Subcellular localization and transcription regulatory potency of KCNIP/Calsenilin/DREAM/KChIP proteins in cultured primary cortical neurons do not provide support for their role in CRE-dependent gene expression


Address correspondence and reprint requests to Tõnis Timmusk and Priit Pruunsild, Department of Gene Technology, Tallinn University of Technology, Akadeemia tee 15, 12618, Tallinn, Estonia. E-mails:;


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.

Abbreviations used



calsenilin like protein


CREB binding protein


chromatin immunoprecipitation


cAMP/Ca2+-response element


CRE binding protein


CRE binding protein 1


CRE modulator




dentate gyrus


days in vitro


downstream regulatory element


DRE antagonist modulator


calcium binding EF-hand motif


EF-hand mutated


electrophoretic mobility shift assay


endoplasmatic reticulum


glial fibrillary acidic protein


inducible cAMP early repressor






Kv4 interacting protein


Kv4 interacting protein


voltage-dependent K+ channel


N-myrsitoylation motif


neuronal calcium sensor


S-palmitoylation motif


signal peptide


transcription factor


transmembrane helix


V5 epitope


wild type

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.

Materials and methods

Plasmid constructs

PCR with human brain cDNA as a template was performed with Expand High-Fidelity PCR system kit (Roche Applied Science, United Kingdom) according to manufacturer’s instructions to isolate cDNAs encoding full-length human CREB1, ICER-Iγ (inducible cAMP early repressor isoform I gamma; Borlikova and Endo 2009) and different KCNIP proteins. PCR fragments were resolved by agarose gel electrophoresis, excised from gel and cloned into pcDNA3.1/V5-His vector (Invitrogen, Carlsbad, CA, USA). All cloned PCR fragments were verified by sequencing. Primers for human CREB1 were CCACCATGACCATGGAATCTGGAGCCGAG and ATCTGATTTGTGGCAGTAAAGGTCC, and for human ICER-Iγ CACCATGGCTGTAACTGGAGATGAC and CTCTACTTTATGGCAATAAAGATC. Primers for human KCNIPs were: hKCNIP1IaS, CACCATGAGCGGCTGCTCCAAAAG, and hKCNIP1AS, CATGACATTTTGAAACAGCTGGAG (for KCNIP1-IaΔII); hKCNIP1IbS, CACCATGGGGGCCGTCATGGGCAC, and hKCNIP1AS (for KCNIP1-Ib and KCNIP1-IbΔII); hKCNIP2IaS, CACCATGCGGGGCCAGGGCCGCAAG, and hKCNIP2AS, GATGACATTGTCAAAGAGCTGC, (for KCNIP2-Ia, KCNIP2-IaΔIIb and KCNIP2-IaΔIIab); hKCNIP2IbS, CACCATGAACCGATGCCCCCG, and hKCNIP2AS (for KCNIP2-IbΔIIab); hKCNIP3IaS, CACCATGCAGCCGGCTAAG, and hKCNIP3AS, GATGACATTCTCAAACAGCTGC, (for KCNIP3-Ia); hKCNIP3IbS, CACCATGGGCATCCAGGGCATGGAGC, and hKCNIP3AS (for KCNIP3-IbΔII); hKCNIP4IaS, CACCATGAATGTGAGGAGGGTGGAAAG, and hKCNIP4AS, AATCACATTTTCAAAGAGCTGC, (for KCNIP4-Ia and KCNIP4-IaΔII); hKCNIP4IbS, CACCATGAGTGGCTGTAGAAAGCG, and hKCNIP4AS (for KCNIP4-IbΔII); hKCNIP4IdS, CACCATGAACTTGGAAGGGCTTG, and hKCNIP4AS (for KCNIP4-IdΔII); hKCNIP4IeS, CACCATGTTGACTCTGGAGTGGGAG, and hKCNIP4AS (for KCNIP4-IeΔII). For cloning cDNAs encoding human KCNIP isoforms without the V5 tag the antisense primers included the stop codon and were as follows: hKCNIP1stopAS, TTACATGACATTTTGAAACAGCTGG; hKCNIP2stopAS, CTAGATGACATTGTCAAAGAGCTGC; hKCNIP3stopAS, CTAGATGACATTCTCAAACAGCTGC; and hKCNIP4stopAS, TTAAATCACATTTTCAAAGAGCTGC. Mutation in the fourth calcium binding EF-hand motif (EF) encoding sequence of KCNIP3-Ia was inserted according to published data (Carrion et al. 1999). KCNIP1-IbΔII, KCNIP2-Ia, and KCNIP4-Ia were mutated similarly. The sense primer sequences for mutagenesis were: CTTCTTCCAGAAAATGGCCAAAATTAAAGATGGCATCGTAAC (KCNIP1), CTTCTTCCAGAAGATGGCCAGAATCAAGGATGGTGTGGTG (KCNIP2), GTTCTTCGAGAAAATGGCCCGGATCCAGGATGGGGTAGTG (KCNIP3), and CATTTTTTCAGAAAATGGCCAAAATTAAAGATGGGGTTGTTAC (KCNIP4). Mutagenesis with complementary primers was performed with Phusion High-Fidelity DNA Polymerase (Finnzymes, Finland). After 25 PCR cycles the samples were treated with DpnI (Fermentas, Lithuania) to degrade the wild type (WT) template plasmid and transformed into TOP10 competent cells (Invitrogen). All mutations were verified by sequencing.

Rat primary cortical neuron culture

The local ethics committee agreed with all animal procedures carried out in this study. Primary cortical neuron cultures were prepared from E21 rat embryo brains (Sprague–Dawley, Scanbur BK AB, Sweden). The cortices, together with hippocampi, were dissociated with 0.25% trypsin (Gibco, Rockville, MD, USA), treated with 0.05% DNase I (Roche) and the cells were plated on poly-L-lysine-coated dishes in Neurobasal A medium (Gibco) with B27 supplement (Gibco), penicillin (PAA, Germany; 100 U/mL), streptomycin (PAA, 0.1 mg/mL), and 1 mM l-glutamine (PAA). 5-fluoro-2′-deoxyuridine (Sigma-Aldrich, Germany) was added to the medium (final concentration 10 μM) at 2 days in vitro (DIV). Neuronal activity was modeled by activation of voltage-sensitive calcium channels with final concentration of 25 mM KCl.


Transfections with different human full length KCNIP isoform cDNAs in pcDNA3.1/V5-His vector (Invitrogen) using FuGENE 6 (Roche) for HiB5 cells (kind gift from Prof. Elena Cattaneo, University of Milan, Italy) and Lipofectamine 2000 (Invitrogen) for rat primary neurons (7 DIV) were performed according to manufacturer’s instructions. 0.5 μg DNA per 0.75 cm² well of a 48-well cell culture plate was used. Immunocytochemistry for endogenous KCNIP proteins in primary neurons was performed without transfection. Twenty-four hours after transfection (or at 8 DIV without transfection), cells were washed twice with phosphate-buffered saline (PBS), fixed 15 min in 4% paraformaldehyde in PBS, treated 10 min with 50 mM NH4Cl in PBS, washed twice with PBS, and permeabilized 10 min in 0.5% Triton X-100 in PBS. The cells were subsequently washed twice with PBS and incubated in PBS containing 1% bovine serum albumin and 0.05% Tween 20 (blocking buffer) for 30 min. The cells were incubated with anti-V5 antibody (ab) (Invitrogen cat# R960-25; diluted 1 : 1000 into blocking buffer) or anti-KCNIP1 ab (anti-KChIP1 K+ channel, NeuroMab Antibodies Incorporated, Davis, CA, USA; clone K55/7; 1 : 500), anti-KCNIP2 ab (anti-KChIP2 K+ channel, NeuroMab Antibodies Incorporated, clone K60/73; 1 : 500), anti-KCNIP3 ab (anti-Calsenilin/DREAM/KChIP3 K+ channel, NeuroMab Antibodies Incorporated, clone K66/38; 1 : 500), anti-KCNIP4 ab (anti-KChIP4 (L-14), Santa Cruz Biotechnology, Santa Cruz, CA, USA; cat# SC-46380; 1 : 125), and anti-MAP2 ab (Chemicon, Temecula, CA, USA; cat# MAB378; 1 : 200) all for 1 h at 20–22°C and washed three times with PBS containing 0.05% Tween 20. Subsequently, the cells were incubated with Alexa fluorescent dye conjugated goat anti-mouse, goat anti-rabbit or rabbit anti-goat secondary ab (all Molecular Probes, Eugene, OR, USA; diluted 1 : 2000 into blocking buffer) for 1 h at 20–22°C, washed three times with PBS containing 0.05% Tween 20 and once with distilled water. The cells were mounted in ProLong Gold antifade reagent with 4′-6-diamidino-2-phenylindole (DAPI; Molecular Probes).

Western blotting

Proteins were separated by 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), transferred to polyvinylidene fluoride membrane (Millipore Corporation, Bedford, MA, USA), blocked in 5% skim milk and 0.05% Tween 20 in PBS (western blocking buffer) for 30 min at 20–22°C and incubated with primary ab for 1 h and then with secondary ab for 1 h at 20–22°C in western blocking buffer. Washing with 0.05% Tween 20 in PBS, three times, followed both incubations. The same primary abs as for immunocytochemistry, and in addition, anti-panKCNIP ab (anti-Pan-KChIP, NeuroMab Antibodies Incorporated, clone K55/82), were used. Dilutions of primary abs used: anti-KCNIP1 1 : 2000, anti-KCNIP2 1 : 2000, anti-KCNIP3 1 : 2000, anti-KCNIP4 (L-14) 1 : 500, anti-panKCNIP 1 : 2000, and anti-V5 1 : 5000. The secondary abs were HRP-conjugated anti-mouse ab (Pierce, Rockford, IL, USA; 1 : 5000) or anti-goat ab (LabAs, Estonia; 1 : 100 000). Chemiluminescent signal was detected with SuperSignal West Femto Chemiluminescent Substrate (Pierce).

In situ hybridization

Full-length KCNIP1-IaΔII, KCNIP2-Ia, KCNIP3-Ia, and KCNIP4-IaΔII cDNAs were cloned into pSC-A vector (Stratagene, La Jolla, CA, USA) and the plasmids were linearized with restriction enzymes from Fermentas: NcoI (KCNIP1), Eco81I (KCNIP2 and KCNIP3), and MlsI (KCNIP4). This generated templates for synthesizing probes enabling detection of all of the transcripts of the respective KCNIP gene. cRNA probes were synthesized using MAXIScript in vitro Transcription Kit (Ambion, Austin, TX, USA) with the T7 RNA polymerase, using [α-35S]UTP (Amersham Biosciences, Switzerland) for labeling. Serial sections (14 μm) from frozen post-mortem human hippocampal tissue were fixed in 4% paraformaldehyde in PBS for 15 min and rinsed twice in PBS and twice in distilled water. Then the sections were treated with 0.1 M HCl for 10 min, acetylated for 20 min with 0.25% acetic anhydride in 0.1 M ethanolamine, dehydrated with ethanol, and air-dried. The sections were hybridized overnight at 52°C in a chamber humidified with 50% formamide and 2.5x SSPE (375 mM NaCl, 25 mM NaH2PO4 and 2.5 mM EDTA) with 250 μL of hybridization buffer (2.5x SSPE, 50% formamide, 12.5% dextran sulfate, 2.5x Denhardt’s solution, 0.5 mg/mL yeast tRNA, and 10 mM DTT) per slide containing 1.1 × 107 cpm/mL [α-35S]UTP-labeled cRNA probe. After hybridization, slides were washed twice in 2x SSC (300 mM NaCl and 30 mM sodium citrate) at 37°C for 20 min, treated with RNase A (20 μg/mL) in 0.5 M NaCl, 10 mM Tris-HCl, 1 mM EDTA at 37°C for 30 min, and washed twice in 2x SSC at 37°C for 5 min. Then, slides were washed with 50% formamide with 2x SSC and 1 mM DTT for 30 min, in 2x SSC for 30 min, and in 0.2x SSC for 30 min, all at 52°C, and again in 50% formamide with 2x SSC and 1 mM DTT at 63°C for 45 min and rinsed first in 2x SSC and then in 0.2x SSC at 20–22°C and dehydrated with ethanol. Emulsion-dipped sections were developed after 4 weeks using D-19 developer and fixed with sodium fixer (both Eastman Kodak, NY, USA).

Luciferase assay

Rat primary neurons were transfected at 7 DIV and PC12 cells [grown in Dulbecco’s modified Eagle’s medium supplemented with 10% horse serum, 5% fetal bovine serum, 100 U/mL penicillin, and 0.1 mg/mL streptomycin (all PAA)] at 70% confluency with pGL4.29[luc2P/CRE/Hygro] (Promega, Madison, WI, USA) plasmid using Lipofectamine 2000 (Invitrogen) and LipoD293 (SignaGen Laboratories, Gaithersburg, MD, USA), respectively, according to manufacturers’ instructions. Empty vector pRC and KCNIP or ICER encoding pcDNA3.1 expression constructs were co-transfected in 1 : 1 ratio with the pGL4.29[luc2P/CRE/Hygro] construct. For normalization, pEF1alpha/pGL4.83[luc2P/Hygro] (1/100 of pGL4.29[luc2P/CRE/Hygro] quantity) was co-transfected. Altogether 0.5 μg DNA per 0.75 cm² well of a 48-well cell culture plate was used. 24 h post transfection neuronal membranes were depolarized by 25 mM KCl and the medium of PC12 cells was supplemented with 5 mM caffeine. Luciferase assays were performed with Dual-Glo Luciferase Assay System (Promega) using cell lysates prepared at indicated time points. At least three independent experiments were performed in duplicate. For presentation of data in relative luciferase units (RLUs), background signals from untransfected neurons were subtracted from signals obtained from transfected cells and firefly luciferase signal values were normalized to EF1alpha promoter-dependent Renilla luciferase signals.


All luciferase assay data were log-transformed and autoscaled using data of pGL4.29[luc2P/CRE/Hygro] and pRC co-transfected cells. Means and standard deviations (SD) were calculated. Analyses of statistical significance were performed by t-tests. The data were back-transformed to original scale. Error bars represent upper and lower limits back-transformed as mean + SD and mean − SD, respectively.

Electrophoretic mobility shift assay

In vitro translated proteins were produced with TnT Quick Coupled Transcription/Translation System (Promega) according to manufacturer’s instructions. Somatostatin CRE oligonucleotide (TCCTTGGCTGACGTCAGAGAGAGA, CRE underlined) was radioactively labeled with T4 polynucleotide kinase (Fermentas) according to manufacturer’s instructions. For oligo binding reaction the in vitro synthesized proteins in 15 μL of binding buffer (10 mM HEPES-KOH pH 7.9, 10% glycerol, 0.1 mM EDTA, 8 mM MgCl2, and 1 mM DTT) containing 0.2 μg of poly(dI-dC), were pre-incubated on ice for 10 min. Then, 0.05 pmol of radioactively labeled double-stranded oligonucleotide was added and incubated 20 min at 20–22°C. For DNA competition, 10-fold excess of unlabeled double-stranded oligonucleotide was added 5 min before radioactive probe. The oligonucleotide with mutated CRE was TCCTTGGCTGAATTAAGAGAGAGA (mutated nucleotides in bold). Where indicated, protein lysates were incubated with 1 μg of ab for 1 h in reaction buffer prior to incubation with the radioactively labeled probe. The abs used were rabbit anti-CREB1 (Millipore cat# 06-863) or anti-Calsenilin/DREAM/KChIP3 K+ channel (NeuroMab Antibodies Incorporated, clone K66/38). DNA-protein complexes were resolved in 5% non-denaturing polyacrylamide gel and visualized by autoradiography.


Localization of over-expressed KCNIP-V5 isoforms in neuronal cells

At least 14 different transcripts are exprerssed in the nervous system from the four KCNIP family genes. Protein isoforms encoded by these mRNAs differ in their N-termini because of alternative promoter usage and splicing (Pruunsild and Timmusk 2005; and Fig. 1). The C-terminal regions of the KCNIP proteins that contain EF-hand motifs and bind calcium are either identical, when encoded by the same gene, or highly similar, when originating from paralogous gene loci (Pruunsild and Timmusk 2005). To gain insights into the subcellular localization of KCNIP family proteins in neurons, we transfected rat neural progenitor HiB5 cells (Renfranz et al. 1991) and rat primary cortical neurons with plasmid constructs encoding different isoforms of human KCNIPs fused with the V5 antigen and subjected the transfected cells to immunocytochemistry with the anti-V5 ab. We analyzed the localization of the V5-tagged KCNIP isoforms by fluorescence microscopy in HiB5 cells and by confocal fluorescence microscopy in primary neurons. Our transfections included three N-terminally different isoforms of KCNIP1: KCNIP1-IaΔII, KCNIP1-Ib [also known as KChIP1b (Boland et al. 2003)] and KCNIP1-IbΔII [KChIP1 (An et al. 2000) and KChIP1a (Boland et al. 2003)]; four isoforms of KCNIP2: KCNIP2-Ia [KChIP2L (Ohya et al. 2001)], KCNIP2-IaΔIIb [KChIP2 (An et al. 2000)], KCNIP2-IaΔIIab [KChIP2S (Ohya et al. 2001)] and KCNIP2-IbΔII [KChIP2g (Decher et al. 2004)]; two isoforms of KCNIP3: KCNIP3-Ia [Calsenilin (Buxbaum et al. 1998), DREAM (Carrion et al. 1999) and KChIP3 (An et al. 2000)] and KCNIP3-Ib; and five isoforms of KCNIP4: KCNIP4-Ia [KChIP4bl (Holmqvist et al. 2002) and CALP250 (Morohashi et al. 2002)], KCNIP4-IaΔII [CALP216 (Morohashi et al. 2002)], KCNIP4-IbΔII, KCNIP4-IdΔII [KChIP4a (Holmqvist et al. 2002)] and KCNIP4-IeΔII (Fig. 1). The results showed that different KCNIPs had distinct subcellular distributions. All of the KCNIP1-V5 isoforms were confined to the cytoplasm, both in HiB5 cells and primary neurons, where the V5 signal was distributed in a punctate pattern, most probably corresponding to KCNIP1-V5 localization to post-ER vesicles (Fig. 1). KCNIP1-IaΔII-V5 and KCNIP-Ib-V5 localized more than KCNIP-IbΔII to the perinuclear compartment in HiB5 cells, whereas in primary neurons this propensity was not obvious. Very faint nuclear staining with the anti-V5 ab in primary neurons was present for all of the over-expressed KCNIP1-V5 isoforms as identified by confocal microscopy (Fig. 1). All analyzed KCNIP2-V5 isoforms had a similar localization pattern, which clearly differed from that of KCNIP1-V5 isoforms. In HiB5 cells, all of the KCNIP2-V5 isoforms were detected diffusely all over the cells, including nuclei. Nuclear localization was more pronounced in case of the KCNIP2-IaΔIIab and KCNIP2-IbΔIIab isoforms. In primary neurons the different KCNIP2-V5 isoforms were diffusely localized in the cytoplasm as well as the nucleus. However, the nuclear signal for the KCNIP2-V5 isoforms was less distinct in primary neurons compared with HiB5 cells (Fig. 1). Localization of KCNIP3-V5 and KCNIP4-V5 isoforms depended on the N-terminus of the respective KCNIP. Whereas KCNIP3-Ia-V5, alias Calsenilin/DREAM/KChIP3, was localized diffusely in cells and, similarly to KCNIP2-V5 proteins was present also in the nucleus both in HiB5 cells and in primary neurons, the KCNIP3-Ib isoform had perinuclear localization both in HiB5 cells and primary neurons. Again, as with KCNIP2-V5 isoforms, KCNIP3-Ia-V5 localization to the nuclei was more apparent in HiB5 cells than in primary neurons (Fig. 1). In case of KCNIP4, different N-termini had divergent effects on the localization of the respective isoform. KCNIP4-Ia-V5 and KCNIP4-IaΔII-V5 isoforms were detected both in the cytoplasm and nuclei in HiB5 cells and in primary neurons. The signal for KCNIP4-Ia-V5 was diffuse but in addition comprised a punctate pattern in the cytoplasm, similar to that found for the KCNIP1-V5 isoforms, and this was more pronounced in primary neurons than in HiB5 cells. KCNIP4-IaΔII-V5 on the other hand had diffuse distribution resembling the localization of the KCNIP2-V5 isoforms. Fluorescent signal for KCNIP4-IbΔII-V5 exhibited punctate arrangement in the cytoplasm, especially in primary neurons, and therefore was located in the same way as the KCNIP1-V5 isoforms. Both in HiB5 cells and in primary neurons KCNIP4-IdΔII-V5 and KCNIP4-IeΔII-V5 had perinuclear localization resembling the distribution of KCNIP3-Ib-V5. As with the KCNIP1-V5 proteins, very weak nuclear staining was detected for over-expressed KCNIP4-IbΔII-V5, KCNIP4-IdΔII-V5, and KCNIP4- IeΔII-V5 in primary neurons by confocal microscopy (Fig. 1). Of note, with primary neurons we modeled neuronal activity by KCl-induced depolarization of neurons to identify possible changes of KCNIP-V5 subcellular localization, but did not detect differences between the control and depolarized cells (data not shown). In summary, these results show that when over-expressed in neuronal cells (i) all KCNIP1 isoforms and the KCNIP4-IbΔII isoform have similar punctate cytoplasmic localization, (ii) all KCNIP2 isoforms, the KCNIP3-Ia and the KCNIP4-IaΔII isoform have diffuse localization in the cytoplasm and the nucleus, (iii) the KCNIP3-Ib isoform and KCNIP4-IdΔII and KCNIP4-IeΔII isoforms are localized in a similar manner to the perinuclear compartment in the cytoplasm, and (iv) the KCNIP4-Ia isoform has a mixture of diffuse and punctate distribution in the cytoplasm and is present in lower amounts also in the nucleus.

Figure 1.

 Subcellular localization of over-expressed Kv4 interacting protein (KCNIP)-V5 isoforms in neuronal cells. HiB5 cells and rat primary cortical neurons were transfected with different human KCNIP-V5 expressing constructs and immunostained with anti-V5 ab (green). 4′-6-diamidino-2-phenylindole (DAPI; pseudocoloured red in HiB5 images) was used to visualize cell nuclei. Anti-MAP2 ab was used to identify neurons in primary neuron cultures (red). Fluorescence microscopy images for HiB5 cells and single section confocal fluorescence microscopy images for primary neurons are shown. Schematic representations of each of the transfected KCNIP isoform are at the left of the images. Open boxes represent variable N-termini of the KCNIP isoforms whereas each box stands for the sequence encoded by the respective exon of the gene. The names of the exons are shown inside the boxes and are according to Pruunsild and Timmusk 2005. Supplementary notes in superscript denote sites of post-translational modification or a linear protein motif encoded by the corresponding exon. When the supplementary note is without parentheses, experimental evidence exists for the usage of the respective site. When the supplementary note is in parentheses, only bioinformatic prediction is available. See the Discussion part of text for details. When available, names of the specific isoform given elsewhere are shown below the schematic of the respective isoform. Scale bars are 20 μm. Asterisks mark representative non-transfected cells. p, S-palmitoylation motif; m, N-myrsitoylation motif; TMH, transmembrane helix; SP, signal peptide, EF, calcium binding EF-hand motif; V5, V5 epitope.

Cytoplasmic distribution of endogenous KCNIPs in rat primary cortical neurons

Ectopic expression of tagged proteins is a good model for analyses of localization of different isoforms not distinguishable by other means, but it does not necessarily reflect the distribution of the endogenous protein in cells. We therefore analyzed the expression of endogenous KCNIPs in rat primary cortical neurons by fluorescent confocal microscopy using anti-KCNIP abs. In addition to control conditions, we used KCl-induced depolarization of neurons to induce calcium signaling in cells. In rat primary cortical neurons immunoreactive signal was detectable with anti-KCNIP1 and anti-KCNIP4 abs, but not with anti-KCNIP2 and anti-KCNIP3 abs (Fig. 2a). This result is in accord with a previous study showing only KCNIP1 and KCNIP4 protein expression in cultured primary neurons (Shibata et al. 2003). Both KCNIP1- and KCNIP4-like immunoreactive signals were restricted to the cytoplasm of neurons and showed negligible, if any, localization to the nuclei. Depolarization of neurons for 1 h did not alter this distribution (Fig. 2a) and depolarization for 2 h also had no effect (data not shown). To confirm that the fluorescence detected was not produced by non-specific binding of secondary abs, we included negative controls where the primary abs were omitted from the immunocytochemistry protocol (Fig. 2a). Also, to control if the absence of immunoreactive signal in case of anti-KCNIP2 ab and anti-KCNIP3 ab was indeed because of undetectable levels of KCNIP2 and KCNIP3 proteins in neurons, we transfected KCNIP encoding constructs, namely KCNIP1-IbΔII, KCNIP2-Ia, KCNIP3-Ia, or KCNIP4-Ia, into neurons and subjected the cells to immunocytochemistry with the anti-KCNIP abs. Firstly, the results substantiated that the anti-KCNIP abs, including anti-KCNIP2 ab and anti-KCNIP3 ab, recognized the corresponding over-expressed human KCNIP protein in cells and thus work in immunocytochemistry (Fig. 2b). Secondly, as the transfected constructs in this experiment encoded for KCNIP isoforms without the V5 tag and the localization of these over-expressed proteins was highly similar to the localization of the corresponding isoforms with the V5 tag (Fig. 2b), these results additionally indicate that the V5 tag does not markedly affect the subcellular localization of KCNIP isoforms. Although the over-expressed KCNIP proteins were present in neuronal nuclei in addition to the cytoplasm, the untransfected cells in the same cultures displayed only cytoplasmic staining with anti-KCNIP1 and anti-KCNIP4 abs (Fig. 2b). Taken together, no nuclear signal was detected with the anti-KCNIP abs for endogenous proteins in rat primary cortical neurons.

Figure 2.

 Kv4 interacting protein (KCNIP)1 and KCNIP4 are expressed with cytoplasmic distribution in rat primary cortical neurons. (a) Rat primary neurons (8 DIV) were immunostained with mouse anti-KCNIP1, anti-KCNIP2, anti-KCNIP3, or goat anti-KCNIP4 ab (all green). DAPI was used to visualize cell nuclei and rabbit anti-MAP2 ab was used to identify neurons (red). Primary abs were omitted where indicated. Neurons were left untreated (CNTR) or treated with 25 mM KCl for 1 h. (b) Rat primary cortical neurons were transfected with human KCNIP1-IbΔII, KCNIP2-Ia, KCNIP3-Ia, or KCNIP4-Ia constructs encoding KCNIP isoforms without the V5 tag and immunostained with anti-KCNIP1, anti-KCNIP2, anti-KCNIP3, or anti-KCNIP4 ab (all green), respectively. Anti-MAP2 ab was used to identify neurons (red). Arrows point to transfected cells and asterisks mark untransfected cells. Single section confocal fluorescence microscopy images are shown in (a) and (b). All the images in (a) were obtained with the same laser and camera settings, whereas in (b) the laser intensity for Alexa488 excitation was lower than in (a) but higher than in Fig. 1, to allow simultaneous imaging of the immunoreactive signals for the over-expressed and endogenous proteins. Scale bar is 20 μm.

Specificity of the anti-KCNIP antibodies and expression of KCNIP isoforms in rat primary neurons and human brain

Next, to compare the localization of endogenous proteins to that of over-expressed ones, we sought to determine which of the isoforms of KCNIP1 and KCNIP4 were expressed in rat primary neurons. First, we controlled by western blotting whether the KCNIP abs used in this study recognize all of the different isoforms of the respective KCNIP and whether the abs cross-react with KCNIPs encoded by the paralogous KCNIP genes. In vitro synthesized human KCNIP isoforms without the V5 tag were produced for that and subjected to analyses with anti-KCNIP1, anti-KCNIP2, anti-KCNIP3, anti-KCNIP4, or anti-panKCNIP abs. We determined that all of the different KCNIP abs used were specific for the respective KCNIP protein and did not recognize any of the other KCNIP proteins (Fig. 3a). Also, the analysis showed that the anti-KCNIP1 ab, anti-KCNIP2 ab, anti-KCNIP3 ab, and anti-KCNIP4 ab recognize all of the N-terminally different KCNIP1 isoforms, KCNIP2 isoforms, KCNIP3 isoforms, and KCNIP4 isoforms, respectively, and that the anti-panKCNIP ab recognizes all the KCNIP isoforms except human KCNIP3 protein isoforms (Fig. 3a). As different KCNIP isoforms had differences in mobility in SDS-PAGE we then used lysates of rat primary neurons and the in vitro synthesized KCNIP proteins to describe which of the isoforms are expressed in rat primary neurons. In parallel, we included lysates of human cortical and hippocampal tissue in the analyses as controls. The results were in accordance with our immunocytochemistry analyses since KCNIP1 and KCNIP4, but not KCNIP2 and KCNIP3, were expressed in rat primary neurons as evidenced by western blotting (Fig. 4b). In case of KCNIP1 the proteins detected from the lysates of primary neurons and human cortical and hippocampal samples were of higher molecular weight than the in vitro synthesized KCNIP1 isoforms (Fig. 3b) suggesting that KCNIP1 is modified post-translationally in neurons in vivo. From primary neuron lysates and human cortical and hippocampal tissue samples, we obtained two immunoreactive signals with the anti-KCNIP4 ab that were in the range of predicted molecular weight of KCNIP4 isoforms (Fig. 3b). One of the signals matched in size with the molecular weight of the in vitro synthesized KCNIP4-Ia and was the predominant KCNIP4 isoform expressed in primary neurons. The other signal was for a slightly lower molecular weight protein not corresponding to the weight of any in vitro synthesized KCNIP4 isoform. This signal was detected at equal levels with the KCNIP4-Ia isoform in cortical and hippocampal tissue (Fig. 3b). The results also showed that the KCNIP2 isoforms detected by western blotting in human cortex and hippocampus were KCNIP2-Ia and either KCNIP-IaΔIIab or KCNIP-IbΔIIab or both, whereas in the hippocampus KCNIP2-Ia was the major isoform expressed and in the cortex the other isoforms predominated (Fig. 3b). With the anti-KCNIP3 ab, we detected relatively higher expression levels of KCNIP3 in the cortex than in the hippocampus, whereas the protein detected from tissue samples had a slightly faster mobility in SDS-PAGE than the in vitro synthesized KCNIP3-Ia isoform (Fig. 3b). Altogether, these results argue that the isoform of KCNIP4 expressed endogenously in rat primary neurons is KCNIP4-Ia and that any of the three KCNIP1 isoforms could be endogenously expressed in rat primary neurons while the isoform of KCNIP1 that is expressed, is probably modified post-translationally. In addition, these results suggest that KCNIP2-Ia could be the most abundant KCNIP isoform expressed in human hippocampus. For further evidence to this notion we analyzed expression of the KCNIP genes in the human hippocampus by in situ hybridization with KCNIP1-, KCNIP2-, KCNIP3-, or KCNIP4-specific probes recognizing all of the transcripts of the respective KCNIP gene. We detected strong expression of KCNIP2 in the dentate gyrus (DG) granule cells and in the pyramidal neurons of the CA regions of the hippocampus (Fig. 3c). KCNIP4 was also relatively strongly expressed in the DG granule cells and the pyramidal neurons of the hippocampus, whereas in the CA2 region KCNIP4 mRNA was much less abundant than in the CA1 and CA3 region. KCNIP3 transcripts were barely detectable only in the DG cells and KCNIP1 expression did not reach the detection limit of our assay (Fig. 3c).

Figure 3.

 Characterization of specificity of anti-KCNIP antibodies and expression of Kv4 interacting protein (KCNIP) isoforms in rat primary neurons and human brain. (a) Western blots showing the differential mobility of the different in vitro synthesized KCNIP isoforms without the V5-tag in sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Note that all of the different KCNIP abs used are specific for the respective human KCNIP protein isoforms and the anti-panKCNIP ab recognizes all but human KCNIP3 isoforms. (b) Western blots showing the expression pattern of different KCNIP protein isoforms in human cortex and hippocampus (50 μg protein loaded for both) and in rat primary cortical neurons (8 DIV, 100 μg protein loaded). Of note, KCNIP4 expression is detected with anti-KCNIP4 ab in tissue samples and primary neurons and with pan-KCNIP ab in in vitro synthesized lysates to avoid background signal generated by the anti-KCNIP4 ab in in vitro synthesized protein lysate lanes. (c) In situ hybridization analysis of KCNIP1, KCNIP2, KCNIP3, and KCNIP4 expression in human hippocampus. Dark-field emulsion autoradiographs of coronal sections of hippocampus are shown. CA1, CA2, and CA3, corresponding hippocampal regions; DG, dentate gyrus; S, subiculum. Scale bar is 1 mm.

Figure 4.

 KCNIP (Kv4 interacting proteins) do not modulate CRE (cAMP/Ca2+-response element)-CREB (CRE binding protein)-CBP-mediated transcription in rat primary cortical neurons and PC12 cells. (a) Rat primary cortical neurons or PC12 cells transfected with CRE-dependent firefly luciferase (Luc) construct, together with empty vector pRC or the respective KCNIP or inducible cAMP early repressor (ICER) encoding plasmid and hEF1alpha promoter-dependent renilla luciferase (hRluc) construct, were left untreated (CNTR) or treated with 25 mM KCl (neurons) or with 5 mM caffeine (PC12 cells) for the time indicated and luciferase activities were measured. CRE-dependent promoter activities are presented in RLUs (defined in Materials and Methods) relative to the promoter activity measured from untreated empty vector pRC co-transfected cells that was arbitrarily set as 1. Results of three independent experiments with duplicate measurements normalized to hRluc activities are shown. Statistical significance denoted by asterisk is relative to pRC transfected CRE-dependent RLU at respective time point (*< 0.05; t-test). (b) EMSA (electrophoretic mobility shift assay) showing CREB1-V5 and ICER-V5 complexes binding to the somatostatin CRE. Input of in vitro synthesized CREB1-V5, ICER-V5, and KCNIP-V5 proteins was ascertained with western blotting with the V5 ab shown above EMSA results. Ratio of CREB1-V5 to ICER-V5, 1:2; ratio of CREB1-V5 to the respective KCNIP-V5, 1:5. (c) EMSA showing that excess of KCNIP3-Ia-V5 in the reaction does not disturb CREB1-V5 binding to the somatostatin CRE. Input is shown as in b. Note that neither 5-fold (x5) nor 10-fold (x10) excess of KCNIP3-Ia over CREB1-V5 in the binding reaction influenced formation of the CREB-CRE complex. (d) EMSA showing the specificity of CREB1-V5 and ICER-V5 binding to the somatostatin CRE. Addition of 10-fold excess (10x) of unlabeled WT or CRE mutated (CREm) oligonucleotide or addition of V5 ab to reaction is indicated above the panel. In vitro synthesized proteins in the reaction are indicated below the panel (CREB1-V5:ICER-V5, 1 : 1). (e) EMSA showing that KCNIP3-Ia is not included in the CREB-CRE complex. Addition of ab to reaction and proteins used are designated as in d (CREB1-V5:KCNIP3-Ia-V5, 1 : 5). Asterisks mark unspecific binding in b–e.

KCNIP proteins do not participate in CRE-dependent gene expression regulation

Previous research on KCNIP3 as the transcriptional regulator DREAM has primarily relied on in vitro DNA binding assays, such as electrophoretic mobility shift assay (EMSA), and KCNIP3 over-expression in cell lines. We reasoned that since KCNIP3-Ia and KCNIP2 isoforms as well as two of the KCNIP4 isoforms can be detected in nuclei of neurons when they are over-expressed, it would be interesting to study transcription regulatory effects of the over-expressed KCNIPs also in primary neurons. The CREB family of TFs are one of the most significant proteins in calcium-dependent neuronal gene expression regulation and therefore we decided to test whether the KCNIPs affect CRE-dependent transcription in primary neurons, as described for KCNIP3-Ia in pheochromocytoma PC12 cells (Ledo et al. 2002). We transfected a vector containing four CRE elements in front of a minimal promoter driving transcription of the luciferase reporter gene along with different KCNIP isoforms, either WT or calcium binding EF-hand mutated (EFm), into primary neurons and examined the effect of over-expressed KCNIPs on CRE-dependent transcription. CRE-dependent transcription was induced by depolarizing neurons with the addition of KCl to the medium, which elicits calcium influx through voltage-sensitive calcium channels (Bading et al. 1993). We monitored luciferase activity starting already from the time point of 0.5 h after depolarization because it has been claimed that KCNIP3-Ia interacts with CREB and slows recruitment of CBP to CREB, thereby influencing the kinetics of initiation of CREB-dependent transcription (Fontan-Lozano et al. 2009). Our results showed that none of the KCNIP isoforms tested was able to negatively influence CRE-dependent luciferase activity levels (Fig. 4a). Conversely, the recognized negative regulator of CREB, inducible cAMP early repressor (ICER), which is an alternative isoform of CREM (Borlikova and Endo 2009) and was used here as a positive control, greatly reduced CRE-dependent transcription (Fig. 4a). Over-expression of ICER-V5 reduced the activity of CRE-dependent promoter four- to tenfold compared to control pRC transfected cells in every time point analyzed (Fig. 4a). Furthermore, the EFm KCNIPs that do not respond to calcium concentration increase in cells and therefore should inhibit CREB-CRE interaction even in depolarized conditions (Carrion et al. 1999; Ledo et al. 2002) had, similar to the WT KCNIP isoforms, no effect on CRE-dependent luciferase expression (Fig 4a). Notably, we used KCNIP3-Ia construct encoding KCNIP3-Ia without V5 tag to control if the V5 tag influences results, but similarly to the KCNIP3-Ia-V5 construct, did not detect any effect of KCNIP3-Ia on CRE-dependent luciferase activity (data not shown). To test if the discrepancy between our results and the results obtained by Ledo et al., could be caused by differences in the cells and treatment used, we transfected PC12 cells with the CRE-luciferase reporter construct together with KCNIP3-Ia, KCNIP3-Ia EFm, or ICER encoding plasmid and stimulated Ca2+/cAMP-singalling with the addition of caffeine as has been described (Ledo et al. 2002). Again, over-expression of KCNIP3-Ia or KCNIP3-Ia EFm proteins in PC12 cells did not affect the induction of CRE-dependent luciferase levels at the time points of 12 h and 24 h of caffeine treatment, whereas over-expression of ICER significantly reduced CRE-dependent luciferase levels in every time point analyzed (Fig. 4a). Thus, we propose that KCNIP3 does not modulate CRE-dependent transcription also in PC12 cells. These results together with our immunocytochemistry results suggest that albeit some of the KCNIP protein isoforms localize to the nuclei when over-expressed in neuronal cells, they do not participate in regulatory mechanisms of CRE-CREB-CBP-mediated transcription in primary neurons and PC12 cells.

It has been shown by EMSA that KCNIP3-Ia blocks both the formation of the CREB-CRE complex and CREB-CBP interaction (Ledo et al. 2002). However, in a later study results of EMSA experiments demonstrated that KCNIP3-Ia interferes with CBP recruitment to CREB without abolishing the CREB-CRE complex by interacting with CREB in CREB-CRE complex (Fontan-Lozano et al. 2009). We sought to clarify these contradictory results by conducting EMSA using an oligonucleotide containing the somatostatin promoter CRE element and in vitro synthesized human CREB1 protein fused to the V5 tag. The CREB-CRE interaction was monitored either in the absence or in the presence of in vitro synthesized human KCNIP-V5 protein isoforms. Again, ICER-V5 was used as a positive control for interference of the CREB-CRE complex formation. As can be seen from Fig. 4b ICER-V5 participated in the formation of CREB-CRE complexes by binding the CRE element both together with CREB1-V5 and without CREB1-V5. Conversely, none of the KCNIP-V5 isoforms used, including KCNIP3-Ia, changed the mobility or abolished the formation of the CREB-CRE complex. Of note, approximately 5-fold more KCNIP-V5 protein was added to the CRE-binding reaction compared to the amount of CREB1-V5, as evidenced from the control of input by western blot with the V5 ab above the panel of EMSA results (Fig. 4b). To exclude the possibility that the ratio of KCNIP3-Ia to CREB in the reaction was too low, we used 10-fold excess of KCNIP3-Ia-V5 compared to CREB1-V5 in CRE-binding assay, but did not detect any effect of KCNIP3-Ia on the CREB-CRE complex when compared to the CREB-CRE formed in the absence of KCNIP3-Ia (Fig. 4c). Next, we determined that the positive control ICER-V5 competes with CREB for binding the CRE site by showing that unlabeled WT but not CRE mutated (CREm) oligonucleotide was able to strongly reduce signals of the ICER-CRE, ICER-CREB-CRE, and the CREB-CRE complexes (Fig 4d). The specificity of these complexes was proven by addition of the V5 ab, which generated a supershift of the complexes (Fig 4d). Finally, to exclude the possibility that KCNIP3-Ia participates in the CREB-CRE complex without changing the mobility of the complex, we added either anti-CREB1 ab or anti-KCNIP3 ab to the CRE-binding reaction together with the CREB1-V5 and KCNIP3-Ia-V5 containing lysates. Anti-CREB1 ab created a supershift of the complex whereas anti-KCNIP3 ab neither reduced nor changed the mobility of the CREB-CRE complex (Fig. 4c). Similar results were obtained with in vitro synthesized KCNIP proteins without the V5 tag, as no influence on CREB-CRE complex was observed (data not shown). Collectively, these results argue that the KCNIP proteins are not recruited to the CREB-CRE complex and do not interfere with the formation of the CREB-CRE complex.


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.


We thank Enn Jõeste from North Estonian Regional Hospital for collaboration; Elena Cattaneo from University of Milan for HiB5 cells; Mari Sepp for opinions and discussions, and Epp Väli and Maila Rähn for technical assistance. This work was supported by Estonian Ministry of Education and Research (Grant 0140143), Estonian Science Foundation (Grant 8844), and Wellcome Trust (Grant 067952). The authors declare no conflict of interests.