Address correspondence and reprint request to Wilma Wasco, Genetics and Aging Research Unit, Department of Neurology, MassGeneral Institute for Neurodegenerative Disease and Harvard Medical School, Building 114, 16th Street, Charlestown, MA 02129, USA. E-mail: firstname.lastname@example.org
Calsenilin/potassium channel-interacting protein (KChIP)3/ downstream regulatory element sequence antagonist modulator (DREAM) is a neuronal calcium-binding protein that has been shown to have multiple functions in the cell, including the regulation of presenilin processing, repression of transcription and modulation of A-type potassium channels. To gain a better understanding of the precise role of calsenilin in specific cellular compartments, an interactor hunt for proteins that bind to the N-terminal domain of calsenilin was carried out. Using a yeast two-hybrid system and co-immunoprecipitation studies, we have identified the transcriptional co-repressor C-terminal binding protein (CtBP)2 as an interactor for calsenilin and have shown that the two proteins can interact in vivo. In co-immunoprecipitation studies, calsenilin also interacted with CtBP1, a CtBP2 homolog. Our data also showed a calsenilin-dependent increase in c-fos protein levels in CtBP knockout fibroblasts, suggesting that CtBP may modulate the transcriptional repression of c-fos by calsenilin. Furthermore, the finding that histone deacetylase protein and activity were associated with the calsenilin–CtBP immunocomplex suggests a mechanism by which calsenilin–CtBP may act to repress transcription. Finally, we demonstrated that calsenilin and CtBP are present in synaptic vesicles and can interact in vivo.
Calsenilin is a neuronal calcium-binding protein that was initially identified as a binding partner of two highly related familial Alzheimer's disease-associated proteins, presenilin (PS)1 and PS2 (Buxbaum et al. 1998). It has been shown to play a role in modulation of the levels of β-amyloid peptide (Aβ), and to enhance apoptosis in cultured cells (Jo et al. 2001, 2003, 2004; Lilliehook et al. 2002;). Raised levels of calsenilin have been reported in the cortex of patients with Alzheimer's disease, and in the neocortex and hippocampus of transgenic mice that overexpress the Swedish mutant form of the Aβ precursor protein (Jin et al. 2005). Altered Aβ formation and long-term potentiation have been reported in a calsenilin knockout model (Lilliehook et al. 2003) and overexpression of calsenilin has been shown to enhance γ-secretase activity associated with the presenilins (Jo et al. 2005).
Calsenilin appears to be a protein with multiple functions. It has also been shown to act as a calcium-dependent transcriptional repressor by binding to the downstream regulatory element sequence (DRE) of the genes that encode prodynorphin, c-fos and Hrk. Consequently, it was termed DRE antagonist modulator (DREAM) (Carrion et al. 1999; Sanz et al. 2001). Calsenilin/DREAM has been also shown to regulate the transcription of arylalkylamine N-acetyl transferase (AA-NAT), inducible cyclic AMP early repressor (ICER) and Fos-related antigen fra-2 (Link et al. 2004). In addition, calsenilin/DREAM has been shown to be involved in thyroid gene expression by binding to the DRE in the 5′ untranslated region of the transcription factors Pax8 and TTF-2/Foxe1 (D'Andrea et al. 2005). Recently, calsenilin/DREAM has been shown to act as a transcription activator for vitamin D and retinoic acid response elements (Scsucova et al. 2005). Calsenilin and two other closely related proteins have also been identified as potassium channel-interacting proteins (KChIPs) and shown to modulate the activity of A-type potassium channels. These studies referred to calsenilin as KChIP3, and the two related proteins as KChIP1 and 2 (An et al. 2000). A fourth protein calsenilin-like protein (CALP)/KChIP4 has also been identified (Morohashi et al. 2002).
Database analysis has shown that calsenilin is a member of the recoverin superfamily of neuronal calcium sensors. Calsenilin/DREAM/KChIP3 and its homologs contain a novel N-terminal extension that is not present in other members of the superfamily. Notably, although all four proteins in this family (calsenilin/DREAM/KChIP3, KChIP1, KChIP2 and KChIP4) display significant homology in their C-terminal regions, their N-termini do not (Zaidi et al. 2002). Calsenilin contains an N-terminus that can be alternatively spliced to produce shorter transcripts (Pruunsild and Timmusk 2005).
To learn more about the normal biological role of this N-terminal domain of calsenilin, we used the yeast two-hybrid system to investigate interactors that bind to this unique region. Our data indicate that the N-terminus of calsenilin interacts with the transcriptional co-repressor C-terminal binding protein 2 (CtBP)2. CtBPs were originally identified as cellular proteins that interact with a C-terminal motif of adenovirus E1A oncoprotein (Boyd et al. 1993; Schaeper et al. 1995; Sundqvist et al. 1998). They are conserved in higher eukaryotes and vertebrates contain two highly homologous CtBP proteins (CtBP1 and CtBP2). CtBP3/brefeldin A-ribosylated substrate (BARS), an N-terminally truncated form of human CtBP1, has been shown to play a role in Golgi fission (Hidalgo Carcedo et al. 2004). Interestingly, CtBP2 is identical to the B domain of RIBEYE, the major protein component of photoreceptor and bipolar cell ribbons in retina (Schmitz et al. 2000). RIBEYE also contains an A domain and is expressed specifically in synaptic ribbons. CtBP2 and RIBEYE are derived from alternative splicing of mRNA from the same gene. (Schmitz et al. 2000).
Data presented in this paper demonstrate that calsenilin interacts with CtBP2 as well as its homolog CtBP1. Calsenilin immunostaining in mouse cerebellum partially co-localizes with CtBP staining, suggesting that this interaction is feasible in vivo. In addition, we have found a calsenilin-dependent increase in c-fos protein levels in CtBP1/2 knockout fibroblasts compared with levels in heterozygous fibroblasts, suggesting that CtBP can modulate the repression of c-fos by calsenilin. The association of enzymatically active histone deacetylase (HDAC) with the calsenilin–CtBP complex in our study suggests that HDAC may be involved in modulating the transcriptional activity of this complex. Finally, we have demonstrated that calsenilin and CtBP are present in synaptic vesicles and that they can interact in vivo. Given that a number of other neuronal calcium sensor proteins such as neuronal calcium sensor-1 (NCS-1), frequenin and visinin-like protein (VILIP) have been implicated in neurotransmitter release (Taverna et al. 2002), and that RIBEYE is associated with synaptic ribbons, the presence of both CtBP and calsenilin in synaptic vesicle preparations raises the possibility that calsenilin and CtBP may act in concert outside of the nucleus to regulate neurotransmitter release from synaptic vesicles.
Materials and methods
Yeast two-hybrid system
A yeast transcription factor GAL4-based two-hybrid assay system (BD Biosciences Clontech, Palo Alto, CA, USA) was used in this study. Two baits were created by fusing the N-terminal 25 amino acid residues or the N-terminal 50 amino acid residues of calsenilin to the yeast GAL4 DNA-binding domain, as shown in Fig. 1. These fusion baits were tested for protein expression in yeast cells, and were found to express detectable levels of protein without activating transcription. The N-25 or N-50 bait was co-transformed with BD Matchmaker pretransformed human brain library cloned into the yeast GAL4 activation domain according to the manufacturer's protocol. Bait and library fusion protein interaction was detected by activation of yeast genes: ADE2, HIS3, MEL1. Positive clones were picked and library plasmids were rescued and co-transformed with the baits in a second interaction screen. After the verification of true positives, clones were sequenced and identified.
Cell culture and treatment
Naïve H4 neuroglioma cells, H4 neuroglioma cells stably expressing calsenilin and CtBP1/2 knockout mouse embryonic fibroblasts and heterozygous fibroblasts were grown in Dulbecco's modified Eagle's medium (Cambrex Biosciences Waterville Inc., Walkerville, MD, USA) containing 10% fetal bovine serum (Sigma-Aldrich Co., St Louis, MO, USA), 2 mm glutamine, 100 U/mL penicillin (Sigma-Aldrich), 100 µg/mL streptomycin (Sigma-Aldrich) and appropriate antibiotics for maintaining the selection, at 37°C in a 5% CO2 incubator. H4 human neuroglioma cells stably expressing calsenilin were treated with 5 mm caffeine for 24 h at 37°C.
Naïve H4 cells were transiently transfected with either wild-type calsenilin or EF2/3/4 hand mutant calsenilin constructs (Zaidi et al. 2004) and H4 cells stably expressing calsenilin were transiently transfected with KT3 epitope-tagged CtBP1 or CtBP2 using Effectene (Qiagen Sciences, Germantown, MD, USA) according to the manufacturer's protocol. CtBP1/2 knockout mouse embryonic fibroblasts and CtBP1/2 heterozygous fibroblasts (provided by Dr J. Hildebrand, University of Pittsburgh, Pittsburgh, PA, USA) were transfected by electroporation using MEF Nucleofector kit 1 (Amaxa Biosystems, Cologne, Germany). A plasmid encoding enhanced green fluorescent protein was used to determine the transfection efficiency.
Mouse brain homogenates, or lysates from H4 neuroglioma cell expressing calsenilin, were prepared in modified RIPA buffer (50 mm Tris, pH 7.4, 150 mm NaCl, 2 mm EDTA, 1% Triton X-100, 1% Nonidet P40, 0.25% sodium deoxycholate) or in CHAPS buffer (50 mm Tris, pH 7.4, 150 mm NaCl, 1% CHAPS) with protease inhibitor cocktail (CompleteTM; Roche Diagnostics, Mannheim, Germany). Mouse brain tissue and H4 neuroglioma cells transiently transfected with either wild-type or EF hand mutant calsenilin were homogenized in CHAPS buffer either in the presence of 1 mm calcium or 1 mm EGTA. In a separate experiment, total brain homogenates, crude synaptosomes or synaptic vesicles were solubilized in CHAPS buffer. Samples were rotated at 4°C for 1 h and then centrifuged at 14 000 g for 10 min. Supernatant corresponding to 1 mg total protein was incubated with calsenilin polyclonal antibody overnight at 4°C. A 50-μL aliquot of protein A magnetic beads (Qiagen, Hilden, Germany) was added to each sample and incubated for 2 h at 4°C. Immuocomplexes were washed three times each for 5 min with appropriate buffer and eluted with reducing sodium dodecyl sulfate (SDS) sample buffer.
SDS–polyacrylamide gel electrophoresis (PAGE) and western blotting
Cell lysates or immunocomplexes mixed with SDS–PAGE sample buffer were analyzed by SDS–PAGE on either a 14% or a 4–20% gradient Tris glycine gel (Invitrogen, Carlsbad, CA, USA). Samples were transferred to polyvinylidene difluoride membrane using a semidry transfer apparatus (Bio-Rad Laboratories, Hercules, CA, USA). Blots were blocked in 5% non-fat dry milk in TBST (25 mm Tris, pH 7.5, 135 mm NaCl, 0.15% Tween 20) for 2 h, followed by incubation with the appropriate antibody at 4°C for 16 h. Antibodies used for western blotting included calsenilin monoclonal (characterized by Zaidi et al. 2002), CtBP monoclonal (Santa Cruz Biotechnology, Santa Cruz, CA, USA), KT3-tag monoclonal (Covance Research Products Inc., Denver, PA, USA), c-fos polyclonal (eBioscience, San Diego, CA, USA), β-actin monoclonal (Sigma-Aldrich), nucleoporin p62 monoclonal (BD Transduction Laboratories, Lexington, KY, USA), synaptophysin polyclonal (Chemicon International Inc., Temecula, CA, USA) and a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) monoclonal (Abcam Inc., Cambridge, MA, USA). Blots were washed three times each for 15 min with TBST and incubated with an appropriate secondary antibody conjugated with horseradish peroxidase (1 : 5000) at room temperature for 1 h. Membranes were washed three times each for 15 min with TBST and developed using an enhanced chemiluminescence detection system (Amersham Biosciences UK Ltd, Little Chalfont, UK) according to the manufacturer's protocol.
Synaptic vesicle preparation
Synaptic vesicles were prepared following a protocol described by Lau et al. (1996) and Blackstone et al. (1992) with minimal modifications. All reagents contained protease inhibitor cocktail and the procedure was performed at 4°C. In brief, mouse brain was homogenized in 10 volumes HEPES-buffered sucrose (0.32 m sucrose, 4 mm HEPES, pH 7.4). The homogenate was centrifuged at 800 g for 10 min to remove the nuclear fraction. The supernatant was then spun at 9000 g for 15 min to yield the crude synaptosomal fraction. This pellet was resuspended in 10 volumes of HEPES-buffered sucrose and then respun at 10 200 g for 15 min. The resulting pellet was lysed by hypo-osmotic shock in water, rapidly adjusted to 1 mm HEPES, and mixed constantly for 30 min. The lysate was centrifuged at 25 000 g for 20 min; the resulting supernatant contained crude synaptic vesicle fraction. The supernatant was then spun at 165 000 g for 2 h to retrieve synaptic vesicles. The resulting pellet contained synaptic vesicles and was suspended in 50 mm HEPES, pH 7.4, 2 mm EDTA with protease inhibitors.
Extraction of synaptic vesicle-associated calsenilin
Synaptic vesicles were treated with urea as described earlier (Zaidi et al. 2002). In brief, the vesicle preparations were incubated with 8 m urea with gentle rocking at 4°C for 1 h and then spun at 165 000 g for 1 h. Supernatants and pellets were collected and analyzed by SDS–PAGE along with untreated vesicles.
Measurement of HDAC activity
HDAC activity was assayed using a colorimetric HDAC activity assay kit (Biovision Research Products, Mountain View, CA, USA) according to the manufacturer's protocol. In this assay, HDAC in the test samples deacetylates an acetylated lysine side-chain substrate that can be measured following treatment with a lysine developer, which produces a chromophore that can be analyzed spectrophotometrically at 400 nm.
Immunocytochemistry and confocal microscopy
Human H4 neuroglioma cells stably expressing calsenilin were plated in chambered slides, and the next day cells were treated with 5 mm caffeine for 24 h. Both control and caffeine-treated cells were fixed with 4% paraformaldehyde, permeabilized using detergent Nonidet P40 (Roche Diagnostics), washed with phosphate-buffered saline and then incubated with a polyclonal calsenilin antibody (1 : 500) and a CtBP monoclonal antibody (1 : 500) at 4°C for 16 h. Calsenilin immunofluorescence was visualized using goat anti-rabbit cy3 conjugate (1 : 500; Jackson Immuno Research Laboratories Inc., West Grove, PA, USA), whereas CtBP immunoreactivity was detected by Bodipy anti-mouse (1 : 200; Molecular Probes, Eugene, OR, USA). Paraformaldehyde-fixed mouse brain sections were stained with a polyclonal calsenilin antibody and a monoclonal CtBP antibody, and processed in the same manner. Immunostaining was analyzed using confocal microscopy.
CtBP2 interacts with the N-terminus of calsenilin in a yeast two-hybrid assay
The N-terminus of calsenilin shows no significant homology to the N-terminus of other members of the calsenilin/KChIP3/DREAM family. To identify proteins that might provide clues about the function of this unique domain, the N-terminus of calsenilin was used in a GAL4-based yeast two-hybrid screen. A pretransformed BD Matchmaker adult human brain library was screened with two N-terminal calsenilin baits (N-25 and N-50), as shown in Fig. 1. Before the library screening, the baits were transformed in yeast host cells and were found to express the calsenilin fusion proteins. Neither bait was toxic to the host yeast strain nor did they show significant autonomous transcriptional activation of the reporter gene lacZ (data not shown). After determining the transformation efficiency, 1.5 × 107 transformants were screened and 21 positive clones were identified using stringent selection conditions in the presence of X-α-Gal to assay for α-galactosidase. False positives were eliminated and clones that were reconfirmed in a second screen were rescued, retested and sequenced. Clone 18 was chosen for further characterization because it was picked at a higher frequency than other positives, it interacted with the N-50 bait in a specific manner, and it did not interact non-specifically with control vector (Fig. 2). Sequence analysis identified this clone as human CtBP2. CtBP2 and its homolog CtBP1 have been shown to act as co-repressors for a variety of transcriptional repressors including Kruppel (Turner and Crossley 1998), Net (Criqui-Filipe et al. 1999), ZEB (Postigo and Dean 1999), Ikaros (Koipally and Georgopoulos 2000) and Evi-1 (Izutsu et al. 2001).
Endogenous calsenilin forms a complex with endogenous CtBP(s) in vivo independent of calcium
To assess whether an interaction between calsenilin and CtBPs could be detected at endogenous levels, we carried out co-immunoprecipitation analysis from mouse brain extracts as well as H4 neuroglioma cells stably expressing calsenilin using a polyclonal calsenilin antibody. Western blots were probed with either a calsenilin CtBP monoclonal antibody or a monoclonal antibody that recognizes both CtBP1 and CtBP2. The immunoprecipitation of calsenilin by the calsenilin antibody resulted in co-immunoprecipitation of endogenous CtBP1 and/or CtBP2 in both cases (Fig. 3a), suggesting that calsenilin and CtBP(s) interact physiologically and at endogenous levels. The interaction of calsenilin with CtBP was found to be independent of calcium as co-immunoprecipitation of CtBP along with calsenilin was observed in the presence of calcium or EGTA. Furthermore, a calsenilin EF2/3/4 hand mutant, which is incapable of binding calcium, also interacted with CtBP (Fig. 3b).
Calsenilin interacts with both CtBP1 and CtBP2
Human H4 neuroglioma cells express high levels of endogenous CtBP1 and CtBP2. Therefore, to determine whether calsenilin could interact with CtBP1 as well as CtBP2, we transiently transfected KT3 epitope-tagged CtBP1 or CtBP2 in H4 neuroglioma cells expressing calsenilin and subjected extracts to immunoprecipitation using a calsenilin polyclonal antibody. Blots were probed with a KT3-tag monoclonal antibody to determine whether tagged CtBP1 or CtBP2 could be detected after immunoprecipitation. As shown in Fig. 4, the calsenilin antibody immunoprecipitated calsenilin along with either tagged CtBP1 or CtBP2, demonstrating that calsenilin can interact with both CtBP1 and CtBP2.
Calsenilin and CtBP immunostaining partially co-localize in mouse cerebellum
A number of studies, including our own, have described the localization of calsenilin/DREAM in mouse brain (Zaidi et al. 2002; Hammond et al. 2003; Lilliehook et al. 2003), and it has been found to be present in the cerebellum, hippocampus, cortex, olfactory bulb and a variety of other brain regions. To date there is little information available about the cellular localization of CtBP in the brain. To determine whether calsenilin and CtBP are expressed in the same cells we chose to look in cerebellum, a brain region with high levels of calsenilin. Co-immunostaining in mouse brain sections using calsenilin and CtBP antibodies showed that, as expected in the cerebellum, calsenilin was predominantly expressed in the granular cell layer. Although CtBP staining was most prominent in the molecular layer with no staining in the Purkinje cell layer (Fig. 5), it was also expressed in the granular cell layer. A partial co-localization of calsenilin and CtBP was observed in the granular cell layer.
Calsenilin co-localizes with CtBP after nuclear translocation
At steady state in cultured H4 neuroglioma cells stably expressing calsenilin, calsenilin was primarily expressed in the cytoplasm whereas CtBPs were primarily expressed in the nucleus (Fig. 6, upper panel). However, because calsenilin can act as a transcriptional modulator, it has the ability to translocate to the nucleus. Our previous studies demonstrated that treatment with agents that increase intracellular calcium levels, such as caffeine, results in the translocation of calsenilin from the cytoplasm to the nucleus (Zaidi et al. 2004). To determine whether calsenilin and CtBP co-localize in the nucleus, cells were treated with 5 mm caffeine and processed for immunostaining with antibodies to calsenilin and CtBP. As shown in Fig. 6 (lower panel), following treatment with caffeine, a partial co-localization of calsenilin and CtBP was observed in the nucleus. This suggests that, following translocation to the nucleus, calsenilin can interact with CtBPs in this compartment.
CtBP modulates the transcriptional repression activity of calsenilin
The endogenous levels of CtBPs in cultured cell lines are high, making an evaluation of the ability of CtBP to modulate the transcriptional repression activity of calsenilin difficult. To circumvent this problem, embryonic fibroblasts from CtBP1/2 knockout or heterozygous mice were transiently transfected with calsenilin, and the levels of c-fos protein, a target gene for repression by calsenilin, were assessed in the presence and absence of CtBP1/2. A small but significant increase (20% increase; p < 0.001, Student's t-test) in c-fos levels was observed in the CtBP1/2 knockout fibroblasts compared with levels in heterozygous fibroblasts (Fig. 7). This result suggests that transcriptional repression of the c-fos gene by calsenilin may be modulated by its interaction with the co-repressor CtBP. Levels of GAPDH, an unrelated protein, remained unchanged in the presence and absence of CtBP, showing that the effect of CtBP on calsenilin-mediated c-fos transcription is a specific event and is not due to a global increase in protein levels in CtBP null fibroblasts.
The role of calsenilin/DREAM in transcription repression of the prodynorphin gene has been studied in great detail in recent years (Carrion et al. 1999; Ledo et al. 2000). Unfortunately, the levels of endogenous prodynorphin in the mouse embryonic fibroblasts lines used in our study were below the detectable limit, thereby making it difficult to study endogenous prodynorphin in CtBP heterozygous and knockout fibroblasts. Other studies also failed to observe detectable levels of prodynorphin mRNA in rat fibroblasts (Rosen et al. 1990).
Enzymatically active HDAC is associated with calsenilin–CtBP complex
CtBPs have been shown to be associated with HDACs and the finding that this HDAC–CtBP complex contains significant HDAC activity (Subramanian and Chinnadurai 2003) suggests that this association is physiologically relevant. To determine whether calsenilin could associate with the HDAC–CtBP complex, we examined whether calsenilin antibody could precipitate this complex. Our co-immunoprecipitation studies using H4 neuroglioma cells stably expressing calsenilin showed that the antibody precipitated calsenilin along with CtBP and HDAC. Similar experiments carried out using naïve H4 neuroglioma cells containing very low levels of endogenous calsenilin also co-immunoprecipitated HDAC, but we detected significantly more HDAC protein (2.9-fold, as determined by quantitation of western blots) in the presence of overexpressed calsenilin (Fig. 8a). We also detected higher HDAC activity (39% increase; p < 0.028, Student's t-test) in the immunocomplex obtained from H4 neuroglioma cells stably expressing calsenilin compared with that from naïve H4 neuroglioma cells (Fig. 8b). This indicates that calsenilin, CtBP and HDAC can exist as a complex in vivo, that the HDAC associated with this complex is functionally active, and that this complex and the activity associated with HDAC could represent part of the mechanism by which calsenilin represses transcription.
CtBP and calsenilin are associated with synaptic vesicles
RIBEYE, an alternatively spliced product of the CtBP2 locus, is expressed in ribbon synapses in the retina (Schmitz et al. 2000), where it is believed to play a structural role and may be involved in tethering synaptic vesicles to the ribbon and/or in preparing vesicles for exocytosis. Interestingly, recent reports have shown that CtBP1 is also present at conventional synapses (tom Dieck et al. 2005). Taken together, these findings suggest a new role for the CtBP family in the molecular assembly and function of CNS synaptic transmission (tom Dieck et al. 2005). Because we found that CtBP1 and CtBP2 interact with calsenilin in a yeast two-hybrid system as well as in co-immunoprecipitation experiments, we examined whether, like the CtBPs, calsenilin is also associated with synaptic vesicles. As can be seen in Fig. 9(a), western blot analysis of synaptic vesicle preparations indicated that a small percentage of calsenilin was present in these preparations. Western blots using a nucleoporin p62 antibody demonstrated that the synaptic vesicle preparations did not contain this nuclear protein and were therefore unlikely to be contaminated with other cellular organelles.
Our previous studies (Zaidi et al. 2002) have shown that calsenilin is very tightly associated with intracellular membranes. Treatments such as high salt (1 m NaCl) or high pH (100 mm sodium bicarbonate, pH 11) resulted in only partial removal of calsenilin from the membrane fraction. In contrast, treatment with 8 m urea dissociated most of the calsenilin from the intracellular membranes. Therefore, to study the nature of the association of calsenilin with synaptic vesicle membranes, synaptic vesicle preparations were treated with 8 m urea. Synaptic membrane-associated calsenilin appeared to be very tightly bound as it was found to be resistant to urea treatment, whereas CtBP was not as tightly associated and could be easily dissociated by urea treatment (Fig. 9b).
CtBP and calsenilin can interact at conventional synapses
Because both calsenilin and CtBP were detected in our synaptosomal and synaptic vesicle preparations, we looked to see whether these protein can interact in vivo in these samples. Co-immunoprecipitation in CHAPs-solubilized synaptosomes and synaptic vesicles showed that calsenilin and CtBP could be co-immunoprecipitated from these preparations (Fig. 9c). These findings, which localize both calsenilin and CtBP to synaptic vesicles, suggest that outside of the nucleus the calsenilin–CtBP interaction may play a role in the maintenance, release or trafficking of synaptic vesicles.
In this study, we describe CtBP(s) as novel interactors for calsenilin/DREAM/KChIP3. Co-immunoprecipitation from cell lines as well as mouse brain extracts was used to demonstrate that these proteins interact at endogenous levels in a calcium-independent manner. Data presented here also indicate that calsenilin partially co-localizes with the CtBPs in mouse cerebellum, suggesting that this interaction is feasible in vivo. The CtBPs have been known for quite some time to act as transcription co-repressors, but their interaction with calsenilin or other members of the neuronal calcium sensor family of proteins has not been reported. To date, two highly related homologs, CtBP1 and CtBP2, have been described in vertebrates and a single CtBP has been described in Drosophila. An N-terminally truncated form of human CtBP1, called CtBP3/BARS, has been shown to play a role in Golgi fission (Hidalgo Carcedo et al. 2004). CtBP shows significant sequence homology with NADH-dependent 2-hydroxyacid dehydrogenase and has low levels of dehydrogenase activity. CtBPs are known to act as co-repressors for a variety of transcription repressors in both vertebrates and Drosophila, and have been shown to modulate the expression of Hairy, Knirps, Kruppel and Snail in Drosophila (Poortinga et al. 1998; Turner and Crossley 1998; Keller et al. 2000) and CtBP-interacting protein, BKLF, FOG, NET, ZEB, Evi-1 and Ikaros in vertebrates (Turner and Crossley 1998; Criqui-Filipe et al. 1999; Postigo and Dean 1999; Koipally and Georgopoulos 2000; van Vliet et al. 2000; Izutsu et al. 2001; Katz et al. 2002). CtBP recognizes a consensus motif PXDLS (where X is any amino acid residue) and, although many DNA-binding factors that recruit CtBP contain related motifs, not all of the proteins that bind to CtBP contain such a consensus motif. For instance, FHL3, a member of four and a half LIM domain containing protein, binds to CtBP2 but does not contain a PXLDS motif (Turner et al. 2003). In some cases these proteins may contain a motif that is structurally similar but not easily identifiable based on its primary amino acid sequence. Calsenilin does not have a PXDLS motif and to date it is not clear which amino acid residues are involved in the interaction with CtBP.
Given that calsenilin and CtBP have the ability to interact at endogenous levels, we hypothesize that in the nucleus CtBP binds calsenilin and functions as a co-repressor by modulating transcriptional repression by calsenilin of target genes such as those encoding c-fos and prodynorphin. Our data, generated using CtBP1 and CtBP2 knockout mouse embryonic fibroblasts (Hildebrand and Soriano 2002) demonstrate that a calsenilin-dependent increase in the level of c-fos protein is observed in CtBP knockout fibroblasts compared with heterozygous fibroblasts (Fig. 7). This finding suggests that, in the absence of cellular CtBP, transcriptional repression of c-fos by calsenilin is not as efficient, and that this may lead to an increase in c-fos levels.
One mechanism by which CtBP could mediate its transcriptional co-repressor activity is by interaction with proteins such as HDACs (Sundqvist et al. 1998; Shi et al. 2003; Subramanian and Chinnadurai 2003). HDACs are thought to regulate transcriptional repression by deacetylating histones and increasing the positive charge, thereby causing compaction of chromatin which, in turn, prevents transcription factors from accessing DNA. It has been shown that CtBP1 associates with class I HDACs (HDAC-1, HDAC-2 and HDAC-3). Some of these HDACs also interact with the Drosophila CtBP homolog, dCtBP. The CtBP protein complex exhibits significant HDAC activity in vitro, suggesting that association of CtBP with HDACs may be functionally relevant (Subramanian and Chinnadurai 2003). However, CtBP can also repress transcription by associating with other histone modifiers such as histone acetyltransferases (Kim et al. 2005). Our data, which show that calsenilin can co-immunoprecipitate CtBP along with HDACs, indicates that these proteins can form a complex in vivo. Because this calsenilin–CtBP–HDAC complex has significant HDAC activity, it may be relevant functionally and it may be associated with the mechanism by which calsenilin acts as a transcriptional repressor.
RIBEYE is an alternatively spliced product of the CtBP2 locus that is expressed in ribbon synapses in retina (Schmitz et al. 2000). CtBP1 has been shown to be present at conventional synapses, and it has been suggested that CtBP proteins may play a role in the molecular mechanisms associated with the assembly and function of CNS synaptic transmission (tom Dieck et al. 2005). Because CtBP1/2 can interact with calsenilin in a yeast two-hybrid system as well as in co-immunoprecipitation experiments, we hypothesized that calsenilin might associate with CtBP in synaptic vesicles. Our western blot analysis showed that calsenilin and CtBPs are in fact associated with synaptic vesicle preparations along with the presynaptic terminal marker synaptophysin, and that calsenilin is more tightly associated with the synaptic membranes than CtBP. Furthermore, our co-immunoprecipitation studies indicated that calsenilin and CtBP can interact in vivo in our synaptic vesicle preparation.
Recent studies have suggested that the RIBEYE–CtBP interaction may be involved in preparing synaptic vesicles for exocytosis and that these proteins may function as lysophosphatidic-acyl CoA transferase and thus modulate the curvature of lipid membranes (Kooijman et al. 2003). Interestingly, CtBP1 has been implicated in membrane fission processes at the Golgi complex, where it is involved in a fission reaction that generates aligned vesicles from continuous tubular structures (Weigert et al. 1999). Other members of the recoverin superfamily of calcium-binding proteins, such as NCS-1, frequenin and VILIP, have been shown to be associated with synaptic vesicles and have been suggested to have a role in neurotransmitter release (Taverna et al. 2002). These previous findings, taken together with our current data demonstrating that calsenilin and CtBPs are interacting proteins and that both proteins are present in synaptic vesicle preparations, suggest that calsenilin and CtBP may act in concert to regulate neurotransmitter release from the vesicles. Future studies will address this hypothesis.
We thank Dr J. Hildebrand, University of Pittsburgh, for his generous gift of CtBP constructs and CtBP1/2 knockout and heterozygous embryonic fibroblasts. This work was supported by grants from the National Institute of Aging (RO1-AG016361 to WW and RO1-AG021792 to JDB).