was completely prevented by nuclear microinjection of BAPTA-D70 (Fig. 8B). Thus, a rise in nuclear calcium mediates the KCl response, which does not require the SRE element but does require the CRE element. Conversely, a rise in nuclear Ca 2ϩ is not necessary for basal c-fos expression, which does not require the CRE element but does require the SRE element. DISCUSSION Important elements of Ca 2ϩ signaling in neuroendocrine cells can be studied only in individual cells. To link such elements to the control of transcription, tools are required to study the latter at the level of the single cell. In the past repeated attempts have been made to study gene transcription at the single cell level, and several very different techniques have been engaged.
Some techniques aim to directly read out the mRNA of interest. Those are based on "in vivo " hybridization of fluorescent DNA or RNA probes. Hybridization generates a fluorescence signal by eliminating self-quenching in probes with a "molecular beacon" design (44). Alternatively, hybridization of two probes in close vicinity can be monitored by FRET from the FRET photon donor probe to the FRET photon acceptor probe (45). Direct detection of mRNA in live single cells using those techniques has not been reported by many groups, most likely because of technical difficulties. For example, probes need to be injected, and their design is difficult because they have to hybridize selectively at low temperatures; furthermore, only a substantial increase in mRNA will generate a detectable signal.
The few studies addressing gene transcription control mechanisms in single cells are based mostly on the quantification of a "reporter" protein. Three types of reporter genes have been used, namely luciferase and ␤-lactamase that code for enzymes whose activity can be monitored (25,26,42,43) or GFP whose product can be directly detected by its fluorescence. Finally, in some studies exogenous gene products carrying an epitope tag have been quantified by immunocytochemistry (5).
Among reporter proteins, ␤-lactamase has some clear advantages, namely the high sensitivity of its detection with the non-toxic fluorescent ␤-lactamase substrate CCF-2 and the possibility of quantification based on a ratio-metric measurement that eliminates nonspecific effects on total fluorescence. By comparison with microinjected luciferase reporter vectors using the same regulatory elements 2 or very similar elements (42), we estimate that ␤-lactamase provides for a 10 -50-fold increase in sensitivity over luciferase, which turns out to be a reporter too weak to be used to study induction of IEGs (46).
In this single-cell study of transcription control it was indeed possible to exploit individual cell activities and relate them to gene expression (Figs. 3 and 4). Also, signal compartmentalization was addressed using microinjection, a powerful single cell technique. Note that microinjection of the reporter plasmids should permit reporter studies with preparations close to the physiological reality (e.g. fully differentiated neurons in brain slices). The fluorescence readout of reporter ␤-lactamase activity provides only a semi-quantitative result, the relevance of which relies on statistics. This limit, however, is inherent to all reporter gene studies; furthermore, the majority of currently published work on gene expression uses largely semi-quantitative data such as densitometric evaluation of Northern blots.
With the ␤-lactamase reporter gene, c-fos expression was studied in individual cells of the pituitary line AtT20. The releasing factor PACAP as well as serum caused a strong induction of c-fos reporter expression. Spontaneous Ca 2ϩ transients caused by Ca 2ϩ action potentials (21)  reporter transcription (Fig. 4). Mutation of the c-fos-promoter, which inactivates the SRE element, strongly reduced basal transcriptional activity; in contrast, an inactivating mutation on the CRE element had no effect (Fig. 5). Nuclear injection of a slowly diffusing Ca 2ϩ chelator, BABTA-D70, strongly attenuated spontaneous as well as depolarization-induced nuclear Ca 2ϩ transients measured with coinjected fura-D70. Nuclear BAPTA-D70 had no significant effect on basal c-fos reporter transcription, whereas it strongly reduced KCl-stimulated c-fos-␤L because of sustained depolarization, consistent with the notion that cytoplasmic rather than nuclear Ca 2ϩ signals are responsible for basal c-fos transcription (Figs. 6 and 7).
IEG transcription factors such as c-fos link cellular activity to gene expression. Transcription of IEGs is triggered by short lived intracellular signals that accompany cell activity; IEG transcription factors in turn control a multitude of cell typespecific genes, the transcription of which is adapted to cell activation. We show here a link between Ca 2ϩ action potentials and basal c-fos transcription. Physiologically, such a link can provide for a precise transcriptional readout of a finely tuned phasic Ca 2ϩ signal. Indeed, frequency, rhythm, and amplitude of Ca 2ϩ action potentials are extremely well controlled by a multitude of factors that regulate hormone secretion by these same cells (34). It thus appears that secretory activity and immediate gene expression are sensitive to the same Ca 2ϩ signals, which is a means for coordinate control.
What is the potential role of the "basal" expression of IEGs such as c-fos in non-stimulated neuroendocrine cells? In the pituitary, inhibitors such as somatostatin and dopamine play an important part in the regulation of hormone secretion. Their action is based largely on the suppression of spontaneous action potentials. Indeed, spontaneous action potentials are likely a rare physiological event for cells in an intact pituitary in which tonic inhibition by dopamine and adenosine predominate most of the time. In the AtT20 cell model, spontaneous Ca 2ϩ transients that cause the basal c-fos expression just illustrate how IEG levels may reflect precisely the "activity status" of each individual cell.
The c-fos promoter element SRE appears essential to sustain a basal transcription of the c-fos gene because of action potential-linked Ca 2ϩ transients in AtT20 cells. The transcription factors SRF and it's cofactor TCF (i.e. ELK-1) bind to the SRE element of the c-fos promoter. In response to serum or purified growth factors, the transcription activation by SRF is principally controlled by the ras-raf-ERK MAPkinase pathway or by rhoA GTPase-dependent mechanisms (35,36).
The activation of SRF by Ca 2ϩ signals was demonstrated with a c-fos reporter gene including a minimal promoter and a single SRE inserted 42 base pairs upstream of the transcription start site (16). Moreover, in a promoter lacking the SRE element, even in presence of the CRE the Ca 2ϩ -activated transcription was significantly reduced. Activation of SRF by Ca 2ϩ depends on CaMKinases, and not on the ras-raf-ERK pathway known to mediate Ca 2ϩ activation of TCF in PC12 cells (37). In AtT20 cells the Ca 2ϩ -dependent c-fos gene transcription activation via SRE appears to be TCF-independent (17). The expression of constitutively active CaMKinases II and IV in AtT20 cells (38) showed that CaMKinase IV repressed transcriptional activity of a c-fos-SRE-CAT reporter activity, whereas CaMKinase II activated a c-fos-SRE-CAT reporter transcription.
SRE-dependent transcription of the c-fos gene in AtT20 cells appears possible by Ca 2ϩ signals of moderate amplitude that are limited to the cytosol, as is evident from our data as well as from an earlier study (5). It is therefore likely that Ca 2ϩ transients are linked to c-fos transcription via CaMKinase II. This enzyme is located in the cytosol (8), and it's activity is modulated by the pattern of Ca 2ϩ oscillations (39). A further potential mediator is the Ca 2ϩ -activated tyrosine kinase PYK-2, a member of the non-receptor tyrosine kinase family Fak, which is highly expressed in brain (40). Ca 2ϩ -activated PYK-2 leads to the activation of the Ras-MAP kinase signaling pathways.
Ca 2ϩ signals outside the nucleus can lead to enhanced transcription also via CRE (41). This has been elegantly demonstrated by a recent study in neurons (47) where calmodulin associated to L-type voltage-gated Ca 2ϩ channels and released upon Ca 2ϩ flux through them is capable of causing the phosphorylation of CREB and thereby the activation of CRE-driven reporter gene transcription. Interestingly, this selective activation mechanism based on a very localized rise in Ca 2ϩ involves the ERK-MAP kinases cascade.
For the spontaneous Ca 2ϩ transients and "basal" c-fos transcription studied here, CRE does not seem to be an important target, as it can be inactivated by mutation without significant consequences. However, this finding does not exclude a certain contribution of an ERK-CREB-CRE pathway to the physiological regulation of steady-state c-fos mRNA.
In conclusion, this study shows how Ca 2ϩ action potentials, which are a reflection of neuroendocrine cell activity even in a non-stimulated "basal" state, control transcription of the immediate early gene c-fos via cytoplasmic Ca 2ϩ transients signaling to the SRE element in the c-fos promoter. In our view this mechanism is involved in the adaptation of gene expression to cell activity and may well apply to many more IEG transcription factors in all cells that show electrical excitability of their plasma membrane.