Calcium Regulation of Gene Expression: Isn't That Spatial?

Calcium may enter a neuron via several portals including voltage-sensitive calcium channels (VSCC), such as the L-type calcium channel, and ionotropic neurotransmitter receptors, such as the NMDA subtype of glutamate receptors. Alternatively, calcium may be released from its internal storage depot, the endoplasmic reticulum. Interestingly, depending on its route of entry into the neuron, calcium may influence distinct transcriptional responses in neurons. For example, calcium influx via VSCCs is more effective than calcium influx through NMDA receptors in triggering activation of expression of the BDNF gene in cortical neurons (reviewed inGhosh and Greenberg 1995xGhosh, A. and Greenberg, M.E. Science. 1995; 268: 239–247CrossRef | PubMedSee all ReferencesGhosh and Greenberg 1995). Moreover, calcium influx via VSCCs is more effective than calcium influx via NMDA receptors in activating CRE-mediated transcription of c-fos in hippocampal neurons (Bading et al. 1993xBading, H., Ginty, D.D., and Greenberg, M.E. Science. 1993; 260: 181–186CrossRef | PubMedSee all ReferencesBading et al. 1993). Yet calcium entry via either VSCCs or NMDA receptors is sufficient to trigger SRE-mediated transcription of c-fos. How can the same second messenger ion differentially regulate CRE-mediated and SRE-mediated transcription within the same neuron? Are the levels of calcium attained different depending on its mode of entry, or does the route of calcium entry influence distinct biochemical processes?One potential explanation is that the CRE and the SRE can discriminate between spatially distinct pools of calcium. This possibility was addressed recently in experiments in which calcium within the nucleus of AtT20 cells, a pituitary cell line, was buffered to prevent its accumulation following activation of VSCCs (Hardingham et al. 1997xHardingham, G.E., Chawla, S., Johnson, C.M., and Bading, H. Nature. 1997; 385: 260–265CrossRef | PubMed | Scopus (558)See all ReferencesHardingham et al. 1997). The buffer, BAPTA coupled to a 70kDa dextran molecule, was microinjected into the nucleus, and it did not leak from the nucleus to the cytoplasm. Therefore, the buffer largely prevented accumulation of nuclear calcium ions, but it did not affect accumulation of cytoplasmic calcium following activation of VSCCs. Under these conditions, SRE-mediated gene expression was normal following calcium entry via VSCCs. In dramatic contrast, CREB-mediated transcription was inhibited. Therefore, an increase in cytoplasmic calcium is not sufficient to promote CREB-mediated transcription; an increase in nuclear calcium is critical for CREB-mediated gene expression. Whether an increase in cytoplasmic calcium is sufficient to promote TCF-dependent transcription, TCF-independent transcription, or both remains to be determined. Since calcium-sensitive, TCF-dependent transcription is mediated by a member of the MAPK family of protein kinases, and since MAPK can translocate from the cytosol to the nucleus, it is reasonable to suspect that an increase in calcium levels exclusively within the cytoplasm may be sufficient to trigger MAPK-mediated, TCF-dependent transcription. Thus, spatially distinct pools of calcium do influence gene expression in unique ways.An increase in nuclear calcium concentration is critical for CREB-mediated transcription, and since phosphorylation of CREB on Ser133 influences CREB-mediated transcription, one may predict that a rise in nuclear calcium levels is also critical for phosphorylation of CREB Ser133. However, in hippocampal neurons, an increase in calcium concentration within the cytoplasm at the synapse, but not within the nucleus, appears to be necessary to promote phosphorylation of CREB on Ser133 (Deisseroth et al. 1996xDeisseroth, K., Bito, H., and Tsien, R.W. Neuron. 1996; 16: 89–101Abstract | Full Text | Full Text PDF | PubMed | Scopus (492)See all ReferencesDeisseroth et al. 1996). At first glance, these two observations may seem contradictory. One might immediately dismiss the significance of this apparent discrepancy by suggesting that calcium signaling mechanisms are profoundly different between these two cell types. For example, AtT20 cells are non-neuronal cells that do not express NMDA receptors, and calcium entry via NMDA receptors into hippocampal neurons is necessary for activity-dependent phosphorylation of CREB on Ser133. However, other explanations exist. For example, phosphorylation of CREB Ser133, while critical for CREB-mediated gene expression, may not be sufficient to promote IEG transcription. This idea is best exemplified by the observation that many extracellular stimuli induce robust phosphorylation of CREB on Ser133, yet only certain of these stimuli can induce CREB-mediated transcription in the absence of other transactivators. Thus, other critical regulatory event(s) are likely to accompany phosphorylation of CREB Ser133 during calcium activation of CREB-mediated transcription. Other events may include additional modifications of CREB, modifications of CBP or P/CAF, or associations between CBP and other regulatory proteins (e.g.,Nakajima et al. 1996xNakajima, T., Fukamizu, A., Takahashi, J., Gage, F.H., Fisher, T., Blenis, J., and Montminy, M.R. Cell. 1996; 86: 465–474Abstract | Full Text | Full Text PDF | PubMed | Scopus (219)See all ReferencesNakajima et al. 1996). Therefore, an increase in cytoplasmic calcium concentration in hippocampal neurons may be necessary or perhaps even sufficient to trigger phosphorylation of CREB Ser133, but it may not be sufficient to promote CREB-mediated transcription. A current major challenge is to identify and characterize regulatory events, in addition to phosphorylation of Ser133, that influence CREB-mediated IEG expression.One exciting prospect of the recent work on calcium signaling to the nucleus is that spatially distinct calcium signals may influence the activities of transcription factor complexes in unique ways. Thus, synaptic input that influences nuclear and cytoplasmic calcium levels differently may regulate distinct complex programs of gene expression, and therefore, distinct long-term neuronal changes. In other words, two excitatory synapses that differ only in their proximity to the cell body of the post-synaptic neuron may have profoundly different influence over the nature of the nuclear response. For example, a synapse in a distal dendritic process may have strong influence over SRE-mediated gene expression, and one in which the synapse is more proximal to the cell body may have control over SRE- as well as CRE-containing genes (Figure 2Figure 2). Since some IEGs contain multiple SRE-like elements, they may be more sensitive to changes in cytoplasmic levels of calcium. Other IEGs, including somatostatin, proenkephalin and other neuropeptide genes, contain one or more CREs, but no SRE. As such, expression of these genes may be influenced by synaptic activity that affects nuclear calcium levels, but not by synaptic activity that increases calcium levels only within distal processes. This idea may explain the observation that LTP-inducing stimuli more efficiently triggers activation of Zif/268, which contains four putative SREs and only one or two CREs, than c-fos, which contains only one SRE and three functional CREs (Cole et al. 1989xCole, A.J., Saffen, D.W., Baraban, J.M., and Worley, P.F. Nature. 1989; 340: 474–475CrossRef | PubMedSee all ReferencesCole et al. 1989). Perhaps LTP-inducing stimuli more efficiently increase calcium concentration within distal spines and cytoplasm than in the nucleus. Thus, while much emphasis has been placed on CREB and its role in activity-dependent gene transcription, it is likely that SRF and p62TCF family members are also critical for activity- dependent expression of IEGs in neurons, and for mediating activity-dependent long-term neuronal adaptive responses. Future experiments should clarify the mechanisms by which the calcium signal is propagated from the synapse to the nucleus to trigger TCF-dependent and TCF-independent transcription, and the roles of SRF, Elk-1, Sap-1a and other p62TCFs in activity-dependent neuronal plasticity.Figure 2Transactivating Factors Bound to the c-fos SRE and CRE May Respond to Distinct Pools of CalciumView Large Image | View Hi-Res Image | Download PowerPoint SlideThus, it is possible that expression of entire complex genetic programs that control neuronal adaptive responses may be influenced by spatial aspects of calcium signaling, and the nature of excitatory synaptic input. While this idea remains to be experimentally verified, it seems clear that IEGs have the capacity to distinguish between different forms of calcium signaling, at least in part, because they can respond to spatially distinct pools of calcium.