- Top of page
- Materials and methods
KCNIP3/KChIP3 (voltage-dependent K+ channel interacting protein 3), alias Calsenilin and downstream regulatory element antagonist modulator (DREAM), is a multifunctional protein that modulates A-type potassium channels, affects processing of amyloid precursor protein and regulates transcription. KCNIP3 has been described to negatively influence the activity of CREB (cAMP/Ca2+-response element binding protein), an essential factor in neuronal activity-dependent gene expression regulation. However, reports on intracellular localization of KCNIP3 in neurons are diverse and necessitate additional analyses of distribution of KCNIPs in cells to clarify the potential of KCNIP3 to fulfill its functions in different cell compartments. Here, we examined localization of the entire family of highly similar KCNIP proteins in neuronal cells and show that over-expressed isoforms of KCNIP1/KChIP1, KCNIP2/KChIP2, KCNIP3/KChIP3, and KCNIP4/KChIP4 had varied, yet partially overlapping subcellular localization. In addition, although some of the over-expressed KCNIP isoforms localized to the nucleus, endogenous KCNIPs were not detected in nuclei of rat primary cortical neurons. Moreover, we analyzed the role of KCNIP proteins in cAMP/Ca2+-response element (CRE)-dependent transcription by luciferase reporter assay and electrophoretic mobility shift assay and report that our results do not support the role for KCNIPs, including DREAM/Calsenilin/KChIP3, in modulation of CREB-mediated transcription in neurons.
Neuronal calcium sensor (NCS) proteins are calcium binding proteins that have numerous distinct functions in the brain (Burgoyne and Weiss 2001; Burgoyne 2007). One of the NCSs, named either Calsenilin, downstream regulatory element antagonist modulator (DREAM) or Kv4 (voltage-dependent K+ channel 4) interacting protein 3 (KChIP3) depending on the role it was independently described to have, is outstanding among others in that it itself performs discrete functions in different cell compartments. Calsenilin was first reported to interact with presenilins and affect the processing of the amyloid precursor protein (Buxbaum et al. 1998). Then, DREAM was demonstrated to act as a transcriptional repressor (Carrion et al. 1999) and KChIP3 was shown to be a direct modulator of A-type potassium channels (An et al. 2000). The three proteins are identical in fact, and are produced from the same gene officially designated potassium channel interacting protein 3 (KCNIP3), which belongs to a family of four KCNIP genes along with KCNIP1, KCNIP2, and KCNIP4 (Pruunsild and Timmusk 2005).
As a result of different roles described for KCNIP3, it functions either in the cytoplasm in the vicinity of cell membranes or in the nucleus. Intriguingly, however, data on KCNIP3 localization in cells are controversial. Studies that have been centered on KCNIP3 as Calsenilin have shown that KCNIP3 functions as a cytoplasmic protein and is predominantly localized to the perinuclear region in the cytoplasm where it is associated with intracellular membranes (Lilliehook et al. 2002; Zaidi et al. 2002, 2004; Woo et al. 2008; Jang et al. 2011). However, nuclear localization of KCNIP3 has been detected when KCNIP3 has been stably or transiently over-expressed in H4 neuroglioma cells or in HeLa cells, respectively, and intracellular calcium levels have been experimentally elevated (Zaidi et al. 2004; Woo et al. 2008). Research on KCNIP3 and its paralogs KCNIP1, KCNIP2, and KCNIP4 as Kv4 interacting proteins has demonstrated that KCNIPs associate with Kv4 α subunits, are integral components of the endogenous A-type Kv channel complexes and as auxiliary subunits of these channels localize close to intracellular membranes and the cell membrane (Shibata et al. 2003; Misonou and Trimmer 2004; Rhodes et al. 2004). It has been determined that targeting of KCNIP2 and KCNIP3 to membranes is controlled by palmitoylation (Takimoto et al. 2002) and that KCNIP1 is localized to the post-endoplasmatic reticulum (post-ER) transport vesicles by myristoylation (O’Callaghan et al. 2003), making nuclear functions for these proteins unlikely. Yet, when over-expressed in COS1 or COS7 cells without the expression of Kv4 channels, diffuse distribution all over the cell including the nucleus has been detected for KCNIPs (Shibata et al. 2003; Han et al. 2006).
KCNIP3 as the transcriptional repressor DREAM is unique among other calcium-dependent transcription factors (TFs) because it binds calcium and has been demonstrated to be primarily regulated by calcium binding as such and not by signaling pathways that include phosphorylation and/or dephosphorylation (Carrion et al. 1999). In vitro experiments have established that in calcium free state the C-terminal part of KCNIP3 binds to a DNA sequence termed downstream regulatory element (DRE) (Ledo et al. 2000b; Osawa et al. 2001), and along with KCNIP3, other KCNIPs are capable of binding DRE in vitro and repress transcription in a DRE-dependent manner (Link et al. 2004). Although initially described as a repressor, KCNIP3 has been found to function also as an activator of transcription (Scsucova et al. 2005) and its role in gene regulation has been shown in various tissues and cell types including hematopoietic progenitor cells (Sanz et al. 2001), pineal gland, and retina (Link et al. 2004), cerebellar granule neurons (Gomez-Villafuertes et al. 2005), thyroid carcinoma cells (Matsuda et al. 2006), and cortical progenitors (Cebolla et al. 2008). To act as a transcriptional regulator, nuclear localization is a prerequisite and accordingly KCNIP3 has been detected to bind the arylalkylamine N-acetyltransferase promoter (Link et al. 2004), the cyclin-dependent kinase inhibitor 1A promoter (Scsucova et al. 2005), and the GFAP (glial fibrillary acidic protein) promoter (Cebolla et al. 2008) for example, by chromatin immunoprecipitation (ChIP). Furthermore, KCNIP3 has been found to interact with the CREB (cAMP/Ca2+-response element binding protein) family of TFs. Firstly, interaction of KCNIP3 with αCREM (cAMP/Ca2+-response element modulator α isoform) has been shown to mediate derepression of DRE-dependent regulation when the proteins are over-expressed in HEK293 cells (Ledo et al. 2000a). And secondly, it has been demonstrated that when KCNIP3 is not binding calcium it interacts with CREB1 (also known as CREB) and impairs recruitment of CBP (CREB binding protein) to CREB1 and impedes cAMP/Ca2+-response element (CRE)-dependent transcription in PC12 cells (Ledo et al. 2002). Provided that the CREB family members have crucial developmental and functional roles in the nervous sytem (Lonze and Ginty 2002), that data on intracellular localization of KCNIP3 are contradictory and that the CREB target gene BDNF (brain-derived neurotrophic factor) is not regulated by KCNIPs in primary cortical neurons (Pruunsild et al. 2011), it is essential to determine whether KCNIP3 or other KCNIP proteins are modulating the transcriptional activity of CREB family proteins in neurons. For that we analyzed the localization of different isoforms of the KCNIPs in neuronal cells and tested their potential to interfere with CRE-dependent transcription in rat primary cortical neurons and PC12 cells.
- Top of page
- Materials and methods
In this study, we demonstrated that the KCNIP proteins localize differently when over-expressed in primary cortical neurons. In addition, our results showed no KCNIP-like immunoreactive signal for endogenous proteins in the nuclei of rat primary cortical neurons and provided evidence that the KCNIPs, including KCNIP3-Ia alias DREAM, are not involved in CRE-dependent transcription in primary cortical neurons.
We observed that all KCNIP1 isoforms and the KCNIP4-IbΔII isoform have similar punctate cytoplasmic localization in neuronal cells. The punctate most likely correspond to post-ER transport vesicles, where KCNIP1 has been described to be localized by N-myristoylation (O’Callaghan et al. 2003). Although KCNIP1-Ib and KCNIP1-IbΔII contain the consensus N-myristoylation motif (O’Callaghan et al. 2003), KCNIP1-Ia and KCNIP4-IbΔII isoforms do not and the molecular mechanism of their subcellular localization has yet to be determined. However, it is noteworthy that the N-termini of KCNIP1-Ia and KCNIP4-IbΔII isoforms are the only homologous N-termini among the KCNIP proteins (Pruunsild and Timmusk 2005) and that a putative S-palmitoylation motif resides in both of them according to analysis with the CCS-Palm software (Zhou et al. 2006). Besides that, the N-termini of KCNIP1-Ia and KCNIP4-IbΔII contain lysine and arginine residues that might constitute the canonical class 2 nuclear localization signal (Pruunsild and Timmusk 2005). However, the subcellular localization of these isoforms makes the functionality of the predicted nuclear localization signal motifs doubtful and rather suggests functional S-palmitoylation. We found that all KCNIP2 isoforms, KCNIP3-Ia, and KCNIP4-IaΔII are localized to the nuclei of neuronal cells in addition to diffuse distribution in the cytoplasm. Considering that KCNIP2-Ia, KCNIP2-IaΔIIb, and KCNIP3-Ia contain S-palmitoylation motifs, which have been demonstrated to be functional (Takimoto et al. 2002), and that in the N-terminus of KCNIP2-IbΔIIab a motif for S-palmitoylation can also be predicted, this localization pattern is unanticipated. Diffuse localization of the KCNIPs in the cytoplasm, or both in the cytoplasm and the nucleus, when over-expressed in cells has been demonstrated before, however, and the common observation has been that localization of KCNIPs changes to membranous compartments when proteins interacting with them, for example the Kv4 channels or presenilins, are co-expressed in cells (Buxbaum et al. 1998; An et al. 2000; Shibata et al. 2003; Han et al. 2006; Quintero et al. 2008). Interestingly, we found here that in rat neural progenitor cell line HiB5, nuclear localization of over-expressed KCNIP2 isoforms, KCNIP3-Ia, and KCNIP4-IaΔII was more pronounced than in primary neurons. One of the reasons might be that as HiB5 cells are immature neuron-like cells, they express less Kv4 channels than primary neurons and therefore, the KCNIPs together with their associating proteins distribute less to the membranes and more KCNIP protein is available for nuclear import.
Our immunocytochemistry results revealed also that one of the KCNIP4 isoforms, namely KCNIP4-Ia, has a combination of diffuse and punctate pattern in neuronal cells. Punctate localization resembling that of KCNIP1 isoforms could be explained by S-palmitoylation of the N-terminus of KCNIP4-Ia (Takimoto et al. 2002). Moreover, reversible nature of S-palmitoylation might render the KCNIP4-Ia isoform less associated with membranes than irreversible N-myrsitoylation that causes the clear punctate pattern for N-myrsitoylated KCNIP1 isoforms that has been detected previously also by Hasdemir et al. in cultured hippocampal neurons (Hasdemir et al. 2005). Three of the KCNIP isoforms, KCNIP3-Ib, KCNIP4-IdΔII, and KCNIP4-IeΔII, had almost exclusive perinuclear localization in HiB5 cells and primary neurons. We analyzed the N-termini of these isoforms using the eucaryotic linear motif resource for functional sites in proteins (Puntervoll et al. 2003) and found that the N-termini of KCNIP3-Ib and KCNIP4-IeΔII contain a transmembrane helix and in addition, the N-termini of KCNIP3-Ib and KCNIP4-IdΔII contain a signal peptide sequence that mediates targeting of nascent secretory or membrane proteins to the ER, explaining the localization detected for these KCNIP isoforms. It should be added that in the N-termini of KCNIP1-Ia, KCNIP3-Ib, and KCNIP4-Ie, which all showed clear cytoplasmic distribution, and in the constitutive part of KCNIP4 encoded by exon III of KCNIP4, a leucine-rich nuclear export signal sequence can be predicted. However, it remains to be explored if these nuclear export signal motifs are functional. Thus, taken together, in each of the N-termini of KCNIP isoforms, except in that of KCNIP2-IaΔIIab and KCNIP4-IaΔII, a well recognizable signal for localization to membranous compartments in cells can be identified. This is compatible with previous studies showing that in vivo in the central nervous system neurons KCNIP2, KCNIP3, and KCNIP4 localize to neuronal processes (Hammond et al. 2003; Rhodes et al. 2004) where they associate with membranes (Zaidi et al. 2006; Zhang et al. 2007; Duncan et al. 2009) and co-localize with the Kv4 channels (An et al. 2000; Rhodes et al. 2004), and KCNIP1 localizes to the perinuclear cytoplasm and dendrites where it also co-localizes with the Kv4 channels (An et al. 2000; Rhodes et al. 2004; Strassle et al. 2005).
How do the KCNIPs, particularly KCNIP3-Ia/DREAM, regulate transcription when they are associated with membranes? An answer might be that certain signals facilitate movement of KCNIPs to the nucleus. Indeed, it has been demonstrated that KCNIP3-Ia translocates to the nucleus when calcium levels are elevated in H4 neuroglioma cells (Zaidi et al. 2004, 2006) and rat neonatal cardiomyocytes (Ronkainen et al. 2011), and that peripheral inflammation or acute pain induces KCNIP3 translocation to the nuclei of cells in the spinal cord (Zhang et al. 2007; Long et al. 2011). On the other hand, in primary neurons we did not detect redistribution of KCNIP isoforms, either over-expressed or endogenous, upon induction of calcium signaling. Interestingly, in contrast with neurons, KCNIP3-Ia has been detected in nuclei in glial cells where it is translocated out of the nuclei upon activation of glutamate receptors (Edling et al. 2007; Chavira-Suarez et al. 2008). Furthermore, KCNIP3-Ia has been shown to bind the GFAP promoter and activate GFAP gene expression during astrocyte differentiation in cortical progenitor cells (Cebolla et al. 2008). It could be that the function and localization of KCNIP3-Ia depends stringently on the cell type and that in glial cells KCNIP3-Ia acts as a transcriptional regulator. Notably, in studies where nuclear localization of KCNIP3-Ia has been detected in vivo, brain or spinal cord tissue has been analyzed by nuclear fractionation followed by western blotting (Zhang et al. 2007; Alexander et al. 2009; Long et al. 2011), which does not discriminate between neurons and glial cells. Glial expression of KCNIP3-Ia should also be taken into account when results of EMSA or ChIP assays with nuclear extracts of in vivo brain samples are analyzed. Intriguingly, although all in vivo data on KCNIP localization in neurons indicate that cytoplasmic distribution of KCNIPs predominates, as discussed before, a recent study has shown that sumoylation regulates nuclear localization of KCNIP3 and that endogenous KCNIP3 localizes to the nucleus in rat trigeminal ganglia neurons (Palczewska et al. 2011). Whether sumoylation-mediated KCNIP3 localization to the nucleus is an exception or is valid for other KCNIP isoforms and in other neuron types in addition to trigeminal ganglia neurons, remains to be explored. However, our results on endogenous KCNIP localization in primary cortical neurons demonstrate no nuclear localization of KCNIP1 and KCNIP4. Whereas the anti-KCNIP1 abs used here have been verified (Rhodes et al. 2004), we admit that the anti-KCNIP4 abs might recognize other proteins in addition to KCNIP4 in immunocytochemistry. Nevertheless, as we did not see KCNIP4-like immunoreactive signal for endogenous proteins in nuclei, we reason that there are no detectable amounts of nuclear KCNIP4 in primary neurons. Our western blotting experiments suggest that the isoform of KCNIP4 that was expressed in primary neurons was KCNIP4-Ia, which is located to the nucleus in addition to the cytoplasm when over-expressed in neurons. This puts forward the possibility that the KCNIP proteins are small enough to passively shuttle through the nuclear pore (Dingwall and Laskey 1986), and emphasizes that over-expression of KCNIP proteins might not ideally reflect localization of the endogenous protein.
On the basis of our western blotting and in situ hybridization results, we suggest that in the human hippocampus KCNIP2-Ia and KCNIP4-Ia could be the most abundantly expressed KCNIP isoforms, although this has yet to be confirmed by quantitative analysis. Interestingly, with anti-KCNIP4 ab we detected a protein of slightly faster mobility in SDS-PAGE than the in vitro synthesized KCNIP4-Ia in addition to a KCNIP4 isoform matching the mobility of KCNIP4-Ia. Likewise, in case of KCNIP3, only the marginally faster migrating form of KCNIP3-Ia was detected in human brain samples. This phenomenon resembles calcium-dependent mobility shift in SDS-PAGE, which has been described for NCS proteins (Garrigos et al. 1991; Teng et al. 1994) including KCNIP3-Ia (Buxbaum et al. 1998) and we could mimic this by adding CaCl2 to in vitro synthesized KCNIP proteins before electrophoresis (data not shown). However, it remains to be elucidated whether in the human brain in normal conditions the majority of KCNIP3-Ia is binding calcium.
We found that none of the KCNIP proteins was able to interfere with CRE-dependent transcription in neurons and we were unable to replicate interaction of KCNIP3-Ia with CREB by EMSA. In addition, we have demonstrated before that transcription of the BDNF gene, an established CREB target, is not regulated by the KCNIP proteins in primary cortical neurons (Pruunsild et al. 2011). However, because CREB is a central TF in calcium-regulated processes in neurons (Silva et al. 1998; Lonze and Ginty 2002; Hagenston and Bading 2011), it was important to control the proposed role of KCNIP3 as a modulator of CRE-dependent gene expression (Ledo et al. 2002; Fontan-Lozano et al. 2009). Using KCNIP3 knockout mice it has been suggested that absence of KCNIP3 accelerates CREB-dependent transcription during learning, producing enhanced memory and improved synaptic plasticity (Fontan-Lozano et al. 2009). Importantly, however, demonstration of KCNIP3 occupancy on endogenous CREB-dependent promoters was missing in this study and only in vitro EMSA results of KCNIP3 interaction with CREB on CRE element was shown, making it difficult to decisively conclude that KCNIP3-Ia functions as a CREB modulator in neurons in vivo. On the other hand, research on KCNIP3-Ia as a TF that binds the DRE element has provided several studies confirming presence of KCNIP3 on endogenous promoters, for example on the prodynorphin promoter in neuroblastoma cells (Campos et al. 2003), on the arylalkylamine N-acetyltransferase promoter in pineal gland and retina (Link et al. 2004) and on the IL-2, IL-4, and IFNγ promoters in lymphocytes (Savignac et al. 2005). In view of our results that KCNIP proteins are not involved in the regulation of CRE-dependent transcription in neurons, an alternative explanation would be expected for the enhanced memory phenotype of the KCNIP3 knockout mice, and accordingly, it has been proposed that the memory enhancement might be because of combined effects of altered Kv4.2 functioning and prodynorphin expression during hippocampus dependent learning (Alexander et al. 2009). Finally, we would like to draw attention to the fact that some of the discrepancies in the results of previous studies on KCNIP3 might be because of usage of a KCNIP3 construct encoding a protein that has 30 additional N-terminal amino acids not present in the WT KCNIP3-Ia or KCNIP3-Ib isoforms (Carrion et al. 1999). Several studies have referred to having used this KCNIP3-Ia construct afterwards (Ledo et al. 2000a, 2002; Savignac et al. 2005) without determining if the artificial N-terminus affects the results.
In conclusion, we showed here that different KCNIP isoforms display variable localization in the nucleus and cytoplasm when over-expressed in neuronal cells. However, we also showed that endogenous KCNIP proteins were not detectable in the nuclei of rat primary cortical neurons. Lastly, our luciferase-reporter experiments and EMSA results do not support the role of KCNIP proteins in the modulation of CRE-dependent transcription in cortical neurons.