10 November 2006. Two recent studies are changing traditional ideas about transcriptional regulation. In the November 3 Cell, researchers led by Ricardo Dolmetsch at Stanford University, California, report that a voltage-gated calcium channel fragment can directly act as a transcriptional regulator in the nucleus. The finding raises new questions about the role of these channels, which are known to activate CREB, BDNF, and other factors that are linked to neurodegenerative diseases, including Alzheimer disease (AD). The second study, led by Rajvir Dahiya at the University of California at San Francisco, finds a new role for small, double-stranded RNAs (dsRNAs). Infamous for their ability to induce gene silencing through RNA interference (RNAi), 21-nucleotide dsRNAs can also activate expression, report Dahiya and colleagues in the October 30 PNAS online. According to the authors, “RNA activation,” or RNAa, may prove to have therapeutic benefit because it could, in theory, activate genes that have been improperly silenced, much like RNAi is now being aggressively pursued as a means of shutting off gene expression.
A new mechanism for calcium signaling in neurons
Voltage-gated calcium channels are a critical part of the calcium signaling system in neurons. By ushering extracellular calcium into the cell in response to depolarization, the channels elicit changes in gene expression, facilitating a variety of adaptive responses, including neuronal plasticity. L-type channels are particularly effective at inducing transcriptional changes (Bading et al., 1993), a property that has been attributed to two mechanisms. First, through a calcium “wave” that propagates through the cell to the nucleus, the channels can activate nuclear calcium-dependent transcription factors such as CREB (Hardingham et al., 2001). Second, L-type channels can locally activate signaling proteins that then transduce the signal to the nucleus (Deisseroth et al., 1998; Dolmetsch et al., 2001). L-type channels have also been implicated in AD pathology. There is evidence that in response to amyloid-β toxicity, voltage-gated calcium channels allow an influx of calcium that leads to apoptosis, or cell death (see Ueda et al., 1997). In fact, L-type channel blockers are currently in clinical trials for AD.
But in this study, Dolmetsch and colleagues challenge convention by showing that a C-terminal fragment of the L-type voltage-gated channel CaV1.2 can act directly as a transcriptional regulator. Previous work in other laboratories found that this fragment is produced normally in vivo by proteolytic cleavage (De Jongh et al., 1994; Gerhardstein et al., 2000). In fact, just recently researchers at the University of Washington in Seattle discovered that the fragment can bind to and inhibit truncated calcium channels (Hulme et al., 2006), suggesting that the fragment fulfills a regulatory function.
To further investigate this auto-inhibitory process, first author Natalia Gomez-Ospina and colleagues generated an antibody to the C-terminal of the channel. Surprisingly, when they used the antibody to localize the fragment in subcellular fractions of rat cortex, they found that the peptide turned up predominantly in the nucleus. Subsequently, the authors found that the fragment, which they dub the calcium channel associated transcriptional regulator, or CCAT, is normally found in the nucleus of unstimulated neurons; that depolarizing conditions cause CCAT to be transported out of the nucleus; and that CCAT regulates expression of endogenous genes. In addition, they found that expressing CCAT in cerebellar granule neurons in culture resulted in a doubling of the average length of neurites.
Calcium channel associated transcriptional regulator (CCAT)
Following the subcellular fractionation study, the researchers used the antibody to localize CCAT in rat brain sections. They found that the fragment is strongly nuclear in a subset of neurons in the thalamus, inferior colliculus, inferior olivary nucleus and olfactory bulb, as well as in a small number of neurons in the cortex and hippocampus. CCAT costained with glutamic acid decarboxylase, indicating that it exists mostly in inhibitory GABAergic neurons.
Because antibodies can cross-react with proteins that contain identical epitopes, the researchers confirmed the nuclear localization of CCAT using a second approach—they made CaV1.2 fusions with yellow fluorescent protein (YFP) and expressed them in neurons, myocytes, and HEK293T cells. When they fused YFP to the C-terminal end of the calcium channel, they found fluorescence in the cytoplasm and nucleus, but when the probe was fused to the N-terminal end of the channel, yellow fluorescence only appeared at the cell membrane. A fusion with only the last 503 amino acids also turned up in the nucleus as punctate staining. A similar pattern emerged when the authors used confocal imaging to localize endogenous CCAT. Significantly, this punctate pattern was enhanced when the cells were grown in low calcium, suggesting that it may be driven by cell signaling events.
To test this idea, Gomez-Ospina and colleagues challenged cortical neurons with a variety of calcium mobilizing agents. When the extracellular calcium concentration was decreased by chelation, thereby decreasing the intracellular calcium concentration, a strong increase in nuclear CCAT fluorescence was seen. Conversely, KCl, to mimic tonic depolarization, or glutamate caused significant decreases in the nuclear CCAT fluorescence. This loss of CCAT was not due to protein degradation, leading Gomez and colleagues to conclude that depolarizing conditions drive CCAT export from the nucleus.
What is the significance of CCAT accumulation in, and export from, the nucleus? Because the responses to changes in calcium suggest that CCAT is indeed involved in cell signaling, the authors looked for nuclear binding partners that might mediate such effects. Using immunoprecipitation and mass spectrometry they identified p54(nrb)/NonO as a CCAT binding partner in Neuro2A cells. Since p54(nrb)/NonO is a nuclear protein that regulates a variety of signaling pathways, the authors hypothesized that CCAT might regulate transcription and so used oligonucleotide microarrays to determine if any messenger RNAs (mRNAs) are transcriptionally regulated by CCAT overexpression. Cells which expressed full-length CCAT were compared with those that expressed a dysfunctional CCAT fragment. The combined results revealed that 16 mRNAs were significantly upregulated and 31 were significantly downregulated by intact CCAT. Seven of these mRNAs were confirmed by RT-PCR. Those genes that appeared upregulated include the gap junction protein connexin 31.1, the axon guidance factor Netrin4, and RGS5, a regulator of G-protein signaling. Downregulated genes included many ion channels, such as the sodium-calcium exchanger, the cation channel TRPV4, and the 2D subunit of the NMDA receptor. “These results suggest that CCAT can both increase and decrease the expression of a wide set of genes that regulate neuronal differentiation, connectivity, and function,” write the authors. Interestingly, the NMDA receptor has also been linked to Aβ mediated toxicity (see ARF related news story).
Lastly, Gomez-Ospina and colleagues found that expressing CCAT in cerebellar granule neurons in culture resulted in a doubling of the average length of neurite growth, as well as in a small but statistically significant increase in the number of neurites. CCAT expression did not appear to affect granule cell survival.
“We propose that CCAT both regulates transcription and reduces calcium influx through CaV1.2,” the authors write. “This hypothesis is appealing in light of the observation
that CCAT is exported from the nucleus by elevations in intracellular calcium, suggesting that under conditions of tonically elevated calcium, CCAT would both alter the transcription of specific genes and inhibit the activity of CaV1.2. Thus CCAT may be an important part of a negative-feedback pathway regulating both gene expression
and calcium influx in the neurons,” they conclude.
Small, double-stranded RNAs (dsRNAs) induce gene activation
In the second study, Dahiya and colleagues report that small, double-stranded RNAs (dsRNAs) can induce activation of genes, a striking contrast to the well-known paradigm in which dsRNAs silence genes through RNA interference (RNAi). First author Long-Chen Li and colleagues designed and synthesized 21-nt dsRNAs which were meant to target selected promoter regions of human E-cadherin, p21WAF1/CIP1, or vascular endothelial growth factor (VEGF) genes. In each case, expression of the targeted genes was enhanced between two- and 10-fold. Interestingly, the effects of the RNA activation (RNAa) were detectable for nearly 2 weeks after the experiment was performed, suggesting that RNAa may be more long-lived than RNAi, which typically fades after less than a week.
In mechanistic experiments, the researchers determined that the 5’ end of the antisense strand of the dsRNA must be strictly homologous to the targeted gene for RNAa to work, but mismatches to the 3’ end are tolerated. In addition, the process of RNAa requires the Argonaute 2 (Ago2) protein, a member of the RNAi pathway, suggesting perhaps that RNAa and RNAi work in a similar fashion.
What, then, determines which process is triggered by a dsRNA? Li and colleagues admit that it is not entirely clear, but there do seem to be subtle mechanistic differences. RNAa tolerates mismatches between the dsRNA and the target sequence, for example, suggesting that Ago2 acts different from its role in RNAi, where it cleaves matched mRNAs. So why the need for Ago2 in RNAa, then? The authors suggest that, as in RNAi, Ago2 might be needed to process the dsRNA into an active form by cleaving and discarding the “passenger” strand.
Li and colleagues note previous work that showed that dsRNAs can silence transcription. This, of course, is very different from the translational silencing brought about by RNAi, and is most likely a function of dsRNAs with extremely high GC-rich sequences that bind to CG islands within promoters and thereby recruit histones and/or methylases that silence the gene (see Morris et al., 2004). Although Li and colleagues avoided CG-rich targets by closely following rational siRNA design rules, including the use of low GC content, DNA modification may play a role in RNAa. They found that loss of histone 3 methylation is associated with activation, for example. Such epigenetic changes may help to explain the long-lived effects of RNAa and also raise the intriguing possibility that gene activation could propagate through cell divisions.
Interestingly, this is not the first time that a small dsRNA has been shown to turn genes on. Work from Fred Gage’s lab at The Salk Institute revealed a small dsRNA that activates the neuron restrictive silencer element (NRSE/RE1) in neuronal stem cells. That dsRNA induces expression of neuron-specific genes (Kuwabara et al., 2004) that help determine cell fate, but it appears to work through a dsRNA-protein interaction, rather than by modifying DNA.
RNAa appears entirely different. “Although the exact mechanism is unknown at present, the identification of such a phenomenon may still have significant therapeutic potential,”
write the authors. This may be particularly true for cancers in which proapoptotic or “normalizing” genes have been improperly silenced.—Jillian Lokere and Tom Fagan.
Gomez-Ospina N, Tsuruta F, Barreto-Chang O, Hu L, Dolmetsch R. The C terminus of the L-type voltage-gated calcium channel ca(v)1.2 encodes a transcription factor. Cell. 2006;127:591-606. Abstract
Li LC, Okino ST, Zhao H, Pookot D, Place RF, Urakami S, Enokida H, Dahiya R. Small dsRNAs induce transcriptional activation in human cells. PNAS published November 3, 2006, 10.1073/pnas.0607015103. Abstract