Synaptic activity induces signalling to CREB without increasing global levels of cAMP in hippocampal neurons

Authors

  • Anna Pokorska,

    1. MRC Laboratory of Molecular Biology, Division of Neurobiology, Hills Road, Cambridge, UK
    Search for more papers by this author
    • The present address of Anna Pokorska is Biofocus Science Park, Milton Road, Cambridge, CB4 0FG, England.

  • Peter Vanhoutte,

    1. MRC Laboratory of Molecular Biology, Division of Neurobiology, Hills Road, Cambridge, UK
    Search for more papers by this author
    • The present address of Peter Vanhoutte and Hilmar Bading is Department of Neurobiology, Interdisciplinary Center for Neurosciences (IZN),University of Heidelberg, Im Neuenheimer Feld 364, 69120 Heidelberg, Germany.

  • Fiona J. L. Arnold,

    1. MRC Laboratory of Molecular Biology, Division of Neurobiology, Hills Road, Cambridge, UK
    Search for more papers by this author
  • Francesca Silvagno,

    1. MRC Laboratory of Molecular Biology, Division of Neurobiology, Hills Road, Cambridge, UK
    Search for more papers by this author
  • Giles E. Hardingham,

    1. MRC Laboratory of Molecular Biology, Division of Neurobiology, Hills Road, Cambridge, UK
    Search for more papers by this author
    • The present address of Giles E. Hardingham is Department of Preclinical Veterinary Sciences, Royal (Dick) School of Veterinary Studies, Edinburgh University, Edinburgh EH9 1QH, UK.

  • Hilmar Bading

    1. MRC Laboratory of Molecular Biology, Division of Neurobiology, Hills Road, Cambridge, UK
    Search for more papers by this author
    • The present address of Peter Vanhoutte and Hilmar Bading is Department of Neurobiology, Interdisciplinary Center for Neurosciences (IZN),University of Heidelberg, Im Neuenheimer Feld 364, 69120 Heidelberg, Germany.


Address correspondence and reprint requests to Hilmar Bading, Department of Neurobiology, Interdisciplinary Center for Neurosciences (IZN), University of Heidelberg, Im Neuenheimer Feld 364, 69120 Heidelberg, Germany. E-mail: Hilmar.Bading@uni-hd.de

Abstract

Nuclear calcium signals associated with electrical activation of neurons can control the activity of the transcription factor cAMP-response element binding protein (CREB). Yet, cAMP is thought to be the key messenger that links synaptic activity to the regulation of CREB-mediated transcription. It is generally assumed that synaptic activity increases the intracellular levels of cAMP; this causes activation of the cAMP-dependent protein kinase (PKA) that regulates CREB-mediated transcription either directly or through controlling nuclear signalling of the MAP kinases/extracellular signal-regulated kinases (ERK1/2) pathway. Here we show that, in hippocampal neurons, synaptic activity failed to increase global levels of cAMP that would be required for the cAMP-PKA system to induce nuclear events. Even near-continuous bursting of action potentials, giving rise to large nuclear calcium signals and robust CREB-dependent transcription, left global intracellular levels of cAMP unchanged. These results suggest that the cAMP-PKA system does not function as the transducer of synaptic signals to the nucleus. They indicate that the known inhibitory effects of blockers of PKA on gene expression and long-lasting plasticity triggered by calcium entry reflect a gating function of basal activity of PKA that renders neurons permissive for nuclear calcium-regulated, CREB/CBP-dependent gene expression.

Abbreviations used
CBP

CREB-binding protein

CREB

cAMP-response element binding protein

ERK1/2

extracellular signal-regulated kinases

PKA

cAMP-dependent protein kinase

The generally accepted view is that transcriptional responses associated with electrical activation of neurons are triggered by calcium influx through NMDA receptors and/or l-type voltage-gated calcium channels (Szekely et al. 1989; Murphy et al. 1991; Lerea et al. 1992; Bading et al. 1993; Lerea and McNamara 1993; Bading et al. 1995; Bito et al. 1996; Deisseroth et al. 1996; Impey et al. 1996; Fields et al. 1997; Impey et al. 1998). However, it is controversial which second messenger links calcium entry to genomic events, in particular to the transcription factor complex CREB/CREB-binding protein (CBP) that plays a key role in learning processes in Drosophila, Aplysia and in the mammalian brain (Milner et al. 1998). One widely believed mechanism of signal propagation involves cAMP and the cAMP-dependent protein kinase PKA (Milner et al. 1998), a classical activator of CREB/CBP activity (Mayr and Montminy 2001). The production of cAMP following calcium entry results from the stimulation of calcium/calmodulin-dependent adenylyl cyclases (Poser and Storm 2001); globally elevated levels of cAMP activate PKA causing its translocation to the nucleus and activation of CREB/CBP-mediated gene expression (Mayr and Montminy 2001). PKA has also been reported to regulate nuclear import of molecules of the MAP kinase (ERK1/2) signalling pathway (Impey et al. 1998); this signalling pathway is activated by calcium entry into hippocampal neurons (Bading and Greenberg 1991) and can signal to CREB (Xing et al. 1996), although it is not sufficient to induce CREB/CBP-mediated transcription (Sheng et al. 1988; Chawla et al. 1998; Hardingham et al. 2001a, 2001b).

The cAMP-PKA signalling mechanism has received much attention in learning-related gene expression because inhibition of PKA in transgenic mice overexpressing the mutant regulatory subunit of PKA specifically impairs late-phase, transcription-dependent LTP and the animals' performances of learning tasks (Abel et al. 1997). In addition, several learning mutants in Drosophila (including dunce and rutabaga) have mutations in genes involved in cAMP metabolism (Milner et al. 1998), and, in the excitable cell lines PC12 and AtT20, inhibition of PKA by pharmacological or genetic means blocks calcium-influx induced gene expression (Ginty et al. 1991; Silvagno and Bading, unpublished observation). Thus, a large number of studies appear to indicate that cAMP functions as the messenger that links synaptic activity to genomic responses. However, this concept is challenged by observations that in certain neurons and excitable cell lines electrical stimuli only generate very small and short-lived or no increases in cAMP (Chetkovich et al. 1991; Frey et al. 1993; Impey et al. 1998).

An alternative concept for synapse-to-nucleus communication suggests that signal propagation is mediated by calcium itself: synaptically evoked calcium signals are propagated (perhaps in form of a regenerative calcium wave) to the cell nucleus (Hardingham et al. 2001a). Nuclear calcium stimulates nuclear calcium/calmodulin-dependent protein kinase IV, a potent activator of gene expression by CREB/CBP (Bading 2000). Nuclear calcium-induced, CREB-mediated gene expression is further supported by the MAP kinase (ERK1/2) pathway that is triggered by a submembranous calcium microdomain in the immediate vicinity of the site of calcium entry (Hardingham et al. 2001b); this signalling pathway, while not sufficient to activate CREB-dependent gene expression by itself (Sheng et al. 1988; Chawla et al. 1998; Hardingham et al. 2001a,b), can support nuclear calcium-mediated responses by prolonging the transcriptionally active (i.e. serine 133-phosphorylated) state of CREB following brief synaptic stimuli (Hardingham et al. 2001b; Wu et al. 2001). This study was aimed at clarifying which messenger transduces signals from the synapse to the nucleus.

Experimental procedures

Cells, calcium imaging, and multi-electrode array recordings

Hippocampal neurons were cultured from newborn Long-Evans rats as described (Bading and Greenberg 1991) except that growth media was supplemented with B27 (Gibco/BRL, Rockville, MD, USA). Hippocampal neurons were analysed after 10–14 days of culturing. During the culturing period hippocampal neurons develop a rich network of processes, express functional NMDA-type and AMPA/kainate-type glutamate receptors, and form synaptic contacts (Bading et al. 1995). Calcium imaging and multi-electrode array recordings were performed as described (Hardingham et al. 2001a).

cAMP measurements and nuclear signalling

Levels of cAMP were measured using a cAMP(125I) scintillation proximity assay system from Amersham (Amersham Pharmacia Biotech, Piscataway, NJ, USA). On day 10–14 of in vitro culturing hippocampal neurons were stimulated for various lengths of time followed by cell lysis and measurement of cAMP levels. The cAMP(125I) scintillation proximity assay system is based on the competition between unlabelled cAMP and a fixed quantity of 125I-labelled cAMP for a limited number of binding sites on a cAMP-specific antibody. The amount of radioactive ligand bound to the antibody was inversely proportional to the concentration of non-radioactive cAMP in the sample. The antibody-bound cAMP reacted with a reagent containing anti-rabbit secondary antibodies bound to fluoromicrospheres. Light produced by the fluoromicrospheres was measured in β-scintillation counter and the amount of bound labelled cAMP was calculated. The concentration of unlabelled cAMP in every sample was determined by interpolation from a standard curve. CREB phosphorylation on serine 133 was analysed by immunoblotting as described (Hardingham et al. 1997, 1999; Chawla et al. 1998). The antibodies to CREB phosphorylated on serine 133 and to calmodulin were from UBI.

Results

Intracellular levels of cAMP were measured in cultured hippocampal neurons before and after stimulating glutamatergic synaptic transmission using bath application of bicuculline. Bicuculline blocks GABAA receptors, thereby eliminating the tonic inhibition of synaptic activity imposed onto the neuronal network by GABAergic inhibitory interneurons that represent about 10% of the neuronal cell population (Vanhoutte and Bading, unpublished results). Bicuculline triggers bursts of action potential firing that occur periodically with frequencies of 0.05–0.15 Hertz (Hardingham et al. 2001a). These bursts are accompanied by large cytoplasmic and nuclear calcium transients and give rise to robust activation of CREB-mediated transcription (see Fig. 2b; Hardingham et al. 2001a). In some experiments hippocampal neurons were stimulated by exposing them to elevated levels of extracellular KCl (causing membrane depolarisation) or to glutamate. Either treatment causes large global calcium transients and robust genomic responses (Bading et al. 1993, 1995; Hardingham et al. 1999).

Figure 2.

(a) Multi-electrode array recordings of bursts of action potential in hippocampal networks following treatment with bicuculline in the presence or absence of 4-AP. (b) Imaging of global calcium transients in hippocampal neurons exposed to bicuculline (BiC, 50 μm) in the presence or absence of 4-AP; a typical example is shown, similar results were obtained in other experiments (see also Hardingham et al. 2001a,b). (c) Analysis of the intracellular levels of cAMP in hippocampal neurons before and after stimulation for 1, 2, 5, 10, and 20 min with bicuculline in the presence or absence of 4-AP. (n = 5).

Levels of cAMP were measured before and at various times (up to 20 min, the period relevant for immediate, signal-induced transcriptional events) after stimulation using a very sensitive scintillation proximity assay. We found that treatment of hippocampal neurons with 10 μm forskolin, an activator of adenylyl cyclases, caused a readily detectable increase in the levels of cAMP (Fig. 1a). In striking contrast, neither induction of bursts of action potentials with bicuculline, nor treatment with 50 mm KCl or 20 μm glutamate caused a change in the intracellular levels of cAMP (Figs 1b, c and d). Even under conditions where phosphodiesterases were blocked with 0.5 mm 3-isobutyl-1-methylxanthine (IBMX), that enhanced forskolin-induced increases in cAMP levels, stimulations with either bicuculline, glutamate or KCl failed to increase levels of cAMP (see Fig. 3a; data not shown).

Figure 1.

Analysis of the intracellular levels of cAMP in hippocampal neurons before and after stimulation for 1, 2, 5, 10, and 20 min with the indicated compounds. (a) n = 7; (b) n = 5; (c) n = 4; (d) n = 5.

Figure 3.

(a) Analysis of the intracellular levels of cAMP in hippocampal neurons before and after stimulation for 5 min with the indicated compounds. IBMX, 0.5 mm; 10 μm forskolin/0.5 mm IBMX; 50 μm bicuculline (Bic)/0.5 mm IBMX; 20 μm glutamate (Glu)/0.5 mm IBMX (n = 6). (b) Immunoblot analysis of CREB phosphorylation on serine 133 and calmodulin expression (loading control) in unstimulated hippocampal neurons and in hippocampal neurons treated for 5 min with the indicated concentration of forskolin or 10 μm forskolin/0.5 mm IBMX. (c) Quantitative analysis of the phospho-CREB immunoblot (n = 3). The signals obtained with the calmodulin antibodies were used to normalise the phospho-CREB immunoblot signals for protein loading.

To investigate the possibility that more intense neuronal firing is required to generate increased levels of cAMP, we treated hippocampal neurons with bicuculline and 4-amino pyridine (4-AP). 4-AP, a weak potassium channel blocker, increases the frequency of bicuculline-induced bursting (Hardingham et al. 2001a). Multi-electrode array recordings illustrate the high frequency bursting of hippocampal neurons treated with bicuculline/4-AP that give rise to large global calcium transients (Figs 2a and b). However, even those conditions failed to cause a detectable increase in the levels of cAMP (Fig. 2c).

Our experiments do not rule out the possibility that the stimulations studied generate a very small increase in cAMP that is not detectable with the methods used. We therefore established conditions that generate small but detectable increases in cAMP and tested whether those levels of cAMP are sufficient to signal to CREB. Titration of the concentration of forskolin used to stimulate the neurons revealed that 2 μm forskolin generated a two-fold increase in the levels of cAMP (Fig. 3a). However, 2 μm forskolin was not sufficient to cause phosphorylation of CREB on its activator site serine 133 (which represent a key signal in the induction of CREB-dependent transcription) (Figs 3b and c). Thus, small yet clearly detectable increases in cAMP are unable to trigger signalling pathways to CREB. Given that none of the treatments used to electrically activate hippocampal neurons led to a detectable increase in the levels of cAMP, it appears very likely that messengers other than cAMP link synaptic activity to CREB-mediated gene expression.

Discussion

Two types of signals, elevated levels of cAMP and nuclear calcium, are known to be sufficient to induce CREB/CBP-mediated gene expression (Bading 2000; Mayr and Montminy 2001). Previous studies have extensively documented that electrical stimuli activating CREB/CBP-dependent gene expression are associated with calcium signals (Bading et al. 1995; Deisseroth et al. 1996; Fields et al. 1997; Hardingham et al. 1997, 1999; Nakazawa and Murphy 1999; Hardingham et al. 2001a) and that inhibition of nuclear calcium blocks transcriptional responses mediated by CREB/CBP (Hardingham et al. 1997; Chawla et al. 1998). While mechanisms involving calcium/calmodulin-sensitive adenylyl cyclases have been described through which a calcium signal can be converted into a cAMP signal (Poser and Storm 2001), the role of cAMP in the transmission of synaptic calcium signals to the nucleus was unclear. Our study documents that several different stimulation paradigms that electrically activate hippocampal neurons and evoke genomic responses do not lead to a detectable increase in global levels of cAMP. These findings strongly suggest that nuclear calcium rather than cAMP links synaptic activity to CREB/CBP-mediated transcription.

While our results do not explain the known inhibition by blockers of PKA of calcium influx-induced transcriptional responses and plasticity, they suggest that these effects are due to a reduction of basal levels of PKA activity in neurons. Basal PKA activity may be required for rendering neurons permissive for CREB/CBP-dependent transcriptional responses. This raises the possibility that PKA has a ‘gating function’ that allows synaptic activity and nuclear calcium signalling pathways to stimulate gene expression mediated by CREB/CBP. The mechanisms through which PKA gates activity-induced transcriptional responses are unknown; they could involve the inactivation of a putative nuclear factor that prevents CREB from interacting with CBP (Mayr et al. 2001). Alternatively, it is conceivable that calcium-induced nuclear translocation of molecules of the MAP kinase (ERK1/2) cascade requires basal levels of PKA activity (Impey et al. 1998). Stimulation of the MAP kinase (ERK1/2) pathway prolongs the transcriptionally active (i.e. serine 133-phosphorylated) state of CREB in response to short-lasting calcium signals (Hardingham et al. 2001b; Wu et al. 2001). Therefore, the possible inhibition, by reducing basal levels of PKA activity, of nuclear import of one or several components of the MAP kinase (ERK1/2) cascade would primarily affect CREB/CBP-mediated transcription induced by very brief synaptic stimuli. This mechanism cannot explain why transcriptional responses induced by prolonged increases in intracellular calcium are also compromised by PKA inhibitors, as under those conditions the activation of the MAP kinases (ERK1/2) pathway is dispensable for induction of CREB-regulated gene expression (Johnson et al. 1997; Hardingham et al. 1999, 2001a).

The lack of increases in global levels of cAMP following electrical activation of hippocampal neurons let it appear unlikely that the cAMP-PKA system is the signal transducer in synapse-to-nucleus communication. However, our results do not rule out the possibility that synaptic activity and calcium entry can, through stimulating calcium/calmodulin-dependent adenylyl cyclases, give rise to localised increases in the levels of cAMP and PKA activity. Such spatially confined, signal-induced stimulations of the cAMP-PKA signalling pathway may account for some of the phosphorylation events on ion channels and neurotransmitter receptors (Sigel 1995; Swope et al. 1999). In addition, increases in cAMP levels in subcellular microdomains could synergise with nuclear calcium signals in activity-dependent gene expression in neurons.

Acknowledgements

This work was supported by the MRC, Affymax, a Marie Curie Fellowship from the European Community (PV), and the Alexander von Humboldt Foundation (HB).

Ancillary