Calumenin, a Ca2+-binding Protein Retained in the Endoplasmic Reticulum with a Novel Carboxyl-terminal Sequence, HDEF*

We have identified and characterized a cDNA encoding a novel Ca2+-binding protein named calumenin from mouse heart by the signal sequence trap method. The deduced amino acid sequence (315 residues) of calumenin contains an amino-terminal signal sequence and six Ca2+-binding (EF-hand) motifs and shows homology with reticulocalbin, Erc-55, and Cab45. These proteins seem to form a new subset of the EF-hand protein family expressed in the lumen of the endoplasmic reticulum (ER) and Golgi apparatus. Purified calumenin had Ca2+-binding ability. The carboxyl-terminal tetrapeptide His-Asp-Glu-Phe was shown to be responsible for retention of calumenin in ER by the retention assay, immunostaining with a confocal laser microscope, and the deglycosylation assay. This is the first report indicating that the Phe residue is included in the ER retention signal. Calumenin is expressed most strongly in heart of adult and 18.5-day embryos. The calumenin gene (Calu) was mapped at the proximal portion of mouse chromosome 7.

The endoplasmic reticulum (ER) 1 is involved in synthesis and modification of secretory and membranous proteins as well as resident proteins in the lumen of the ER, Golgi apparatus, or lysosomes (1). The ER is also known as the major Ca 2ϩ storage compartment in eukaryotic cells. By pumping cytosolic Ca 2ϩ into the ER lumen, cells keep their cytosolic concentration of free Ca 2ϩ at extremely low levels so that they can use Ca 2ϩ as intracellular signal (2). Besides, the ER itself needs luminal Ca 2ϩ for its normal functions such as protein folding and protein sorting (3)(4)(5)(6). Among many ER-resident proteins, endoplasmin (GRP94) (7,8), Bip (GRP78) (9,10), protein-disulfide isomerase (ERp59) (11,12), and calreticulin (CRP55) (13) are Ca 2ϩ -binding proteins that are associated with Ca 2ϩ -dependent folding and maturation of secretory proteins in the ER lumen (14,15). In addition, two Ca 2ϩ -binding ER-resident proteins, reticulocalbin (16) and Erc-55 (17), have been isolated recently. They have multiple EF-hand motifs and constitute a new subset of the EF-hand superfamily, together with a homologous protein in the lumen of the Golgi apparatus, Cab45 (18). However, their physiological functions are still unknown.
ER-resident proteins generally carry a retention signal at their carboxyl terminus. This ER retention signal is first identified to be tetrapeptides, Lys-Asp-Glu-Leu (KDEL) in mammalian cells and His-Asp-Glu-Leu (HDEL) in yeast (19,20). Soluble ER-resident proteins are trapped by binding to the KDEL receptor expressed in the cis-Golgi and retrieved to the ER (21)(22)(23). Further studies have demonstrated that HDEL and several variants of the KDEL sequence can also work as the ER retention signal in mammalian cells (24,25). Comparison of variants of the KDEL sequence suggests that the replacement of Lys and Asp residues with other amino acid residues does not abolish the ER retention activity. However, the carboxyl-terminal two residues are considered to be critical because the third and fourth positions in all of the ER retention signals are Glu/Asp and Leu/Ile, respectively. During the embryogenesis, the heart begins to beat already in the 8.5-day mouse embryo, whose heart is still a two-chambered tube with one atrium and one ventricle (26). Beating of cardiac myocytes is maintained by the strict regulation of their cytoplasmic Ca 2ϩ concentration. To achieve this regulation, cardiac myocytes develop their specialized ER, called the sarcoplasmic reticulum, as the Ca 2ϩ storage compartment and produce the rhythmical Ca 2ϩ oscillation between the sarcoplasmic reticulum and the cytosol (27). Molecules involved in this Ca 2ϩ oscillation are reported to be present very early in mouse cardiogenesis (28,29).
Since we are interested in molecules involved in heart embryogenesis, we screened in a signal sequence trap library (30,31) of mouse embryonic heart to isolate cDNAs encoding amino-terminal hydrophobic signal sequences. We here report cloning and characterization of cDNA encoding calumenin that binds Ca 2ϩ and carries a new ER retention signal, HDEF, at the carboxyl terminus. Calumenin is a novel member of the reticulocalbin family, a new subset of the EF-hand superfamily in the ER. Calumenin is most strongly expressed in the heart of adult and 18.5-day embryos.

EXPERIMENTAL PROCEDURES
cDNA Cloning-Poly(A) RNA from approximately 100 hearts of 9.5day postcoitus (dpc) mouse embryos was extracted with TRIzol reagent (Life Technologies, Inc.) and Oligotex™-dT30ϽSuperϾ (Roche). cDNA was synthesized from 1.35 g of poly(A) RNA using Super Script II (Life Technologies, Inc.). First strand cDNA was synthesized with 25 pmol of URPX3 primer: GAG-ACG-GTA-ATA-CGA-TCG-ACA-GTA-GCT-CGA-GXX-XXX-XXX-X (where X represents one of the following: A, G, C, or T). After alkali lysis of RNA and the poly(A) tailing procedure, second strand cDNA was synthesized with 25 pmol of ESTN primer: CCG-CGA-ATT-CTG-ACT-AAC-TGA-(T) 17 XX. Then cDNA of 400 -800 base pairs were fractionated by agarose gel electrophoresis and subjected to polymerase chain reaction (PCR) using ExTaq (TaKaRa, Japan) and 25 pmol of ESP primer (CCG-CGA-ATT-CTG-ACT-AAC-TGA-TT) and Ad-P1 primer (GAC-GGT-AAT-ACG-ATC-GAC-AGT-AGG) under the following conditions for Thermal-Cycler (TaKaRa, Japan): 94°C for 5 min and then 94°C for 45 s, 52°C for 60 s, and 72°C for 2 min for 25 cycles and 72°C for 10 min. After cloning cDNA unidirectionally between the EcoRI and XhoI sites of pSuc2t7F1ori vector, screening procedures were performed as described (31).
To clone full-length cDNA, the 3Ј-rapid amplification of cDNA ends * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U81829.
(RACE) method was performed with LAtaq (TaKaRa). Mouse 13.5-dpc embryonic heart cDNA library in Uni-ZAP XR vector was used (Stratagene). C39N1b outer primer (biotinylated) (GTG-GAG-CTC-CCG-GGA-AAG-GTT-ATC-ATG) and C39N2 inner primer (CAT-GGA-CCT-GCG-TCA-GTT-TC) were designed as gene-specific primers and used with standard T7 primer (CGC-GTA-ATA-CGA-CTC-ACT-ATA-GGG-C). Conditions adopted for Perkin-Elmer 9600 were as follows: 95°C for 2 min and then 95°C for 30 s, 68°C for 30 s, and 72°C for 2 min for 25 cycles. The reaction mixture of first PCR with C39N1b and T7 primers was subjected to purification with M280 Dynabeads (Dynal) to decrease backgrounds due to nonspecific amplification by T7 primer. Second PCR was performed with C39N2 and T7 primers using the purified mixture. For sequencing, amplified cDNA was cloned into pGEM-T vector (Promega).
Sequencing was performed with an automated sequencer (model 373A; Applied Biosystems), and sequence analysis was done with the computer analysis program, GeneWorks (IntelliGenetics, Inc.). Homology search was performed with BLAST and FASTA, using GenBank and EMBL as DNA data bases and PRF, PIR, and SwissProt as protein data bases. Motif search and localization analysis were performed on line at Prosite.
Other Methods-The procedures used to stain cells were essentially as described (18,33). Stained cells were analyzed with a confocal scanning laser microscope (LSM410 UV, Carl Zeiss Inc.) following the protocol for double stain by fluorescein isothiocyanate and rhodamine.
The Slow Fade Antifade kit (Molecular Probes, Inc.) was used to prevent photobleaching. Retention of calumenin in ER was assayed as follows. Cells transfected with FLAG-calumenin and FLAG-calumenin-⌬HDEF were further incubated for 4 h in serum-free medium. The media were then concentrated using Centricon-30 (Amicon). Cell extracts and concentrated media were analyzed by 12% polyacrylamide gel electrophoresis with SDS followed by Western blotting. The reverse transcriptase-PCR method. After first strand synthesis using random 9-mer, PCR was performed with C39N3 primer (GGA-AGA-TGG-ACA-AGG-AAG-AGA-CC) and C39C1b primer (biotinylated) (AGA-GTT-GTT-CCT-CGG-AAG-CTC-G). Transfection of COS-7 cells was performed using lipofectamine (Life Technologies) according to the manufacturer's protocol. Protein synthesis was stopped before protein purification and immunostaining assay by treating cells with 300 M cycloheximide for 2 h (34). The filter for Northern blot was prepared as described (35) and hybridized using Quick-Hyb solution (Stratagene). The 45 Ca 2ϩ binding assay was performed as described (36). Membranes of 45 Ca 2ϩ binding assay and Northern blotting was analyzed using an image analyzer (BAS 2000, Fuji Film). Deglycosylation assay was done as described (18) and analyzed by SDS-polyacrylamide gel electrophoresis and Western blotting. For Western blotting, ECL Western blotting detection reagents (Amersham Life Science, Inc.) was used. Chromosomal mapping of the calumenin gene was done as described (35,37).

RESULTS
Cloning and Sequencing of a Novel cDNA C39 -4.4 ϫ 10 5 yeast transformants from 9.5 dpc embryonic heart cDNA library were screened, and 386 positive clones were obtained by the signal sequence trap method described previously (30,31). Among these, nucleotide sequences of 17 clones were not reported in any mammalian species, 15 clones were homologous to sequences reported in mouse or other mammals and the rest were redundant clones. One of the novel clones (C39) was picked up for further studies because its mRNA was most strongly expressed in heart (see below). The full-length cDNA of C39 was isolated from a 13.5 dpc embryonic heart cDNA library by the 3Ј-RACE method. Nucleotide sequences of two independent clones were determined to avoid sequence errors FIG. 2. Alignment of C39, Erc-55, reticulocalbin, and Cab45. All family members share significant homology even outside the EF-hand motif. EF-hand motifs are shaded. Amino acid residues identical to the C39 protein are boxed. and shown to encode a 315-amino acid protein (Fig. 1A). The translation start site was assigned at nucleotide positions 36 -38 because of the presence of an NH 2 -terminal signal sequence and the comparison with other family members to be described below.
The C39 Protein Is a Ca 2ϩ -binding Protein-cDNA sequence showed that the C39 protein does not have any other hydrophobic stretch long enough to anchor the protein in the membrane (Fig. 1B). The C39 protein carries six potential Ca 2ϩbinding EF-hand motifs and a putative N-glycosylation site. Homology search with the BLAST and FASTA computer program revealed the presence of many Ca 2ϩ -binding proteins homologous to the C39 protein in the EF-hand region. Among these, reticulocalbin (16) and Erc-55 (17) in the ER and Cab45 (18) in the Golgi apparatus had homology in size and sequence to the C39 protein even outside of the EF-hand region (Fig. 2). These proteins appear to form one subfamily of Ca 2ϩ -binding proteins.
Each EF hand in the C39 protein has the general feature for high affinity Ca 2ϩ -binding according to Kretsinger's rule (38); the presence of the helix-loop-helix motif in which 5 oxygencontaining residues and the central glycine are conserved for coordination of Ca 2ϩ binding (Fig. 3A). Although the central glycine of the third motif is replaced by leucine, the same replacement was seen in other Ca 2ϩ -binding proteins (39,40) and considered to retain the Ca 2ϩ -binding activity by the secondary structure prediction of computer analysis (41).
To test whether six EF hands in the C39 protein really have Ca 2ϩ -binding ability, the 45 Ca 2ϩ binding assay was performed as described (36). Strong signal was detected in the FLAG-C39 fusion protein but none in the FLAG-bacterial alkali phosphatase fusion protein (Fig. 3B). The positive band was approximately 57 kDa in size, which was equal to the size of the FLAG-C39 fusion protein detected by Amido Black staining, indicating that EF hands in the C39 protein indeed have Ca 2ϩ binding ability.
Subcellular Localization of the C39 Protein-To determine the intracellular localization of the C39 protein, COS-7 cells transfected with FLAG-C39 were doubly stained with an anti-FLAG antibody and an anti-protein-disulfide isomerase antibody and observed under a confocal laser microscope. The staining profile of anti-FLAG antibody (Fig. 4B) was almost identical to that of anti-protein-disulfide isomerase antibody (Fig. 4A), showing diffuse ER staining patterns as reported previously for protein-disulfide isomerase, a well characterized protein in the ER (42). The identical staining patterns with both antibodies (Fig. 4C) indicate that the C39 protein is an ER-resident protein. Since the C39 protein has Ca 2ϩ -binding ability and is localized in the ER lumen, we named the C39 protein calumenin.
The HDEF Sequence Is a Novel ER Retention Signal-To examine whether the C-terminal tetrapeptide HDEF of calumenin can serve as a novel intracellular retention signal, we constructed two expression vectors: FLAG-calumenin, a fusion protein of FLAG epitope and calumenin, and FLAG-calumenin-⌬HDEF, a fusion protein of the FLAG epitope and calumenin lacking its C-terminal HDEF sequence. FLAG epitopes were incorporated 5 amino acids downstream of the putative signal sequence cleavage site in each construct. COS-7 cells were transfected by these constructs, and concentrated media and cell extracts were analyzed by Western blotting. Most of the FLAG-calumenin-⌬HDEF protein was secreted into the medium, while almost all of the FLAG-calumenin protein was # and E indicate a consensus hydrophobic amino acid and amino acid with the oxygen-containing side chain, respectively. Glycine (G) in the loop domain is often but not necessarily conserved in active EFhand motifs. B, a 45 Ca 2ϩ binding assay was performed as described (36). The purified FLAG-C39 fusion protein (lanes 1 and 3) and the FLAGbacterial alkali phosphatase (FLAG-BAP) fusion protein (lanes 2 and 4) were electrophoresed in 12% SDS-polyacrylamide gel electrophoresis, transferred to a nitrocellulose filter, and incubated with 45 Ca 2ϩ . After washing, the filter was exposed in an image analyzer ( 45 Ca; lanes 3 and 4). The filter was also stained with Amido Black after the exposure (Amido Black; lanes 1 and 2). The arrowhead indicates the FLAG-C39 fusion protein. retained in the cell (Fig. 5). These results indicate that HDEF is essential to maintain calumenin within the cell.
We further examined intracellular localization of calumenin by treatment with deglycosidases. Proteins translocated into the ER are exposed to core glycosylation. At this moment, glycosylated proteins are sensitive to the deglycosidase endoglycosidase H, which cleaves high mannose oligosaccharides on proteins. However, once proteins are transported into the medial Golgi and modified by Golgi-mannosidase II, they become resistant to endoglycosidase H. However, Golgi proteins are still sensitive to N-glycosidase F, a glycosidase whose activity is not disturbed by the following carbohydrate modifications in the medial-and trans-Golgi. Calumenin had one putative N-linked glycosylation site. Affinity-purified FLAGcalumenin protein was incubated with either endoglycosidase H or N-glycosidase F and subjected to Western blotting with the anti-FLAG M2 antibody. We also performed the assay against FLAG-calumenin-rHDEL, whose C-terminal tetrapeptide HDEF was replaced with HDEL, known as the ER retention signal. Both FLAG-calumenin and FLAG-calumenin-rH-DEL showed strong bands at 57 kDa without the glycosidases (Fig. 6, lanes 1, 3, 5, and 7). Digestion with either endoglycosidase H or N-glycosidase F shifted the size of the strong bands to 52 kDa (lanes 2, 4, 6, and 8). We consider that 57-kDa bands correspond to the glycosylated form of calumenin, and 52-kDa bands correspond to the deglycosylated form. Calumenin is indeed translocated and glycosylated in the ER and not modified in the Golgi apparatus, because Endo H completely cleaved oligosaccharides on FLAG-calumenin (lanes 2 and 6) and FLAG-calumenin-rHDEL (lanes 4 and 8). These results indicate that calumenin resides in the ER and that the C-terminal tetrapeptide HDEF works as the ER retention signal like HDEL (16,17).
Expression and Chromosomal Localization of the Calumenin Gene-Northern blotting analysis showed that calumenin was expressed ubiquitously in all tissues examined. However, the expression of calumenin was strong especially in heart and lung. Compared with expression in adult heart, expression in 18.5-dpc heart was slightly stronger (Fig. 7A). The reverse transcriptase-PCR method showed that calumenin mRNA was already expressed as early as 8.5 dpc (Fig. 7B).
To determine the chromosomal localization of the calumenin gene (Calu), 2 strain distribution patterns of restriction fragment length polymorphisms of the calumenin gene were determined in 24 independent recombinant inbred strains derived from crosses between AKR/J and DBA/2J (AXD) ( Table I). Analysis of the distribution pattern revealed the linkage of the calumenin gene with markers located at the proximal region of chromosome 7 (Fig. 8). DISCUSSION A Novel Ca 2ϩ -binding Protein in the ER Lumen-We isolated and characterized a novel Ca 2ϩ -binding protein calumenin from mouse embryonic heart. Calumenin is located in the ER and homologous to previously reported Ca 2ϩ -binding proteins such as reticulocalbin and Erc-55 in the ER (16,17), and Cab45 (18) in the Golgi apparatus. Since all of these proteins including calumenin have six EF hands and Ca 2ϩ -binding activity, they constitute one subset of the EF-hand superfamily. A data base search also revealed that cDNA (SCF13) cloned as DNA supercoiling factor of silkworm (43) also belongs to this family. The deduced amino acid sequence of SCF13 has an aminoterminal signal sequence, six EF hands, and the C-terminal HDEF sequence, which we proved to be the ER retention signal.
Although many ER Ca 2ϩ -binding proteins have been reported so far, their functions are largely not yet well understood. Two possible functions of ER Ca 2ϩ -binding proteins have been suggested. First, calreticulin, one of the Ca 2ϩ -binding proteins in the ER, is reported to regulate the capacity of Ca 2ϩ in the ER (44,45) as well as to have chaperone function (46,47). Another luminal Ca 2ϩ -binding protein, calsequestrin, is also reported to regulate Ca 2ϩ flow from the ER (48,49). These reports suggest that Ca 2ϩ fluxes may be regulated somehow by 2 The gene name Calu for calumenin has been approved by Lois J. Maltais, the Nomenclature Coordinator for the Mouse Genome Data base (The Jackson Laboratory, Bar Harbor, ME). Ca 2ϩ -binding proteins located in the luminal side of the ER. Second, Ca 2ϩ in the ER is reported to be necessary for normal functions of the ER such as protein folding and protein sorting (3)(4)(5)(6), suggesting the existence of ER-resident molecules whose function is regulated by Ca 2ϩ . In fact, Bip, a Ca 2ϩ -binding protein in the ER, is reported to function in the Ca 2ϩ -dependent manner (50).
Further study is needed to determine whether or not calumenin is involved in either of the two possible functions. Re-cently, interaction molecules of Erc-55 have been reported: the E6 protein of papilloma virus, which has p53-independent tumorigenic activity (51), and taipoxin, which blocks neuromuscular transmission at the presynaptic site (52). As discussed above, calumenin belongs to the same subset of the EF-hand superfamily as Erc-55. Thus, the search for proteins that interact with calumenin could provide a clue to elucidate biological functions of this subset.
A Novel ER Retention Signal-We report a novel carboxyl- FIG. 7. Expression pattern of calumenin. A, Northern blotting of mouse RNA with calumenin cDNA as probe. 20 g of total RNA from various mouse tissues were applied to each well (lower column), and Northern blotting was performed using a 512-base pair C39 fragment obtained in the initial screening as a probe (upper column). B, reverse transcriptase-PCR was performed using 1 g of total RNA and calumenin-specific primers. After 28 cycles of PCR, products were electrophoresed and stained with ethidium bromide.

TABLE I Strain distribution patterns of calumenin (Calu) and closely linked markers in RI strains AXD
Restriction fragment length polymorphisms were found by ApaI cleavage of the calumenin in AKR/J (3.8 kilobase pairs) and DBA/2J (4.8 kilobase pairs). Determination and analysis of the restriction fragment length polymorphism distribution pattern among recombinant inbred strains derived from AKR/J and DBA/2J were performed as described (35,37). Strain-specific alleles are abbreviated as A (AKR/J) and D (DBA/2J). Unknown alleles that differ from A and D are identified with U. Putative crossing over points are indicated by X. Pmv4 terminal tetrapeptide HDEF of calumenin as a new ER retention signal to prevent proteins from being secreted and to keep them in the ER. We observed that the HDEF sequence had the intracellular retention activity (Fig. 5). The deglycosylation assay (Fig. 6) and the immunostaining profile (Fig. 4) suggest that the HDEF sequence works as the ER retention signal. The C-terminal tetrapeptide KDEL is the first ER retention signal reported in mammalian cells (20). The C-terminal tetrapeptide HDEL, which was originally identified in yeast, was also found to work as the ER retention signal in mammalian cells (16,17). Other reports show that the first and second positions of the C-terminal tetrapeptide are not strongly conserved to maintain the ER retention activity. In contrast, replacement of the third and fourth residues is strictly restricted. The third position requires an acidic residue such as Glu or Asp, and the fourth position requires a hydrophobic residue containing the aliphatic side chain such as Leu or Ile, which cannot be replaced by either Val or Ala (24,25). Since Cab45, a soluble protein in the Golgi apparatus has HEEF sequence at the C terminus, this tetraresidue sequence was presumed to serve as the Golgi retention signal. The C-terminal Phe, a hydrophobic residue containing an aromatic side chain, was thought to work somehow differently from Leu (18). However, since our study clearly showed that the HDEF sequence worked as the ER retention signal, the retention signal to the Golgi apparatus may be localized in another portion of Cab45. Proteins with ER retention signals such as KDEL are retrieved from Golgi. The convincing mechanism of this retrieval is that such proteins interact with the KDEL receptor distributing in the cis-Golgi and are retrieved to the ER (22,23). We showed that the C-terminal HDEF sequence worked as the ER retention signal, but further investigation is needed to determine whether HDEF interacts with the KDEL receptor or with an unknown receptor.
Chromosomal Mapping of the Calumenin Gene (Calu)-The calumenin gene was mapped on the proximal portion of mouse chromosome 7. This region is well conserved between human and mouse, and the syntenic comparison suggests that this region corresponds to human 19q13. 2-13.3 (53), at which the gene for progressive familial heart block type I, autosomal dominantly inherited cardiac bundle-branch disorder, is mapped (54). Since the causal gene of this disease has not been determined so far, further investigation is needed to determine whether the calumenin gene is involved in this disease.