Neuronal processes in associative learning and memory such as classical and operant conditionings are likely underlain by the integration of different intracellular signals evoked in response to separate synaptic inputs (LeDoux 2000; Kandel 2001; Kelley 2004). In long-term forms of learning and memory, expression of the gene involved in synaptic plasticity can be up- or down-regulated as a target of these converging signals, and can work as a signal integrator of distinct synaptic inputs. A typical model for the relevant mammalian signal transduction from the neuronal cell surface to the nucleus is as follows: the second messengers Ca2+ and cAMP activate multiple kinase pathways including calcium/calmodulin-dependent protein kinase (CaMK), protein kinase A (PKA), and extracellular signal-regulated protein kinase (ERK), which in turn phosphorylate transcription factors such as the cAMP responsive element (CRE)-binding protein (CERB). This results in the activation of target genes for growth factors such as brain-derived neurotrophic factor and transcription factors including Fos and Zif268/Egr1, which can further control a number of downstream genes (Hyman et al. 2006; Flavell and Greenberg 2008; Pape and Pare 2010). Given that this model is also applicable to the molecular mechanisms of associative learning and memory, fundamental questions are raised regarding these processes. What is the combination of synaptic inputs? What is the target gene that functions as a signal integrator? Surprisingly, to date, the knowledge of genes that are regulated cooperatively by different neurotransmitters is very limited.
To address this problem, we first screened genes regulated by Ca2+ and cAMP via complementary DNA (cDNA) microarray analysis. Subsequent investigations for related neurotransmitters revealed that the secretogranin II (SgII) gene (Scg2) is activated cooperatively by glutamate and dopamine, and repressed by GABA. SgII is a member of the granin family of acidic secretory proteins (Fischer-Colbrie et al. 1995; Taupenot et al. 2003), and is processed to a number of small peptides called secretoneurin (Kirchmair et al. 1993; Wiedermann 2000), BM66 (Anouar et al. 1998), and manserin (Yajima et al. 2004). Various physiological roles have been assigned to SgII. In particular, the most well-characterized peptide, secretoneurin, has been found to be involved in hormone secretion (Nicol et al. 2002), angiogenesis (Kirchmair et al. 2004), neurotransmitter release (Saria et al. 1993; You et al. 1996), neurite outgrowth (Gasser et al. 2003), and neuroprotection (Shyu et al. 2008). Previous studies showed that Scg2 is activated by the cAMP pathway via a CRE in the Scg2 promoter (Cibelli et al. 1996; Scammell et al. 2000) and by the Ca2+ pathway depending on the voltage-gated Ca2+ channel (VGCC) (Fujita et al. 1999). However, the cognate extracellular agonists have not been reported, except for gonadotropin-releasing hormone, which activates Scg2 in pituitary gonadotrope lineage cells via the PKA/CREB pathway (Song et al. 2003) and the mitogen-activated protein kinase (MAPK)/activating transcription factor 3 pathway (Xie and Roberson 2008). This study indicates that Scg2 is activated cooperatively by representative fast (depolarizing) and slow (modulatory) neurotransmitters, that is, glutamate and dopamine, respectively. This finding, taken together with the characterization of the Scg2 regulatory mechanisms mediated by these neurotransmitters, leads us to postulate that Scg2 is a good candidate for the signal integrator required for activity-dependent plasticity such as associative learning and memory.
- Top of page
- Materials and methods
- Supporting Information
In this study, we showed that Scg2 mRNA levels in primary cultured hippocampal neurons were elevated by glutamate, dopamine, and noradrenaline, and lowered by GABA. The converse regulatory effects of the fast neurotransmitters, that is, glutamate (activating or depolarizing) and GABA (inhibiting or hyperpolarizing), are concordant with the previously suggested notion that Scg2 expression is regulated in a neuronal activity-dependent manner (Nedivi et al. 1993; Konopka et al. 1995; Fujita et al. 1999), and are also concordant with the present finding that TEA, a K+ channel blocker causing depolarization, activates Scg2. On the other hand, the activation of Scg2 by the slow neurotransmitters dopamine and noradrenaline suggests that the modulation of Scg2 expression is complex. Interestingly, Scg2 was cooperatively activated by glutamate and dopamine; therefore, we characterize Scg2 as a candidate for the signal integrator of these synaptic inputs. However, it remains to be confirmed that the cooperative activation of Scg2 by glutamate and dopamine is executed in each individual neuron. While it cannot be definitively ruled out that glutamate and dopamine stimulate separate neurons, the possibility that all the neurons in the culture respond to only glutamate or dopamine is minor.
Interaction of the intracellular signals induced by glutamate and dopamine inputs seems to underlie a number of plastic changes in the CNS; however, it is a major challenge to demonstrate definitively the causal relationship between cellular biochemistry and behavioral physiology in a single neuron. Medium spiny neurons (MSNs) in the dorsal striatum and the ventral striatum (the nucleus accumbens) are characterized by dense corticostriatal/thalamostriatal glutamatergic and mesostriatal dopaminergic inputs (for review, see David et al. 2005), and have served as a model system for investigation of the consequences of these in-phase inputs. One group of MSNs predominantly expresses D1-type dopamine receptors (D1Rs), while the other group expresses D2Rs. Murine D1R and D2R MSNs in the dorsal striatum (Bateup et al. 2010; Hikida et al. 2010; Kravitz et al. 2010), as well as in the nucleus accumbens (Beutler et al. 2011; Pascoli et al. 2012), play distinctive roles in physiological and addictive behaviors. In striatal D1R and D2R MSNs, dopaminergic inputs are required for the induction of NMDAR-dependent LTP and mGluR5-dependent LTD, respectively (Shen et al. 2008). In the CA1 region of the hippocampus, the NMDAR-dependent LTP can be separated into a protein synthesis-independent transient early LTP and a protein synthesis-dependent prolonged late LTP; the latter requires dopaminergic inputs on D1/D5Rs (Navakkode et al. 2007; for review, see Lisman et al. 2011). It remains to be determined whether Scg2 activation by glutamate and dopamine is involved in these plasticities.
Intracellular glutamatergic and dopaminergic signals can converge at multiple steps in their signaling pathways. First, D1R directly interacts with the NMDAR subunits NR1 and NR2A (Lee et al. 2002; Fiorentini et al. 2003), and D2R interacts with the NMDAR subunit NR2B (Liu et al. 2006). Second, the activation of ERK, a MAPK critical for long-term synaptic changes, requires the simultaneous stimulation of D1R and NMDAR in the striatum (Valjent et al. 2005). D1R stimulation leads to cAMP/PKA pathway activation followed by inactivation of the phosphatase cascade of protein phosphatase-1 and striatal-enriched tyrosine phosphatase, allowing effective phosphorylation of ERK by MEK in response to elevation of intracellular levels of Ca2+ channeled via NMDAR. Third, in the regulation of genes that are activated during plastic changes, such as bdnf and c-fos (Flavell and Greenberg 2008; Day and Sweatt 2011), multiple transcriptional and epigenetic regulators can be targets of both glutamatergic and dopaminergic inputs. In this study, the activation of the Scg2 gene by either glutamate or dopamine was inhibited by the Ca2+ chelator BAPTA-AM or the MEK inhibitor PD98059. These results suggest that at least one converging point of glutamatergic and dopaminergic signals is located within the functional realm of the second messenger Ca2+ and its downstream Ras/Raf/MEK/ERK pathway.
Ca2+ mobilization can exhibit differential effects on gene expression, depending on the mobilization route (Greer and Greenberg 2008). Binding of glutamate to NMDAR and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) causes respectively direct and indirect Ca2+ influx via depolarization-triggered opening of L-type VGCC, while GluR2-lacking AMPAR can also conduct direct Ca2+ influx (Jonas et al. 1994). The present study suggests that NMDAR is involved in Scg2 activation (Fig. 5b), although the roles of other glutamate receptors remain to be examined. With respect to dopamine, in addition to the aforementioned direct interaction of D1R or D2R with NMDAR Ca2+ channels, D1R physically interacts with N-type VGCC (Kisilevsky et al. 2008). Stimulation of D1R-D2R and D2R-D5R heteromers causes activation of the phospholipase C/inositol triphosphate (IP3) pathway, leading to activation of the IP3 receptor (IP3R) followed by intracellular store-derived Ca2+ mobilization (Hasbi et al. 2009). In the case of D2R-D5R stimulation, this results in the influx of extracellular Ca2+ (So et al. 2009). In primary cultured striatal MSNs, D1R agonists activate both the phospholipase C/IP3 and cAMP/PKA pathways, which converge on IP3R and cooperatively induce Ca2+ oscillation (Tang and Bezprozvaany 2004). In primary cultured hippocampal neurons, the stimulation of D1R promotes mGluR-induced Ca2+ oscillation in a PKA-dependent manner (Dai et al. 2008), postulating a candidate mechanism underlying the signal integration of glutamatergic and dopaminergic inputs. Further investigations are required to clarify how dopamine mobilizes Ca2+ in our experimental system.
Characteristically, Scg2 seems to be under the control of a transcriptional repressor(s) that is potentiated by the KT5720-sensitive pathway in response to dopamine and noradrenaline (Fig. 7). Previous studies (Scammell and Valentine 1994; Jones and Scammell 1998) that examined the rat pituitary cell line GH4C1 have also suggested the presence of a cycloheximide-sensitive repressor for basal and forskolin-induced Scg2 expression, and its relationship with the putative repressor of this study remains to be examined. The present KT5720- and cycloheximide-sensitive repressor seems to be different from ICER (Sassone-Corsi 1998; Mioduszewska et al. 2003), which had an abundance that was below the detectable level in our experimental system (data not shown). The putative repressor(s) is apparently short-lived and/or requires de novo protein synthesis (Fig. 8a), and is degraded at least in part by proteasomes (Fig. 8b), thus exhibiting properties suitable for a sensitive regulatory point in neuronal responses to the synaptic inputs. It is tempting to speculate that the putative repressor(s) ceases the Scg2 activation caused by in-phase glutamate and dopamine inputs via the Ca2+ pathway. Accumulation of the repressor protein to an effective level may require a lag time compared to the Ca2+-mediated immediate Scg2 transcriptional activation by protein modifications such as phosphorylation. Thus, repressor accumulation may determine the time window permissive to Scg2 activation in response to the in-phase glutamate and dopamine inputs. Without such a programmed repression mechanism, sporadic glutamate or dopamine inputs after the simultaneous inputs might cause unexpected Scg2 reactivation, thus disturbing the signal integration system. The identification and characterization of this repressor will help to uncover the unique regulatory mechanisms of Scg2.
Taken together, the results of this study led us to propose the working model illustrated in Fig. 9. In this model, glutamate inputs mobilize Ca2+ via NMDAR and/or other routes. Dopamine inputs also activate the same and/or distinct Ca2+ mobilization pathway(s). Thus, Ca2+ itself, or its downstream factor leading to the Ras/Raf/MEK/ERK pathway, constitutes a part of the signal integration system of glutamate and dopamine inputs directed to Scg2 activation. The involvement of cAMP in the mobilization of Ca2+ in response to dopamine remains to be confirmed. We propose that the dopamine-activated cAMP signaling diverges to the KT5720-sensitive kinase pathway leading to the accumulation of repressor protein(s), limiting the time window open to Scg2 activation by the in-phase glutamate and dopamine inputs. This programmed repression system prevents accidental Scg2 activation by sporadic glutamate or dopamine inputs, and increases the signal-to-noise ratio for specific activation by the simultaneous inputs. Thus, the regulatory system of Scg2 expression is equipped with machinery that is refined for the signal integration of in-phase synaptic inputs. Further studies are required to confirm this hypothetical mechanism, and to clarify its physiological roles in neuronal plasticity such as associative learning and memory.
Figure 9. Hypothetical mechanism for the regulation of Scg2 as a signal integrator of glutamate and dopamine inputs. Glutamate stimulation of NMDAR and/or other receptors causes Ca2+ influx to activate the MEK/ERK pathway, which further activates Scg2. Dopamine stimulation of D1R and/or other receptors cooperates with the glutamatergic signaling to activate Scg2 via the Ca2+ and MEK/ERK pathway, which thus serves as a part of the signal integration system of glutamate and dopamine inputs. Concurrently, activation of the KT5720-sensitive pathway that diverges from the dopaminergic Ca2+ pathway leads to the accumulation of the putative Scg2 repressor protein (X) that is otherwise short-lived because of rapid degradation by proteasomes. The resulting programmed repression system determines the time window permissive to cooperative activation of Scg2 by in-phase glutamate and dopamine inputs via the Ca2+ pathway.
Download figure to PowerPoint