Address correspondence and reprint requests to Leszek Kaczmarek, Nencki Institute, Laboratory for Molecular Neurobiology, 02–093 Warsaw, Pasteura 3, Poland. E-mail: email@example.com
The acronym ICER (inducible cAMP early repressor) refers to a group of four proteins produced from the CREM/ICER gene due to use of an internal promoter (P2) placed in an intron of the CREM (cAMP responsive element modulator) gene. The ICER proteins contain DNA binding/leucine zipper domains that make them endogenous inhibitors of transcription driven by CREB (cAMP responsive element binding protein) and its cognates, CREM and ATF-1 (activating transcription factor-1). ICER expression is inducible in the brain and in neuronal culture by a variety of stimuli. As a CREB antagonist, ICER appears to be of pivotal importance in neuronal plasticity and programmed cell death.
The discovery of inducible gene expression in the brain in response to physiological and pathological stimuli has provided a major impetus for development of molecular neurobiology (see Kaczmarek and Robertson 2002 for extensive reviews). Originally, nuclear protooncogenes, also called immediate early genes (IEGs), became the focus of multiple investigations. These genes are induced early, transiently and in a de novo protein synthesis-independent fashion in response to a variety of signals reaching the central nervous system. Such a mode of transcription activation requires pre-existing transcription factors to be made functional, usually by means of phosphorylation. CREB (cAMP responsive element binding protein) has become the most widely studied in this regard, although surely others exist which are of no less importance, for example, Elk-1. However, availability of research tools allowing alteration of CREB function, especially by means of mouse and fruit fly genetics, has led to spectacular results, originally suggesting a CREB role in memory formation (Dash et al. 1990; Bourtchuladze et al. 1994; Yin et al. 1994; see, however, Balschun et al. 2003) and also in neuronal survival/apoptosis pathways (Walton and Dragunow 2000; Lonze et al. 2002; Mantamadiotis et al. 2002).
However, to understand the whole picture of CREB complexity it is necessary to investigate thoroughly other members of CREB family, such as CREM (cAMP responsive element modulator) and ATF-1 (activating transcription factor-1). Furthermore, the importance of an endogenous antagonist of CREB, named ICER (inducible cAMP early repressor) should especially be appreciated. Recent results on signalling pathways leading to ICER activation and its expression patterns in the brain, and finally first functional data on a possible role of ICER in neuronal pathology have begun to reveal how important those molecules are in the nervous system. This review aims to show the most basic aspects of ICER in the brain. To fit into the format of a mini-review, we have purposely focused on the information derived from brain studies as well as neurones and/or neurone-like cells in culture.
The four different isoforms of ICER that are known to date to result from alternative splicing are named ICER I, ICER Iγ, ICER II and ICER IIγ (Fig. 1b). ICER I isoforms utilize DBD I, whereas ICER II use DBD II. While mRNA for ICER I contains sequences encoding both DBD I and II, insertion of the DBD II into the protein structure is blocked by a stop codon placed in the C-terminus of DBD I. Additionally, both ICER I and II may be deficient in exon γ (ICER Iγ and IIγ). ICER heterodimerizes with either CREM or CREB proteins, producing strong transcriptional silencers. By analogy to the effects of different DBDs utilization by CREMs, one can speculate that different ICER isoforms can form heterodimers more efficiently with particular isoforms of CREB or CREM (Laoide et al. 1993).
ICER gene and protein regulation
Whereas the major CREM promoter, P1, is rich in GC sequences that are typical for non-inducible, constitutively active promoters, the ICER P2 promoter has a normal AT and GC content and is strongly inducible by various stimuli (see below). P2 contains also two pairs of closely spaced CRE (cAMP regulatory elements) sequences organized in tandems (Molina et al. 1993) called cAMP autoregulatory elements (CARE1-4, see Fig. 1c). Similarly to typical CRE sequences, CAREs are recognized by a variety of CRE binding proteins including CREB, CREM and ICER (Molina et al. 1993; Mao et al. 1998).
Only the CARE3 sequence is identical with the consensus palindromic CRE. CARE4 differs by one base pair, while CARE1 and 2 are only 62% similar to the CRE consensus (Molina et al. 1993; Mao et al. 1998). Detailed gel shift analyses revealed that the different CAREs display differential binding affinities toward CREB, CREM and ICER proteins (Molina et al. 1993; Mao et al. 1998). Furthermore, Monaco and Sassone-Corsi (1997) showed that CARE3–4 were much more efficient than CARE1–2 in driving cAMP-independent transcription of reporter gene stimulated by nerve growth factor (NGF).
The best described signal transduction pathway regulating ICER transcription is the cAMP–protein kinase A (PKA)–CREB–ICER signalling. CREB, phosphorylated by PKA at Ser133, induces transcription from promoters containing CRE sequences (for review see Mayr and Montminy 2001), including ICERs P2 (Molina et al. 1993). After 2–6 h, ICER levels reach a maximum and ICER competes with CREB in binding CRE sequences, blocking transcription from CRE-containing promoters including ICERs own promoter (Molina et al. 1993). Such autoregulatory negative feedback loop serves as a mechanism that prepares cAMP-dependent signalling to be activated again. However, studies by Lamas and Sassone-Corsi (1997) suggest that the ability of the system to reset depends on duration of a stimulus. The cAMP-dependent ICER expression was mostly studied in cells stimulated with forskolin (adenyl cyclase activator), as well as membrane-permeable cAMP analogs, in many different cell types including neuroendocrine cells, and neurones (Molina et al. 1993; Stehle et al. 1993; Tinti et al. 1996; Monaco and Sassone-Corsi 1997; Thommesen et al. 2000).
Although ICER expression was primarily thought to be specifically induced through cAMP and PKA (Molina et al. 1993), recent findings suggest that several different signal transduction pathways may regulate ICER expression. First,Monaco and Sassone-Corsi (1997) demonstrated that ICER expression in PC12 cells might be activated by NGF in a cAMP-independent manner. Next, this finding of cAMP-independent induction was extended to ICER gene activation by epidermal growth factor (EGF) in PC12 and AR42J neuroendocrine-derived cell lines (Thommesen et al. 2000). Both EGF and NGF are known to cause CREB phosphorylation through tyrosine kinase receptors (TrkA, EGFR)-Ras-MAPKinase-RSK signalling cascade (Xing et al. 1996). Indeed, overexpression of dominant negative mutant of Ras in PC12 cells completely blocked NGF- but not cAMP-induced ICER expression (Monaco and Sassone-Corsi 1997). Interestingly, CREB phosphorylation appears to be necessary, but not always sufficient, for the activation of ICER expression (Monaco and Sassone-Corsi 1997). In PC12 cells, which were already differentiated to neurone-like cells, ICER inducibility by NGF was completely abolished. However, the treatment with NGF was still able to increase CREB phosphorylation. On the other hand, stimulation of cAMP-dependent pathway activated both CREB phosphorylation and ICER expression.
In some cell lines, such as PC12, AR42J, RIN5F, phorbol esters (PMA), which are activators of protein kinase C, were also potent inducers of ICER mRNA expression (Thommesen et al. 2000). Similarly, more physiological treatment of AR42J cells with agonists of CCK-A and B receptors (gastrin and CCK-8), that are known to activate both protein kinase C and MAPK kinases, led to an increase in ICER mRNA level. This increase was not blocked by H-89, the inhibitor of PKA (Thommesen et al. 2000). Recently, two additional signal transduction pathways that activate ICER expression were discovered. Mead et al. (2003) showed that ICER expression could be activated by interferon gamma in the J774.2 cell line in a caseine kinase II-dependent manner. Notably, it is known that caseine kinase II is capable to phosphorylate CREB at serines 142 and 143 but not at 133. For a long time, phosphorylation of these two residues of CREB was believed to inhibit CREB activity (Shaywitz and Greenberg 1999). Recently, however, data obtained by Gau et al. (2002) and Kornhauser et al. (2002) suggested that phosphorylation of Ser142 and 143 might act along with phosphorylation of Ser133 to allow CREB to reach maximal activity in neurones. Finally and most surprisingly, Krueger et al. (2000) showed that the activation of P2 promoter may occur independently form CREB by thyrotroph embryonic factor binding to extended CARE3 in a calcium- and CamKIV-dependent manner.
In addition to the regulation of transcription, the abundance of ICER protein in the cell is also regulated by the process of protein degradation. There is some variability in half-life between different ICER isoforms (Folco and Koren 1997). In GH3 cells treated with 8Br-cAMP, ICER I was shown to be the most stable, followed by ICER II. In contrast, ICER isoforms lacking γ domain (ICER Iγ and IIγ) were degraded rapidly after stimulation with a half-life of 3 h. The degradation of all isoforms was strongly inhibited by lactacystin, a selective inhibitor of proteasome-dependent degradation (Folco and Koren 1997). Moreover, coexpression of ICER IIγ and ubiquitin-conjugating enzymes, such as CDC34 or RAD6B resulted in an enhanced rate of degradation of transfected ICER protein that was prevented by inhibition of proteasome activity (Pati et al. 1999). Recently, phosphorylation of ICER at Ser41 by ERK1/2 MAPKinases was shown to target ICER protein for proteasome-dependent destruction (Yehia et al. 2001).
Notably, ICER was also shown to inhibit gene activity by binding to sequences that are recognized by other transcription factors (TFs) (e.g. NFAT/AP1). Additionally, as described in the previous sections, ICER expression may be regulated by signals different from cAMP, and TFs other than CREB. Thus, it seems to be plausible to suggest that ICER may be utilized as a switch between different signalling pathways. In this case, the role of ICER would be to inhibit cAMP- or CREB-dependent gene expression without involving the PKA signalling pathway or to restrain, in a cAMP-dependent manner, the expression of the genes that do not contain any CRE sequences.
ICER expression in the brain
Before presenting data on ICER expression in the brain, it is worth mentioning that because of its close sequence overlap with CREM transcripts and proteins, distinguishing between the CREM versus ICER gene products poses a formidable task. It can, however, be achieved, especially by means of RNase protection assay, RT-PCR, and probably less reliably by in situ hybridization with specially designed gene probes and even by immunocytochemistry and immunoblotting with specific antibodies, although these tools have almost never been used in parallel to assure the specificity of the results. On the other hand, it is widely accepted that the ICER gene products are the major inducible ones among the CREM gene products in the brain and thus inducible expression serves as a hallmark of ICER.
In the central nervous system, ICER is expressed at high level, predominantly in neuroendocrine structures such as the pineal and pituitary glands. Expression of ICER mRNA in the pineal gland gradually increases during the dark phase of the circadian rhythm, leading to a peak early in the second half of the dark period, and decreases to the basal level soon after the dark–light transition (Stehle et al. 1993; Maronde et al. 1999a,b; Pfeffer et al. 1999; Koch et al. 2003). Noradrenalin was found to be the neurotransmitter responsible for regulation of ICER gene expression (Stehle et al. 1993; Pfeffer et al. 1999). Functionally, this pattern of ICER expression has been linked to its antagonistic function against phosphorylated (i.e. active) form of CREB that accumulates at the beginning of the dark period and declines when the ICER raises. It has been suggested that the changing ratio between P-CREB and ICER shapes the in vivo dynamics of expression of arylalkamine-N-acetyltransferase, the rate-limiting enzyme of melatonin synthesis (Foulkes et al. 1996; Maronde et al. 1999a,b; Stehle et al. 2001).
Circadian control over the ICER mRNA expression is also observed in some other brain structures. In the pituitary, the ICER expression pattern is tightly linked to the day–light conditions in which the animals are kept. There are high increases in ICER mRNA at a few hours into the light phase if the animals are on a long day regime, whereas the amplitude of this peak is greatly attenuated under a short photoperiod (Messager et al. 1999, 2000; Johnston et al. 2003). Thus, it appears that ICER in the pituitary may play a role in animal responsiveness to changing day length. In addition, ICER expression in the pituitary gland is inducible by CRF and restrain stress (Molina et al. 1993; Mazzucchelli and Sassone-Corsi 1999). In contrast to the pituitary, circadian induction of ICER mRNA that is observed within suprachiasmatic nucleus a few hours into the light phase of the diurnal cycle is not markedly diminished under short photoperiod conditions (Messager et al. 1999, 2000). In addition, a small induction after light-on can be noted in the supraoptic nucleus and paraventricular nucleus of the hypothalamus, as well as in the central nucleus of amygdala (Spencer and Houpt 2001).
In most other neuronal and nonendocrine tissues tested, ICER mRNA appears to be at basal conditions in uniformly low amounts. However, Staiger et al. (2000) have noted expression of ICER proteins in all cortical layers and marked ICER presence was observed in the olfactory bulb and nucleus of the solitary tract (Stehle et al. 2001). In many brain structures, however, ICER mRNA was found to be inducible by a variety of physiological and pathological stimuli (Table 1). It is of great interest that in most cases the list of the stimuli is overlapping with treatments known to activate such IEGs as c-fos, junB, zif268, etc. For example, ICER/CREM mRNA was found to accumulate in the hippocampus (as well as some other brain regions, e.g. cerebral cortex) in response to seizures evoked by kainate (a glutamate agonist), and electroconvulsive shock (Nedivi et al. 1993; Rydelek-Fitzgerald et al. 1996; Konopka et al. 1998; Storvik et al. 2000). An original report by Nedivi et al. (1993) was based on a gene-screening strategy that aimed at identifying the kainate-responsive genes in the dentate gyrus of the hippocampus. Such specific localization of gene expression would strongly point to its possible involvement in neuronal plasticity as the dentate gyrus granule neurones undergo plastic changes following this treatment (see Zagulska-Szymczak et al. 2001). However, Konopka et al. (1998) have reported that CREM/ICER mRNA kainate-induction is observed in all hippocampal subfields, including CA1 and CA3 that are excitotoxically damaged in response to kainate (see Zagulska-Szymczak et al. 2001). Interestingly, the CREM/ICER mRNA accumulation was more sluggish in comparison with IEG mRNA response to kainate (Konopka et al. 1998; unpublished data).
Table 1. Induction of ICER expression in various brain regions
It thus appears that ICER expression is induced in the aforementioned conditions as a result of massive activation of glutamate receptors that is observed throughout the whole hippocampus. Notably, direct injection of NMDA, an agonist of a specific class of glutamate receptors, into the hippocampus also produced CREM/ICER mRNA accumulation (Konopka et al. 1998). On the other hand, also the noncompetitive NMDA receptor antagonists, such as MK-801, phencyclidine, ketamine, and memantine, were found to be capable of inducing ICER mRNA expression in various brain regions, most strongly in the cingulate and retrosplenial as well as the parietal and temporal cortex, the hippocampus, the thalamus, and the olfactory bulb (Konopka et al. 1998; Storvik et al. 2000). Both glutamate agonists and antagonists affect brain in a very complex manner, with the actions involving neuronal plasticity and apoptotic cell death. Besides the aforementioned intricacy of the kainate model, it is of note that high-dose treatment of MK-801 evokes neuronal loss in the retrosplenial cortex (Fix et al. 1993), and NMDA is also well known for its neurotoxicity.
ICER expression was also found to be markedly up-regulated by several treatments that apparently do not provoke cell death. Konopka et al. (1998) exposed dark-adapted rats to light and found increased ICER mRNA expression in the visual cortex. Again, as in the case of kainate treatment, ICER expression was delayed and prolonged in comparison of such IEGs as c-fos, junB, and zif268. Interestingly, the protein synthesis inhibitor, cycloheximide was also evoking dramatic increase in ICER mRNA levels (Konopka et al. 1998). Such a response is a hallmark of IEG. Sensory stimulation produced by exposing rats to a novel and enriched environment also led to enhanced ICER protein expression in the somatosensory (barrel) cortex, representing mysticial vibrissae, known to be heavily involved in exploring novel environments by rodents (Staiger et al. 2000). In addition, Luckman and Cox (1995) observed ICER gene activation in hypothalamic magnocellular neurones following osmotic stimulation, and Borsook et al. (1999) in the same brain structure after lipopolysaccharide treatment. Notably, ICER mRNA expression observed by Luckman and Cox (1995) was delayed in comparison to c-fos mRNA accumulation. Finally, Spencer and Houpt (2001) reported that LiCl evokes ICER mRNA expression in central amygdala, paraventricular nucleus of the hypothalamus, and supraoptic nucleus. LiCl is routinely used as an unconditioned stimulus in a conditioned taste aversion behavioural training paradigm (Welzl et al. 2001).
Functional role(s) of ICER in the brain
As has been mentioned before, it is interesting to note that all the aforementioned conditions have the IEG activation in common. Furthermore, ICER expression appears to follow the peak of IEGs' mRNA accumulation and this fact, together with ICER ability to antagonize CREB function, as well as with proven role of CREB in activating many IEG implies that the role of ICER is to switch off the IEG expression. However, this hypothesis has apparently been disproved by the study of Spencer and Houpt (2001) who did not find any significant attenuation of c-fos (an IEG) response to LiCl injection at the time when ICER levels were high because of the previous LiCl treatment.
The results of the expression studies point to possible role of ICER in either neuronal plasticity (e.g. sensory stimulation, LiCl injection) or neurodegeneration (kainate and NMDA injections, as well as application of high doses of NMDA receptor antagonists). Although no functional evidence for a role of ICER in neuronal plasticity has been reported as yet, direct support for ICER functional role in neuronal cell death has recently been provided by Jaworski et al. (2003). These authors have shown that endogenous ICER expression is elevated in hippocampal and cortical neuronal cultures when they undergo apoptosis, and, furthermore, the overexpression of ICER delivered in an adenoviral vector also results in apoptosis. The exogenous ICER was found to compete with CREB for CRE DNA sequence, to inhibit the CRE-reporter gene, and to down-regulate bcl-2 gene, known to be up-regulated by CREB. These findings suggest that ICER may play a role in neuronal cell death by antagonizing prosurvival functions of CREB.
Although still limited, the data gathered so far on ICER in neurones and in the brain strongly support the notion that this gene product plays a major role in physiological and pathological responses to external environment. It is expectable that it will be difficult, if not impossible, to appreciate the CREB role in the brain without understanding its alter ego – ICER.
The greatly appreciate comments on the manuscript by Drs R. K. Filipkowski, M. Rylski and J. Platenik. This work has been supported by the Polish grant KBN PBZ-MIN-001/P05/12