Address correspondence and reprint requests to Dr M. Dragunow, Department of Pharmacology, Faculty of Medicine and Health Science, University of Auckland, Auckland 1003, New Zealand. E-mail: email@example.com
This study describes the effect of signalling through muscarinic acetylcholine receptors on two transcription factors implicated in long-term synaptic plasticity and memory formation, EGR1 and the cyclic AMP response element binding protein (CREB). In SK-N-SH neuroblastoma cells, treatment with the cholinergic agonist carbachol led to maximal induction of EGR1 1 h after stimulation. This was preceded by the phosphorylation of CREB, which peaked as early as 5 minutes after carbachol treatment. The levels of both EGR1 and phosphorylated CREB (pCREB) slowly decayed over 4–8 h. CREB phosphorylation and EGR1 induction showed similar sensitivity to carbachol concentration, with EC50 values in the range of 1–10 µM, and the changes in both transcription factors were blocked by the muscarinic antagonist atropine. As has been described elsewhere, EGR1 induction was dependent on activation of p42/44 MAP kinase, as it was blocked by the MEK inhibitor U0126. However, CREB phosphorylation by carbachol was largely unaffected by MAP kinase blockade. As both CREB phosphorylation and EGR1 induction have been linked to long-term potentiation and some forms of memory consolidation, these results may implicate CREB and EGR1 in independent or partially independent cholinergic signalling pathways involved in memory processes.
extracellular signal-regulated protein kinases 1 and 2
muscarinic acetylcholine receptor
mitogen-activated protein kinase kinase
protein kinase C
sodium dodecyl sulfate
The cholinergic system in the CNS has long been considered to have a role in learning and memory processes, and is important from a clinical perspective in neurodegenerative disorders such as Alzheimer's disease (Bartus et al. 1982; Levey 1996; Sirvio 1999). Cholinergic signalling through muscarinic receptors (mAChRs) has been linked to long-term changes in synaptic plasticity which may underlie memory formation (Blitzer et al. 1990; Jerusalinsky et al. 1997; Calabresi et al. 1999), but the molecular basis for its contribution to these processes is still not well understood. Consolidation of newly formed memories requires protein synthesis (Squire and Davis 1981), and the activation of specific inducible and constitutive transcription factors provides a means of initiating programmes of gene expression relevant to memory consolidation. Active research over the last few years has identified some of these candidate ‘memory molecules’.
Another transcription factor which has emerged as a prime candidate memory molecule, not only in mammals but also in flies and molluscs, is the cyclic AMP response element binding protein (CREB; Silva et al. 1998). CREB is a member of the large CREB/ATF family of basic leucine zipper transcription factors, and binds to cyclic AMP response element (CRE) sites in target promoters. Expressed constitutively in most cells, CREB becomes transcriptionally active following phosphorylation at Ser133 (for a review see Mayr and Montminy 2001). In neuronal cultures from the marine snail, Aplysia californica, CREB is required for long-term facilitation, a form of plasticity that parallels reflexive behaviour (Dash et al. 1990). CREB inhibition in Drosophila blocks long-term olfactory memory (Yin et al. 1994), while augmentation of CREB activity can actually enhance memory (Yin et al. 1995). In the mouse, targeted genetic disruption of two main CREB isoforms leads to impaired performance in several behavioural learning tasks and a rapidly decaying form of LTP in the hippocampus (Bourtchuladze et al. 1994). In addition, CREB disruption has been shown to impair long-term spatial memory in the rat (Guzowski and McGaugh 1997). These examples, in addition to demonstrations of increased phosphorylated CREB (pCREB) in the hippocampus following LTP induction and training on behavioural learning tasks (Impey et al. 1996, 1998; Schulz et al. 1999) provide compelling evidence for the involvement of CREB in initiating transcriptional programmes relevant to long-term memory formation.
Recently, studies have demonstrated that activation of mAChRs leads to phosphorylation and activation of the mitogen-activated protein (MAP) kinases, p44 and p42 extracellular signal-regulated protein kinase 1 and 2 (ERK1/2) in the hippocampus (Roberson et al. 1999; Rosenblum et al. 2000), and that this ERK1/2 activation is required for EGR1 induction in SH-SY5Y cells (Grimes and Jope 1999). The MAP kinase cascade has also recently been demonstrated to be involved in mammalian memory and learning (Atkins et al. 1998; Blum et al. 1999). As activation of mAChRs activates EGR1 through phosphorylated ERK1/2 (pERK1/2), and the MAP kinase pathway has also been shown to lead to CREB phosphorylation in hippocampal slices (Roberson et al. 1999), we speculated that mAChR activation induces both EGR1 and pCREB via activation of pERK1/2. To test this hypothesis we used SK-N-SH human neuroblastoma cells, which express muscarinic receptors (Fisher and Heacock 1988; Wall et al. 1991; Baumgartner et al. 1993) and which have been used as a model for mAChR-mediated signalling processes (Stubbs and Agranoff 1993; Shariat-Madar et al. 1997; Popova and Rasenick 2000). In this study we describe the phosphorylation of CREB by mAChR activation in SK-N-SH cells and the relationship of this to EGR1 induction and the activation of ERK1/2.
Materials and methods
All cell culture media components were purchased from Gibco Invitrogen Corp. (Carlsbad, CA, USA) with the exception of trypsin, which was purchased from Difco (Detroit, MI, USA). Carbachol, atropine sulfate, 3,3′-diaminobenzidine (DAB), o-phenylenediamine dihydrochloride (OPD), biotinylated goat anti-rabbit antibody and ExtrAvidin peroxidase were purchased from Sigma (St Louis, MO, USA). U0126 was purchased from Cell Signaling Technology (Beverly, MA, USA). Sources of rabbit polyclonal antibodies were as follows: EGR1 (sc-189) from Santa Cruz Biotechnology (Santa Cruz, CA, USA); Ser133-pCREB from Upstate Biotechnology (Lake Placid, NY, USA); CREB, pERK1/2 and ERK1/2 from Cell Signaling Technology. Peroxidase-conjugated anti-rabbit antibody, polyvinylidene difluoride (PVDF) membrane and enhanced chemiluminescence (ECL) and ECL+Plus reagents were purchased from Amersham Pharmacia Biotech (Piscataway, NJ, USA). Protein assay reagents were purchased from Bio-Rad (Hercules, CA, USA). All other chemicals were purchased from Sigma or BDH (Poole, Dorset, UK).
Human SK-N-SH neuroblastoma were obtained from the American Type Culture Collection (ATCC). They were cultured in 100 mm-diameter dishes at 37°C in a humidified atmosphere (5% CO2/95% air) in RPMI medium supplemented with 10% foetal bovine serum, 100 U/mL penicillin, 100 µg/mL streptomycin and 2 mM l-glutamine. Cells grown to confluency were harvested with 0.2% trypsin in phosphate-buffered saline (PBS), washed with media and plated out 18–24 h prior to drug addition. Cells were plated at 2 × 104 cells/well in 96-well plates for immunocytochemistry and cell-based ELISA, and at 6 × 105 cells/well in 6-well plates for western blot analysis. Where indicated, atropine sulfate (10 µM or 10 nM) or U0126 (10 µM) were added 5 min and 1 h, respectively, prior to addition of carbachol.
Cells were fixed in 2% paraformaldehyde, and immunostaining was carried out as described by Walton et al. (1998) with minor changes. All buffer washes contained 0.2% Triton X-100 and primary antibody incubations were overnight at room temperature (19°C). Dilutions of rabbit polyclonal antibodies were as follows: pCREB, 1/500; EGR1, 1/1000; pERK1/2, 1/250. Antibody binding was visualised by DAB staining following incubation with biotinylated secondary antibody and ExtrAvidin peroxidase.
Following the indicated treatments, cells were washed with cold PBS, lysed in sodium dodecyl sulfate (SDS) sample buffer, and boiled for 5 min. After quantification by Bio-Rad DC protein assay, 20 µg protein samples were separated on SDS polyacrylamide gels and transferred to PVDF membrane using standard procedures. After blocking for 1 h in Tris-buffered saline, 5% skim milk powder, 0.1% Tween 20, membranes were incubated overnight at 4° in primary antibody. Dilutions of rabbit polyclonal antibodies were as follows: CREB, 1/1000; pCREB, 1/1000; EGR1, 1/5000; ERK1/2, 1/1000; pERK1/2, 1/1000. Blots were incubated in peroxidase-conjugated secondary antibody at 1/5000 dilution for 45 min at room temperature and developed using ECL reagents. Where indicated, blots were stripped by incubating in 100 mM 2-mercaptoethanol, 2% SDS, 62.5 mM Tris HCl pH 6.8 for 30 min at 50°.
Densitometric quantification of immunostaining was performed using the MD30+ image analysis system (Leading Edge Software, Australia) on images captured from an Ikegami ICD-42E CCD camera mounted on a Leica DMIRB inverted microscope. Measurements of integrated staining density/total cell area were averaged from 4 fields per well at 200× magnification. Data were collected from 4 wells per treatment and were normalised to control (H2O) values.
Quantification of immunostaining by cell-based ELISA followed the procedure of Versteeg et al. (2000), with minor modifications. PBS-Triton contained 0.2% Triton X-100, the blocking solution used was 10% goat serum in PBS-Triton, and antibodies were diluted in PBS-Triton, 1% goat serum, 0.04% merthiolate. Dilutions of antibodies, optimised for sensitivity and low background, were: pCREB, 1/1000; EGR1, 1/, 2000; peroxidase-conjugated secondary antibody, 1/200. Peroxidase conversion of the substrate OPD was determined by absorbance measurement at 490 nm (A490) corrected for light scatter at 650 nm. Where required, correction for variation in cell number was achieved by crystal violet staining and measurement of A595 (Versteeg et al. 2000). Measurements of A490-A650 (or (A490-A650)/A595) were averaged for at least 4 replicate wells and were normalised to control (H2O) values. Statistical analysis was performed using one-way anova followed by a Tukey-Kramer post hoc multiple comparison test.
Muscarinic signalling leads to CREB phosphorylation and EGR1 expression in SK-N-SH cells
To investigate the effects of muscarinic receptor activation on CREB phosphorylation and EGR1 induction, immunocytochemistry and Western blot analyses were performed on SK-N-SH neuroblastoma cells following treatment with the cholinergic agonist carbachol (100 μM). Carbachol treatment for 10 min resulted in a marked increase in nuclear pCREB immunostaining relative to control (Figs 1a and b). An increase in nuclear EGR1 immunostaining was observed 60 min after carbachol treatment (Figs 1c and d), with staining intensity more variable than that seen for pCREB at 10 min. Similar results were seen on western blots using the same primary antibodies (Fig. 1e), where pCREB was strongly induced 10 min following carbachol treatment, but less-strongly induced at 60 min. Conversely, EGR1 protein was at control levels at 10 min after carbachol treatment but was strongly induced at 60 min. The decline in pCREB immunostaining 60 min after carbachol treatment appeared to be due to a decrease in phosphorylation per se rather than reduction in overall CREB levels, as the loading control immunostaining for total CREB showed comparable levels at 60 min after carbachol addition (Fig. 1e).
Carbachol dose–response curves for CREB phosphorylation and EGR1 induction in SK-N-SH cells were generated by densitometric analysis of immunostained cells (Figs 2a and c) and by cell-based ELISA (Figs 2b and d). For analysis of pCREB and EGR1, cells were fixed 10 min and 60 min, respectively, after carbachol treatment. In the densitometric analysis, both phosphorylation of CREB (Fig. 2a) and induction of EGR1 (Fig. 2c) showed similar sensitivity to carbachol concentration with approximate EC50 values of 10 µm. The muscarinic antagonist atropine (at 10 µm) completely blocked CREB phosphorylation and EGR1 induction when added 5 min prior to carbachol, indicating that carbachol was driving these responses through activation of muscarinic receptors. Following a recently described cell-based ELISA procedure (Versteeg et al. 2000), a similar carbachol EC50 value was obtained for EGR1 (Fig. 2d) but the EC50 for pCREB was closer to 1 µm (Fig. 2b). Pre-treatment with atropine (at 10 nm) produced a right shift in the carbachol dose–response curves, consistent with competitive inhibition of muscarinic receptors. Although these two methods were used in part to cross-validate each other, the cell-based ELISA procedure appeared to give greater accuracy and precision, and was adopted for further experiments.
A more complete time course analysis of CREB phosphorylation and EGR1 induction was carried out by cell-based ELISA of SK-N-SH cells after treatment with carbachol (100 µM) for various times. CREB phosphorylation was rapidly induced by carbachol, with a peak at 5 min and gradual decline back to control levels by about 4 h (Fig. 3a). This increase in CREB phosphorylation preceded an increase in EGR1 immunostaining, which peaked at 60 min after carbachol treatment and declined back to control levels by about 8 h (Fig. 3b).
CREB phosphorylation in SK-N-SH cells by carbachol is not dependent on MEK1/2 activity, in contrast to EGR1 induction
The temporal relationship of CREB phosphorylation and EGR1 induction in SK-N-SH cells raises the possibility that CREB and EGR1 lie on the same signalling pathway downstream of muscarinic receptor activation. It has been reported that induction of EGR1 by muscarinic signalling in the SH-SY5Y cell line, a clonal derivative of SK-N-SH cells, requires activation of mitogen activated protein (MAP) kinase kinases 1/2 (MEK1/2; Grimes and Jope, 1999). The involvement of the MAP kinase cascade in muscarinic cholinergic induction of pCREB and EGR1 in SK-N-SH cells was investigated by preincubating cells with the MEK inhibitor U0126 (10 µM) before addition of carbachol (Favata et al. 1998). The inhibitory action of U0126 was confirmed by immunocytochemical and western blot staining with an antibody to the activated phosphorylated forms of ERK1/2 (pERK1/2), the immediate downstream targets of MEK in the MAP kinase cascade (Fig. 4). Pre-treatment with U0126 substantially blocked an increase in pERK1/2 cytoplasmic and nuclear staining observed 5 min after addition of 100 µM carbachol (Fig. 4A). In addition, the induction of EGR1 nuclear staining seen after 60 min carbachol treatment was also blocked by U0126. However, the increase in pCREB nuclear staining 10 min after carbachol addition was not inhibited by U0126. These results were supported by western blot analysis (Fig. 4b) and cell-based ELISA (Fig. 4c), which both showed inhibition of basal and carbachol-induced pERK1/2 and complete inhibition of carbachol-induced EGR1 by U0126. Furthermore, the western blot analysis indicated that CREB phosphorylation induced by carbachol was resistant to U0126 treatment. Cell-based ELISA analysis of CREB phosphorylation at various time points after treatment with 100 µM carbachol did not show a statistically significant decrease in carbachol-induced pCREB levels by U0126, although a consistent small decrease was evident at all the time points (Fig. 4d). Interestingly, the cell-based ELISA revealed an increase of basal pCREB by U0126 (Fig. 4d, 0 and 10 min, p < 0.01). Although statistically significant, this increase is subtle and has not been observed by the less sensitive western blotting procedure (Fig. 4b). Pooling results from several independent experiments, the carbachol-induced increase in CREB phosphorylation at 10 min with U0126 pre-treatment is 73 ± 6% that of control (mean ± SD, n = 8). This compares with an induction level for EGR1 1 h after carbachol treatment in the presence of U0126 of 0 ± 6% of control (n = 3). Taken together these data suggest that, unlike induction of EGR1, phosphorylation of CREB downstream of muscarinic receptor activation is relatively independent of MEK and the MAP kinase cascade in SK-N-SH cells.
In this study we report the induction of EGR1 and phosphorylation of CREB by mAChR activation in SK-N-SH neuroblastoma cells, and the dependence of EGR1 but not pCREB on the ERK1/2 MAP kinase pathway. EGR1 induction through activation of mAChRs is well established in the literature, both in the CNS (Hughes and Dragunow 1994; Tsiokas and Watson 1995) and in cell culture models relevant to the CNS, such as in rat astrocytes (Arenander et al. 1989), the neuroblastoma/glioma hybrid NG108-15 (Katayama et al. 1993), sublines of neuronal-like PC12 cells (Morita and Wong 1996; Ebihara and Saffen 1997) and the SK-N-SH derivative line, SH-SY5Y (Grimes and Jope 1999). This mAChR-dependent induction of EGR1 has also been demonstrated in transfected cell lines expressing mAChRs (von der Kammer et al. 1998). In addition to the human neuroblastoma SK-N-SH cells reported here, we have also observed EGR1 induction by carbachol in Neuro-2a mouse neuroblastoma cells, and the muscarinic agonist oxotremorine M induces EGR1 in both cell lines (unpublished results).
Unlike the uniform pattern of immunostaining for pCREB in carbachol-treated SK-N-SH cells, EGR1 immunostaining was quite variable. Although the reason for this is not known, it may suggest a degree of cell-cycle dependence of EGR1 expression or some heterogeneity in expression of mAChR subtypes or downstream signalling molecules in these cells. EGR1 induction has been associated with a wide variety of stimuli influencing cell-cycle entry and exit (for a review see Gashler and Sukhatme 1995). There is also variation in mAChR repertoire and expression levels reported in SK-N-SH cells and subcloned derivative lines (Fisher and Heacock 1988; Murphy et al. 1991; Wall et al. 1991; Baumgartner et al. 1993). Whatever the reason is for the variable expression of EGR1 in response to carbachol, this does not apply to the phosphorylation of CREB by the same stimulus.
The induction of EGR1 by carbachol in SK-N-SH cells peaked at 60 min and decayed back to control levels by 8 h. This result is consistent with the work of Grimes and Jope (1999), who showed that carbachol induced EGR1 protein with a similar time course in SH-SY5Y cells, a clonal derivative of SK-N-SH. In that study, the MEK inhibitor PD098059 (50 µM) reduced carbachol-stimulated EGR1 DNA binding activity by 60% Grimes and Jope (1999). However, we find that inhibition of MEK1/2 by 10 µM U0126 is sufficient to completely abolish EGR1 induction by carbachol. This may be due to the greater potency of U0126 over PD098059 as an inhibitor of both MEK 1 and 2 (Favata et al. 1998; Roberson et al. 1999) or slight variations in receptor repertoire or signalling cascades in the two related cell lines.
In different cell types the specific repertoire of muscarinic receptor subtypes will greatly influence the cellular response to receptor activation. SK-N-SH cells are reported to predominantly express the M1 and M3 subtypes of muscarinic receptor, both of which are broadly classed as coupling to G-proteins of the Gq/11 family, leading to phospholipase Cβ activation and mobilisation of intracellular calcium (Fisher and Heacock 1988; Wall et al. 1991; Baumgartner et al. 1993). It is likely that one or a combination of these receptor subtypes are primarily responsible for driving the transcription factor responses that we observe, but the question of which receptor subtypes and whether these responses originate from distinct receptors or G-proteins awaits further analysis.
Treatment of SK-N-SH cells with carbachol resulted in a rapid increase in CREB phosphorylation, with a peak at 5 min, and a gradual decline back to control levels over the next 4–8 h. This result provides an interesting contrast to some effects of mAChR activation in different cell types. Activation of mAChRs in parotid acinar cells decreases CREB phosphorylation (Takuma et al. 1997) and withdrawal from chronic treatment with carbachol in NG108-15 cells results in increased CREB phosphorylation (Thomas et al. 1995). In pancreatic β cells, carbachol stimulates CRE-directed gene expression (Eckert et al. 1996), and in cortical oligodendrocyte precursor cells carbachol treatment leads to CREB phosphorylation (Pende et al. 1997), but current evidence points to the involvement of protein kinase C (PKC) and the MAP kinase pathway in this process (Pende et al. 1997; Sato-Bigbee et al. 1999). Roberson et al. (1999) show that in the hippocampus mAChR activation leads to ERK1/2 activation by a PKC-dependent mechanism, and that CREB phosphorylation downstream of PKC activation can be partially blocked by MEK inhibition. These findings suggest that muscarinic receptor activation leads to CREB phosphorylation, with ERK1/2 activation being a critical step in this process. However, what we find in SK-N-SH cells is that, unlike induction of EGR1, carbachol-induced CREB phosphorylation is largely independent of ERK1/2 activation. Carbachol dose–response curves indicated a slight difference in sensitivity of CREB phosphorylation and EGR1 induction, with approximate EC50 values of 1 µM and 10 µM, respectively. Although this suggests that these responses may involve different combinations of receptors and upstream signalling pathways, it is also possible that this difference in EC50 is a consequence of receptor down-regulation over the longer time period of EGR1 induction. Indeed, the rapid induction of MAP kinase phosphorylation by carbachol has a similar EC50 to that of CREB phosphorylation (results not shown). Interestingly, the time course of elevated CREB phosphorylation, which extends to at least 2 h following carbachol addition and which is largely independent of MAP kinase activation throughout, has the characteristics of a slower more-sustained CREB phosphorylation which is thought to be MAP kinase-dependent (Wu et al. 2001). Taken together, these data provide evidence for an ERK1/2-independent pathway for sustained CREB phosphorylation downstream of mAChR activation in SK-N-SH cells. CREB acts as an integration point for a variety of upstream signals, and the strength and duration of its activation in concert with other transcription factors influences the programme of downstream transcriptional responses (Bonni et al. 1995). In this regard, signalling through mAChRs may modulate neuronal transcriptional responses to glutamate, the primary neurotransmitter implicated in learning and memory processes and induction of long-term synaptic plasticity. Glutamate-mediated induction of EGR1 and phosphorylation of CREB have both been shown to be downstream of the ERK MAP kinase pathway (Sgambato et al. 1998; Perkinton et al. 1999; Vanhoutte et al. 1999; Davis et al. 2000), which itself appears to be essential for some forms of LTP and long-term memory formation (for a review see Impey et al. 1999). EGR1 induction can be driven by ERK phosphorylation of the transcription factor Elk-1 which targets the serum-responsive element (SRE) in the EGR1 promoter (Sgambato et al. 1998; Davis et al. 2000). Signalling through mAChRs may also contribute to glutamate-driven increases in EGR1 and pCREB, but the latter by a distinct ERK1/2-independent pathway.
In conclusion, this report describes the phosphorylation of CREB and the induction of EGR1 downstream of muscarinic acetylcholine receptor activation in SK-N-SH neuroblastoma cells. The novel findings are that pCREB is induced downstream of muscarinic receptor activation and this phosphorylation, unlike EGR1 induction, is independent of ERK1/2 activation. Determining the contribution of acetylcholine-mediated activation of these transcription factors to downstream gene expression and memory processes awaits further study.
This work was supported by The Marsden Fund of New Zealand and the New Zealand Health Research Council. We are grateful for technical assistance from Claire Henderson and Niqi Butterworth.