Molecular Cloning and Characterization of CALP/KChIP4, a Novel EF-hand Protein Interacting with Presenilin 2 and Voltage-gated Potassium Channel Subunit Kv4*

Presenilin (PS) genes linked to early-onset familial Alzheimer's disease encode polytopic membrane proteins that are presumed to constitute the catalytic subunit of γ-secretase, forming a high molecular weight complex with other proteins. During our attempts to identify binding partners of PS2, we cloned CALP (calsenilin-like protein)/KChIP4, a novel member of calsenilin/KChIP protein family that interacts with the C-terminal region of PS. Upon co-expression in cultured cells, CALP was directly bound to and co-localized with PS2 in endoplasmic reticulum. Overexpression of CALP did not affect the metabolism or stability of PS complex, and γ-cleavage of βAPP or Notch site 3 cleavage was not altered. However, co-expression of CALP and a voltage-gated potassium channel subunit Kv4.2 reconstituted the features of A-type K+ currents and CALP directly bound Kv4.2, indicating that CALP functions as KChIPs that are known as components of native Kv4 channel complex. Taken together, CALP/KChIP4 is a novel EF-hand protein interacting with PS as well as with Kv4 that may modulate functions of a subset of membrane proteins in brain.

inherited as an autosomal dominant trait, and mutations in three different genes have thus far been linked to early-onset autosomal dominant forms of familial AD (FAD). Among these, presenilin 1 (PS1) and PS2 account for the majority of the early onset FAD (1). PS1 and PS2 genes encode polytopic integral membrane proteins that are predominantly localized in intracellular membranes and span the membrane six to eight times.
PS proteins undergo endoproteolysis to give rise to N-and C-terminal fragments, which are the preponderant forms of endogenous PS in vivo (2). These fragments form a heterodimer and are incorporated into high molecular weight (HMW) protein complexes (2)(3)(4)(5) that are highly stabilized (t1 ⁄2 ϭ ϳ20 h; Ref. 6), whereas holoproteins of PS are rapidly degraded (t1 ⁄2 ϭ ϳ2 h) (6,7). The steady-state levels of PS fragments seem to be tightly regulated by competition for shared, but limiting, cellular factors, because overexpression of PS in transfected cells does not increase the overall level of PS fragments and replaces endogenous PS (8).
PS plays an important role in the generation of amyloid ␤ peptides (A␤) by facilitating intramembranous ␥-cleavage of ␤-amyloid protein precursor (␤APP), as evidenced by the lack of A␤ production and accumulation of ␤APP C-terminal stubs in cells established from PS-null mice (9 -11). In contrast, FADlinked mutations in PS increase the production of highly fibrillogenic A␤42 (12)(13)(14)(15), which is the initial and predominantly deposited A␤ species in AD brains (16,17) and normally consists of only ϳ10% of total secreted A␤ (18). Moreover, genetic studies in invertebrates and PS-null mice suggested that ␥-cleavage-like proteolytic cleavage at site 3 to release Notch intracellular domain (NICD), which is the prerequisite for Notch signaling (reviewed in Ref. 19), also is facilitated by PS. Furthermore, recent findings that the two intramembranous aspartates within the 6th and 7th transmembrane (TM) domains of PS are required for ␥-secretase activities (20) and that transition state analogue ␥-secretase inhibitors specifically label PS fragments (21)(22)(23)(24) strongly support the notion that the PS-containing macroprotein complex catalyzes ␥-cleavage and that PS may represent the catalytic subunit of ␥-secretase complex. Very recently, ErbB4, a type I single span membrane protein functioning as a tyrosine kinase also was found to be cleaved by ␥-secretase (25,26). These findings suggest that one of the primary functional activities of PS/␥-secretase lies in the intramembranous cleavage of a subset of type I membrane proteins, whereas other functions including regulation of ion fluxes (see Ref. 27 and see "Discussion") or interaction with cytoplasmic (28) or membrane-bound (29) forms of ␤-catenin, which apparently do not involve proteolytic activities, also are implicated.
We have shown previously (30,31) that the C-terminal cytoplasmic region of PS plays an important role in the stabilization and HMW complex formation of PS, which are required for the ␥-secretase activity. The mechanistic role of the C terminus of PS in its metabolism and function still remains unknown, but one possibility is that this portion serves as the binding site for the interacting proteins that regulate the metabolism and functions of PS (6,8). In this paper, we used the yeast two-hybrid system to search for proteins that interact with the C-terminal region of PS2, and we identified a novel PS-binding protein CALP/KChIP4 (calsenilin-like protein) that belongs to the calsenilin/KChIP protein family harboring four EF-hand motifs (32). CALP/KChIP4 did not affect the stability or HMW complex formation of PS, nor did it alter ␥-cleavage of ␤APP or site 3 cleavage of Notch. However, it exhibited a unique character to alter the voltage-gating and inactivation properties of voltage-gated potassium channel subunit Kv4 as observed with other KChIPs.

MATERIALS AND METHODS
Yeast Two-hybrid cDNA Screening-MATCHMAKER LexA Two-hybrid System (CLONTECH, Palo Alto, CA) was used according to the manufacturer's instructions. A cDNA encoding the C-terminal 43 amino acid residues of PS2 (amino acids 406 -448) was subcloned into pLexA as a bait. The bait plasmid and the lacZ reporter plasmid, p8op-lacZ, were transformed into the budding yeast Saccharomyces cerevisiae strain EGY48, which contains a genomic LEU2 reporter gene. Then human brain cDNA library in pB42AD (CLONTECH) was transformed using the lithium acetate method. Transformants (4.8 ϫ 10 9 clones) were selected on Gal/Raf/ϪHis/ϪUra/ϪTrp/ϪLeu plates, and positive clones were chosen after 7-10 days of culture at 30°C. The colony-lift ␤-galactosidase filter assay was performed to exclude false positives. The individual library plasmids of positive transformants were recovered in Escherichia coli strain KC8 and selected based on the digestion pattern by HindIII. Protein interaction was further confirmed by retransformation of the isolated prey plasmid into EGY48 containing pLexA/PS2 C terminus. The confirmed cDNAs were sequenced using Thermosequenase (Amersham Biosciences) on an automated sequencer (Li-Cor, Lincoln, NE).
Cloning of cDNAs Coding for Human and Mouse CALP-Alternative splice variant of human CALP cDNA was cloned by screening the human brain cDNA library (ZAP II, Stratagene, La Jolla, CA) by the plaque hybridization method using the human CALP cDNA obtained by the two-hybrid system as a probe. To identify a sequence including the transcription start site, we amplified cDNAs from the Cap Site cDNA dT (Nippon Gene Co., Ltd., Tokyo) of human brain by nested PCR ("CapSite Hunting"). The following sets of primers were used: 5Ј-GAT-GCTAGCTGCGAGTCAAGTC-3Ј (1RDT) and 5Ј-CGAGTCAAGTCGAC-GAAGTGC-3Ј (2RDT) as "anchor" forward primers; 5Ј-GGTTTCTTCA-TTAACAACACCACTGG-3Ј (moro-1) and 5Ј-TGACGGTGGCCATCTC-CAGTTCATCTT-3Ј (31-3out) as "gene-specific" reverse primers. The reaction solution contained the following reagents: 10 mM Tris-HCl (pH 8.8), 2 mM MgSO 4 , 10 mM KCl, 10 mM (NH 4 ) 2 SO 4 , 400 M dATP, dCTP, dGTP, dTTP, 1 l of Cap Site cDNA dT from human brain (Nippon Gene), 1.25 units PfuTurbo polymerase (Stratagene), 0.5 M each primers (e.g. 1RDT and moro-1 for 1st PCR and 2RDT and 31-3out for 2nd "nested" PCR), 5 l of GC-Melt (CLONTECH). The first PCR was performed for 35 cycles, with each cycle consisting of denaturation for 60 s at 94°C, annealing for 30 s at 60°C, and extension for 90 s at 72°C. The second PCR was performed in the same buffer, using 1 l of the first PCR products as a template and the same program as the 1st PCR. Specific PCR products were subcloned into pCR-Script Amp SK(ϩ) (Stratagene) and subjected to sequencing. To obtain a mouse CALP cDNA, we performed 5Ј-RACE or 3Ј-RACE using mouse brain Marathon-Ready cDNA library (CLONTECH) and CALP-specific primers. The isolated PCR fragments were subcloned and sequenced.
Northern Blot Analysis-Northern hybridization analysis was carried out using a 0.9-kb human CALP cDNA fragment as a hybridization probe, which was hybridized to Northern Lights TM Human Multiple mRNA Blot (Invitrogen, San Diego, CA). The blots were then stripped and reprobed with human ␤-actin cDNA as an internal control.
Cell Culture, Transfection, and Cycloheximide Treatment-Monkey COS-1 cells, mouse neuro2A (N2a), or human HEK293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and penicillin/streptomycin at 37°C in 5% CO 2 atmosphere as described (15,30,31). Transient and stable transfection in COS-1 or N2a cells was performed using LipofectAMINE reagent (Invitrogen). N2a cell lines stably expressing wt or mt PS2 were generated as described (15). N2a cell lines stably co-expressing PS2 and CALP were generated by transfecting the cDNAs encoding fl-or EFmt-CALP in pcDNA3.1/hyg(ϩ) using LipofectAMINE and selection in Dulbecco's modified Eagle's medium containing 300 g/ml G418 and 200 g/ml hygromycin (38). The expression of transgenes was driven by addition of 10 mM butyric acid for 12-24 h. For electrophysiological experiments, transient co-expression of Kv4.2 with CALP or its derivatives in HEK293 cells was performed by the calcium-phosphate method and used after 48 -72 h of cultivation (34). To evaluate the stability of PS or CALP proteins by blocking total cellular protein synthesis, cultured cells were treated with cycloheximide (30 g/ml) for the indicated times and then analyzed by immunoblotting.
Antibodies, Immunoprecipitation, Immunoblot Analysis, and Immunofluorescence Microscopy-The following rabbit and mouse polyclonal antibodies were used: anti-G2N4 against GST fused to 2-59 of human PS2, anti-G2L and mPS2 against GST fused to amino acids 301-361 of human PS2, anti-G1Nr2 and anti-G1L3 against GST fused to amino acids 2-70 and 297-379 of human PS1, respectively (30,31), anti-CALP2 against GST fused to 1-216 of human CALP 216 and anti-N-CALP against a synthetic peptide corresponding to residues 4 -21 of human CALP 216 conjugated to keyhole limpet hemocyanin at the C terminus. Anti-rat Kv4.2 rabbit polyclonal antibody was purchased from Chemicon. A mouse anti-c-Myc monoclonal antibody (9E10) was purchased from Roche Diagnostics. For immunoprecipitation, cells were lysed by TSCC (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% CHAPSO, Complete protease inhibitor mixture (Roche Diagnostics)) and passed through a 27-gauge needle. The solubilized samples were pre-cleared with protein G-conjugated agarose (Invitrogen) for 1 h at 4°C, reacted with antibodies overnight, followed by incubation with protein G-agarose for 2 h at 4°C and wash with TSC (TSCC without CHAPSO). Immunoprecipitates were collected, heated to 37°C for 30 min, and analyzed by immunoblotting. For immunoblot analysis, cells were lysed in 2% SDS sample buffer and briefly sonicated. The samples were separated by SDS-PAGE without previous boiling, transferred to polyvinylidene difluoride membrane, and analyzed by immunoblotting as described (15,30). For immunofluorescence microscopy, transiently transfected COS-1 cells were cultured on glass coverslips, fixed, immunostained, and viewed with a confocal laser scanning microscope (FLUOVIEW, Olympus, Tokyo) as described (38,39), except that secondary antibodies conjugated with Alexa Fluor 488 or 594 (Molecular Probes, Eugene, OR) were used.

Quantitation of A␤ by Two-site ELISAs-Two-site
ELISAs that specifically detect the C terminus of A␤ were used as described (15, 30, 31, 38 -40). BNT77, which was raised against human A␤-(11-28) and recognizes full-length as well as N-terminally truncated A␤ of human and rodent types (40), was used as a capture antibody. BA27 and BC05 that specifically recognize the C terminus of A␤40 and A␤42, respectively, were conjugated with horseradish peroxidase and used as detector antibodies. Culture media were collected after incubation of 24 h and subjected to BNT77/BA27 or BNT77/BC05 ELISAs.
Electrophysiology-Whole-cell voltage clamp was applied to single HEK293 cells with patch pipettes using a CEZ-2400 (Nihon Kohden, Tokyo) amplifier as reported previously (33). A type K ϩ current (I A ) was observed in neither native nor GFP-alone transfected HEK293 cells. GFP signals were detected by use of GFP longpass filter (Nikon, Tokyo). All experiments were done at room temperature (23 Ϯ 1°C). Membrane currents were monitored and stored as reported previously (33). Cell capacitance was measured from the integration of capacitive transient  LexA yeast two-hybrid system Amino acid sequences corresponding to various portions of C-terminal domain of PS2 or PS1 used as baits are shown in the 1st column. Full-length CALP was used as a prey, and an empty vector pB42AD was used as a blank (middle column). p53 and SV40-large T antigen were used as a positive control. Results of ␤-galactosidase (␤-Gal) assay (white, negative; blue ϩϩϩ, strongly positive; blue ϩ, weakly positive) are shown in the last column.

FIG. 2. Expression and protein-protein interaction of CALP and PS2 in cultured cells.
A, schematic depiction of CALP and its derivatives used in this study. The names of cDNAs are indicated at the right of each scheme. The CALP 216 polypeptide contains an N-terminal "variable" domain (that is distinct between KChIP members; box), 1st EF-hand domain (that does not bind calcium; lightly shaded box), and the other three EF-hand domains (heavily shaded boxes). Location of immunogen peptide/protein for antibodies used in this study (␣CALP2 and ␣N-CALP) is shown by solid bars below the scheme. Amino acid substitutions at EF-hand domains are shown by an asterisk, respectively. B, 45 Ca The voltage dependence of I A activation was measured using the conventional double pulse protocol as mentioned previously (41). The membrane potential was changed from Ϫ80 to test potentials for 10 ms to activate I A and then to Ϫ40 mV to measure the tail current. The tail current amplitude was normalized with the maximum in each cell and was plotted against the test potentials. The data were fitted with Boltzmann equation, and the voltage required for the half-maximal activation and the slope factor were determined from the fitting. The double pulse was applied once every 15 s. The voltage dependence of I A inactivation was also determined by the double pulse protocol (41). I A was activated and inactivated by depolarization from Ϫ80 mV to test potentials for 1 s, and then remaining available channels were activated by the following depolarization to ϩ40 mV. The current normalization and the fitting of data with the Boltzmann equation were performed in a similar manner as those for the activation. The double pulse was applied every 30 s.
Statistics-Pooled data were expressed as mean Ϯ S.E., and statistical significance was examined using the unpaired Student's t or Scheffe's test for two or multiple groups, respectively. In the figures, * and ** indicate statistical significance at p values of 0.05 and 0.01, respectively.

Cloning of CALP/KChIP4, a Novel Interacting Protein with the C Terminus of Presenilins-To identify proteins that inter-
act with the C terminus of PS, we screened the human brain cDNA library by the yeast two-hybrid system using amino acid residues 406 -448 of PS2 as bait and obtained Ͼ100 positive clones. We found that ϳ90% of isolated clones in our screen encoded the same polypeptide. BLAST homology analysis revealed that this clone encoded a novel polypeptide showing some homology to calsenilin, which has previously been cloned as a PS C-terminal binding protein (35), as well as to KChIPs that have been identified as components of native Kv4 channel complex (32). We therefore designated this gene as CALP (calsenilin-like protein)/KChIP4 (Fig. 1A). Multiple sequence alignment and BLAST analysis indicated that CALP had high amino acid identity at its C-terminal region with KChIP2 (79.6%) and calsenilin/KChIP3 (77.6%), whereas its N terminus was very divergent from any KChIPs. CALP cDNA derived from our two-hybrid screening contained an in-frame methionine but lacked an apparent Kozak sequence. To isolate the entire CALP open reading frame, we performed plaque hybridization and 5Ј-RACE using additional human cDNA libraries, and we cloned an alternatively spliced form of CALP harboring an N-terminal insert sequence. However, we failed to detect an upstream sequence as well as an in-frame termination codon even when we used a Cap site cDNA dT library, which is suitable for the determination of transcription start site (42). We next cloned a cDNA encoding mouse CALP from a mouse brain cDNA library by the 5Ј-and 3Ј-RACE method, and we found an in-frame termination codon located upstream of the first ATG codon, the latter being in a similar position to the first ATG codon in human CALP cDNA. Thus, we concluded that the human cDNA cloned by two-hybrid system encompassed the entire CALP open reading frame encoding a 216amino acid polypeptide (CALP 216 ); human CALP gene also encoded an alternatively spliced form encoding a 250-amino acid protein (CALP 250 ) with an N-terminal insert, and all mouse CALP cDNA we cloned harbored this insert corresponding to CALP 250 (Fig. 1A). Northern blot analysis of mRNA derived from human tissues revealed that CALP is predominantly expressed in brain (Fig. 1B). Similar results were obtained in Northern blots of mRNA derived from mouse tissues (data not shown). BLAST search of human genome data from the International Human Genome Project of the National Institutes of Health located the CALP gene on Homo sapiens chromosome 4 working draft sequence segment (GenBank TM accession number NT_006138, Locus ID, 80333).
PS2 Interacts with the C-terminal EF-hand Domain of CALP Independent of Calcium Binding-We confirmed the interaction of CALP and PS2 by yeast two-hybrid assay using truncated C-terminal fragments of PS2 as baits (Table I). ␤-Galactosidase filter assay revealed that C-terminal fragments of PS2 corresponding to residues 406 -448 or 406 -441 bind CALP, whereas those corresponding to residues 406 -431 or 406 -421 showed very weak or no reaction, suggesting that a minimal subdomain essential for PS2/CALP binding was located between residues 421 and 431 of PS2. Similar analysis in PS1 confirmed that CALP interacts with the C terminus of PS1 at a comparable subdomain (i.e. between residues 425 and 460).
We next characterized CALP and its derivatives expressed in cultured cells. For this purpose, we constructed expression plasmids encoding full-length (fl) CALP, ⌬N-CALP lacking the N terminus (i.e. amino acid residues 1-30), that is variable among KChIP family proteins, and EFmt-CALP with the highly conserved Asp and Gly residues within the EF-hand motifs (i.e. Asp-99 and Gly-104 in the 2nd, Asp-135 and Gly-140 in the 3rd, and Asp-183 and Gly-188 in the 4th EF hand) being replaced by Ala, based on human CALP 216 (Fig. 2A). We confirmed by 45 Ca 2ϩ overlay assay of GST-fusioned CALP proteins (migrating at ϳ50 kDa) that CALP binds calcium and that EF-hand motifs as well as the conserved Asp and Gly residues therein are responsible for this calcium binding (Fig. 2B).
We then analyzed expression and binding with PS of CALP in transfected mammalian cells. In transiently transfected COS cells, fl-CALP (CALP 216 ) and EFmt-CALP were detected as ϳ25-kDa proteins by ␣CALP2 (raised against GST-fusioned fl-CALP), whereas ⌬N-CALP migrated at 19 kDa (Fig. 2C). Weak additional bands migrating at 19, 22, and 27 kDa were detected upon expression of fl-CALP or EFmt-CALP, which may presumably be derived by post-translational modification or processing/degradation (43). To compare the reactivity of CALP with other of the KChIP family member proteins, we next transfected cDNAs encoding calsenilin/KChIP3, KCh-IP2L, or KChIP2S transiently in COS cells and analyzed by immunoblotting with ␣CALP2 or ␣N-CALP (Fig. 2D). ␣CALP2 detected CALP as well as calsenilin, KChIP2L, and KChIP2S, suggesting that this antibody recognizes the conserved C-terminal region. In contrast, ␣N-CALP raised against the CALPspecific N terminus reacted exclusively with CALP.
To confirm the association of CALP with PS2 in vivo, we doubly transfected cDNAs encoding CALP and PS2 transiently in COS cells and analyzed the CHAPSO-solubilized cell lysates by co-immunoprecipitation (Fig. 2E). ␣G2L (against PS2 loop) immunoprecipitated fl-CALP as well as ⌬N-CALP or EFmt-CALP, suggesting that PS2 interacts with CALP at the Cterminal domain including the EF-hand motifs, although the highly conserved Asp or Gly residues within EF hands as well as the calcium binding capacity mediated by these residues may not contribute to this interaction.
We next examined the intracellular distribution of CALP and PS2 in transiently transfected COS cells by immunofluorescence staining (Fig. 2F). Upon single transfection, CALP was diffusely distributed in cytoplasm as well as in nuclei. Double transfection with PS2 dramatically changed the distribution of CALP into a reticular pattern, overlapping with that of PS2 in the perinuclear area and endoplasmic reticulum membranes. ⌬N-CALP or EFmt-CALP proteins showed similar distribution patterns to those of fl-CALP (data not shown).
We then examined the expression of endogenous CALP protein in mouse tissues by immunoblotting and showed that CALP is predominantly expressed as a ϳ28-kDa protein positively labeled by both ␣CALP2 and ␣N-CALP in the Triton X-100-soluble or -insoluble membrane fraction of brains (Fig.  3). The nature of additional bands inconsistently labeled by ␣CALP2 or ␣N-CALP is unknown, but some may represent post-translational modifications and others may be nonspecific bands. The relatively larger size (i.e. 28 kDa) of mouse endogenous CALP may be consistent with our finding that cDNAs coding exclusively for CALP 250 were cloned from mouse brain cDNA libraries (see Fig. 1A).

Effect of the Overexpression of CALP on Metabolism of PS Polypeptides and ␥-Secretase Cleavage of ␤APP and Site 3
Cleavage of Notch-To verify the effects of overexpression of CALP on the stabilization or half-life of PS complex, we treated N2a cells stably expressing CALP by cycloheximide (Fig. 4A). Overexpression of either fl-CALP or EFmt-CALP did not affect the stabilization of endogenous PS fragments. fl-CALP was less stable compared with PS fragments and degraded rapidly (t1 ⁄2 ϭ ϳ4 h). However, EFmt-CALP was more rapidly degraded compared with fl-CALP (t1 ⁄2 ϭ ϳ1 h), implicating the integrity of the EF-hand motifs in the stabilization of CALP.
To determine the size of the molecular complex harboring CALP in membranous compartments, as well as the effect of overexpression of CALP on the complex formation of PS fragments, we separated 1% CHAPSO-solubilized membrane fraction from N2a cells stably transfected with PS2 and CALP on a linear glycerol velocity gradient (Fig. 4B) (31). In N2a cells singly expressing PS2, endoproteolytic PS2 fragments were predominantly detected in the high molecular mass range of 140 -443 kDa, whereas PS2 holoproteins were fractionated in the low molecular mass range of 67-140 kDa. Fractionation patterns of stable N2a cells co-expressing PS2 and fl-CALP or EFmt-CALP were essentially identical to that without co-expression of CALP in terms of the distribution of holoproteins and fragments of PS2. Moreover, CALP was principally recovered in the low molecular weight range fractions (i.e. ϳ70 kDa or lower), and EFmt-CALP also was detected in the same fractions.
To examine whether CALP affects the ␥-secretase activity of PS complex, we measured the levels of secreted A␤40 and A␤42 in conditioned media from N2a cell lines stably expressing PS2 and CALP by A␤ C-terminal specific ELISAs (Fig. 4C). In conditioned media of cells expressing mt PS2, the percentage of A␤42 as a fraction of total A␤ (ϭ A␤x-40 ϩ A␤x-42/total A␤: %A␤42) was elevated to ϳ80%, whereas %A␤42 in cells expressing wild-type (wt) PS2 was ϳ20% (Fig. 4C, see percentages indicated above each column). Overexpression of fl-or EFmt-CALP did not alter the absolute levels of A␤ secretion of %A␤42 in N2a cells co-expressing wt or mt PS2.
To examine the effect of overexpression of CALP on site 3 cleavage, we then transiently transfected a cDNA encoding Notch⌬E that contains the signal sequence, TM domain, and the intracellular domain of mouse Notch-1 harboring C-terminal Myc epitope tags in N2a cells stably overexpressing PS2 and CALP (Fig. 4D) (36). Consistent with our previous findings (31), production of NICD was impaired in N2a cells stably expressing mt PS2, whereas overexpression of wt PS2 had no significant effect on site 3 cleavage of Notch. The proteolytic release of NICD was not affected by overexpression of fl-or EFmt-CALP in cells expressing wt PS2. Moreover, inhibition of NICD production by FAD-linked mutation in PS2 was not affected by overexpression of fl-or EFmt-CALP, suggesting that overexpression of CALP had no significant effect on site 3 cleavage of Notch. Collectively, overexpression of fl-or EFmt-CALP did not affect the ␥-secretase activities of PS complex.
Electrophysiological Function of CALP in the Regulation of Kv4 Current-To examine if CALP interacts with Kv4 and function as KChIPs, we first examined the interaction of CALP with Kv4.2 in COS cells transiently co-transfected with these two cDNAs. Full-length or ⌬N-human CALP were co-immunoprecipitated with rat Kv4.2, whereas EFmt-CALP was not, suggesting that the binding of CALP with Kv4 is taking place through, and is dependent on the integrity of, the EF-hand motifs (Fig. 5).
We next examined the electrophysiological functions of CALP, ⌬N-CALP, and EFmt-CALP in HEK293 cells expressing Kv4.2 (HEK-4.2). Outward currents were measured from HEK293 cells transfected with cDNAs encoding Kv4.2 and GFP, which were detected by the fluorescent signal. Although the cDNA ratio of pcDNA3.1 (Kv4.2) and pTracer-CMV2 (GFP) used for the co-transfection was 1:4, all the HEK cells having GFP signal (n Ͼ50) in this study (HEK-4.2/pT) showed substantial I A , which was not detected in native HEK and those transfected with pTracer alone. In HEK-4.2/pT, typical A-type K ϩ currents (I A ; early inactivating K ϩ current) were activated by depolarization from a holding potential of Ϫ80 mV to ϩ40 mV by a 10-mV step (Fig. 6A) in a potential dependent manner as shown in current-voltage relationship (Fig. 6D). Similar but larger I A were recorded in HEK-4.2 co-transfected with fl-CALP (HEK-4.2/fl; Fig. 6B) and ⌬N-CALP (HEK-4.2/⌬N; not shown). The current density of I A due to Kv4.2 expression was markedly enhanced by the co-expression with fl-or ⌬N-CALP ( Fig. 6D and Table II). On the other hand, the co-expression with EFmt-CALP (HEK-4.2/EFmt) did not change the I A density (Fig. 6, C and D, and Table II). Cell capacitance was not affected by any arrangement of co-expression components examined.
The time courses of I A activation and inactivation in HEK-4.2/fl or ⌬N appeared to be slower than those in HEK-4.2/pT or EFmt under the peak-matched comparison (not shown). Based on exact analyses, the inactivation phase of I A at potentials positive to Ϫ10 mV was well fitted by sum of two exponential components in all four groups (Equation 1).
where I(t)/I max is the relative amplitude of I A as the function of t versus the maximum. A f and A s values are the relative contribution of the fast and slow inactivating components at time  0). The f at ϩ20 mV was significantly increased by the co-expression with fl-or ⌬N-CALP but not affected by EFmt-CALP (Table II). Neither A f nor s was significantly affected by the co-expression arrangement.
The influence of co-expression with fl-, ⌬N-, or EFmt-CALP on the voltage dependence of I A activation and inactivation was examined using conventional two pulses protocols (see "Materials and Methods"). The relationships between test voltages and the fraction of activation or inactivation of I A were well fitted by Boltzmann equation in all four groups (not shown). The summarized results of the voltages required for the halfmaximal activation or inactivation (V1 ⁄2 ) and the slope factors were listed in Table II. Co-expression with fl-or ⌬N-CALP significantly shifted or tended to shift the activation and inactivation of V1 ⁄2 to negative potentials by several mV but that FIG. 7. The time course of recovery from inactivation of I A . A-C, a paired-pulse protocol was applied to determine the time course of recovery from inactivation of I A at Ϫ80 mV (A, Kv4.2ϩpTracer; B, Kv4.2ϩfl-CALP; C, Kv4.2ϩEFmt-CALP). Cells were depolarized from Ϫ80 to ϩ40 mV for 1 s twice with a certain interval (⌬t). D, summarized data obtained from results typically shown in A-C. The relative amplitude of I A was plotted as IP 2 /IP 1 against ⌬t (ms). IP 1 and IP 2 are the peak amplitude of I A activated by P 1 and P 2 , respectively. The recovery time course was best described by single exponential function, and the fitted curves are illustrated. pT, fl-CALP, ⌬N-CALP, and EFmt-CALP indicate the recovery time courses in Kv4.2ϩpTracer, Kv4.2ϩfl-CALP, Kv4.2ϩ⌬N-CALP, and Kv4.2ϩEFmt-CALP, respectively. The number in parentheses indicates the number of cells used in each group. with EFmt-CALP did not (Table II). One of the most striking effects of fl-and ⌬N-CALP was the changes in the time course of I A recovery from inactivation. The recovery time course was studied using a conventional pairedpulse protocol (Fig. 7, A-C). The peak amplitude of I A elicited by the second pulse was normalized with the first one and plotted against the interval (t). The recovery time course of I A was well fitted by a single exponential function, regardless of the co-expression arrangement (Fig. 7D). The time constant in HEK4.2/pT at Ϫ80 mV was markedly shortened by the coexpression with fl-and ⌬N-CALP but not affected by that with EFmt-CALP (Table II). DISCUSSION Here we describe a novel EF-hand protein that we identified as a binding protein with the C terminus of PS and designated CALP/KChIP4. CALP showed homology to calsenilin, a member of the recoverin superfamily calcium-binding proteins that has been shown to interact with the C-terminal region of PS (35). It has been reported that calsenilin increases the alternative cleavage of 19-kDa PS2 C-terminal fragment as well as the sensitivity for apoptosis (35,43), and that calsenilin reversed the enhancement in calcium signaling caused by expression of mutant PS1 in Xenopus oocyte (27). However, the significance of the association of calsenilin and PS in relation to the metabolism and ␥-secretase function of PS has not been fully understood. In this study, yeast two-hybrid and co-immunoprecipitation analyses showed that CALP interacts with PS in vivo; however, overexpression of CALP did not affect the metabolism and ␥-secretase activities of PS complex. Fractionation analysis of CHAPSO-solubilized membrane fractions suggested that CALP is not a stable component of HMW PS complex that represents the active form of ␥-secretase. These results suggest that CALP is possibly a transient binding partner of a PS complex that may represent the immature form of a functional PS complex; however, the precise function of CALP and other KChIPs in PS complex should further be examined by eliminating these proteins from cells, for example by knockout or RNA interference strategies.
Recently, it has been reported that calsenilin is identical to KChIP3, which binds to and modulates the density and the properties of Kv4 current (32). Because of the high homology in EF-hand motif of CALP to other KChIPs, we have analyzed the effect of the co-expression of CALP on Kv4 current density. Like other KChIPs (32,34), CALP interacted with Kv4.2 polypeptide in vivo and altered the voltage-gating and inactivation properties of Kv4.2. Thus, CALP is a PS-interacting protein as well as a novel KChIP protein, which can be designated as CALP/KChIP4. Although the molecular mechanism underlying the modulation of Kv4 currents by KChIPs remains unclear, it has been envisioned that KChIPs bind to the Kv4 N-terminal domain, facilitate trafficking to the plasma membrane, regulate Kv channel turnover, and/or alter the intrinsic channel property (32, 44 -46). The increase in Kv4 channel density and the modulation of the channel activities by CALP in central nervous system neurons, where these components are highly expressed in combination, may be critical to characterize the neuronal excitability. These results also raise the possibility that CALP and other KChIP proteins including calsenilin interact with various polytopic membrane protein complexes to regulate their metabolism, trafficking, and/or functions in membranous compartments.
CALP polypeptide carries four EF-hand domains and binds calcium. Some PS-associated proteins (e.g. calpain, calmyrin, and sorcin) also bind calcium, and PS has been implicated in intracellular calcium homeostasis (27,(47)(48)(49)(50)(51). In this regard, CALP might have a function in PS-mediated calcium modula-tion. Our data indicate that EF-hand domains of CALP, as well as those of other KChIPs (32), are essential in facilitating the functional expression of Kv4, binding to Kv4, and also modulating the channel kinetics and, thereby, may act as a calcium sensor in the regulation of channel activity. Although it has yet to be determined whether calcium is involved in the regulation of PS function in a similar context to Kv4, it will be important to characterize the effect of CALP on capacitative calcium entry, a novel refilling mechanism for depleted intracellular calcium stores and regulated by PS proteins as recently reported (52,53).
The reason why overexpression of CALP affects the Kv4 current density but not the metabolism and function of PS has yet to be elucidated; a straightforward interpretation of these data would be that CALP is neither the limiting cellular factor that regulates the levels of PS complex nor a component of PS complex that modulates the function of ␥-secretase. In this regard, it would be important to see if CALP, Kv4.2, and PS2 form a ternary complex or if Kv4 or PS2 are incorporated into separate complexes with CALP, although our preliminary immunoprecipitation/Western analyses have so far not shown the presence of a ternary complex (data not shown).
In summary, we identified and characterized CALP/KChIP4 as a novel PS-and Kv4-binding protein that belongs to calsenilin/KChIP protein family and modulates Kv4 functions. Further investigations into the molecular mechanism whereby CALP regulates the function of PS and Kv channels will facilitate our understanding of Alzheimer's disease and normal brain function.