Stomach-specific Calpain, nCL-2/Calpain 8, Is Active without Calpain Regulatory Subunit and Oligomerizes through C2-like Domains*

Calpains constitute a family of intracellular Ca2+-regulated cysteine proteases that are indispensable in the regulation of a wide variety of cellular functions. The improper activation of calpain causes lethality or various disorders, such as muscular dystrophies and tumor formation. nCL-2/calpain 8 is predominantly expressed in the stomach, where it appears to be involved in membrane trafficking in the gastric surface mucus cells (pit cells). Although the primary structure of nCL-2 is quite similar to that of the ubiquitous m-calpain large subunit, the enzymatic properties of nCL-2 have never been reported. Here, to characterize nCL-2, the recombinant protein was prepared using an Escherichia coli expression system and purified to homogeneity. nCL-2 was stably produced as a soluble and active enzyme without the conventional calpain regulatory subunit (30K). Purified nCL-2 showed Ca2+-dependent activity, with half-maximal activity at about 0.3 mm Ca2+, similar to that of m-calpain, whereas its optimal pH and temperature were comparatively low. Immunoprecipitation analysis revealed that nCL-2 exists in both monomeric and homo-oligomeric forms, but not as a heterodimer with 30K or 30K-2, and that the oligomerization occurs through domains other than the 5EF-hand domain IV, most probably through domain III, suggesting a novel regulatory system for nCL-2.

Calpain (Clan CA-C2, EC 3.4.22.17) is an intracellular Ca 2ϩregulated cysteine protease, comprising a superfamily with members in almost all eukaryotes and some bacteria (1)(2)(3). Calpains regulate a wide variety of cellular functions including the cell cycle, signal transduction, apoptosis, and membrane trafficking, through the limited proteolysis of their substrates (1,2,4,5). Dysregulation of calpain activity and/or defective mutations in calpain genes cause lethality (6 -8) or a variety of pathological phenotypes, which include muscular dystrophies in mammals (9,10), degeneration in the developing optic lobes of Drosophila (11), inadaptability to alkaline conditions in fungi and yeasts (12,13), masculinization of nematode hermaphrodites (14), and defective aleurone cell development in maize (15). These observations clearly indicate calpain is indispensable physiologically; however, the specific physiological functions of calpains and the molecular mechanisms underlying their functions remain unclear.
The mammalian calpain family comprises products from 14 independent genes, and can be classified into typical and atypical members according to their domain structures (2,3). The ubiquitousand m-calpains, well characterized typical calpains in mammals, are heterodimers composed of a distinct 80-kDa large catalytic subunit (abbreviated here to CL and mCL, respectively; also known as calpain 1 and 2) and a common 30-kDa small regulatory subunit (30K). 2 The large and small subunits contain four (I-IV) and two (V and VI) domains, respectively: the regulatory N-terminal domain (I), the protease domain (II), the C2-domain-like Ca 2ϩ /phospholipid-binding domain (III), the penta-EF-hand domains (IV and VI), which are missing in atypical calpains, and the Gly-clustering hydrophobic domain (V). The C-terminal fifth EF-hand motifs in domains IV and VI cannot bind Ca 2ϩ , and interact with each other to form the heterodimer. Domain IV is also reported to function in the recognition of substrates and regulatory molecules (16 -18).
In the absence of Ca 2ϩ , the protease domain (II) is separated into two subdomains, IIa and IIb, and residues in the catalytic triad are prevented from assuming the correct positions for hydrolysis. The binding of Ca 2ϩ to domains IIa, IIb, III, IV, and VI induces conformational changes that allow domain II to form a single active domain (19 -23). Previous analyses revealed that CL and mCL show full activity in the absence of 30K if correctly folded (24), and cells from 30K gene-deficient mice, which die at E10.5-11.5, show neither detectableor m-calpain activity nor detectable CL or mCL protein, even though their mRNAs are expressed normally (7,8). These results indicated that 30K functions as a molecular chaperon in vivo, at least for CL and mCL. nCL-2/calpain 8, identified as a stomach-specific calpain, is a typical calpain with four domains (25) (see Fig. 3B). nCL-2 is most similar to mCL (amino acid identities in the full-length protein and in domains II and IV are 61.5, 73.4, and 51.8%, respectively). Possible functions for nCL-2 in the pituitary gland and in the embryogenesis of Xenopus laevis have been reported (26,27). We recently showed that, in the stomach, nCL-2 as well as nCL-4/calpain 9, a digestive tract-specific calpain, are expressed specifically in the gastric mucus cells (pit cells), quite unlike theand m-calpains, which are expressed diffusely in all cells, and that nCL-2 may be involved in the regulation of vesicle trafficking in the pit cells (28). These reports highlight potentially important nCL-2 functions that cannot be compensated for byand m-calpains, not only in the stomach but also in other tissues.
However, the enzymatic properties of nCL-2, including its regulatory mechanisms, have not been reported, except for one yeast two-hybrid analysis suggesting that nCL-2 does not interact with 30K despite the high similarity between nCL-2 and mCL (29). Thus, in this study, using Escherichia coli-expressed and purified nCL-2, the detailed biochemical properties of this enzyme were examined, and several unique characteristics were revealed. Unexpectedly, nCL-2 was found to exist as both a monomer and as homo-oligomers, and the domain responsible for oligomerization was, unlike in theand m-calpains, the C2-like domain (III), not the penta-EF-hand domain (IV), suggesting that nCL-2 has a novel regulatory mechanism.
Cloning and Construction of Expression Plasmids-cDNAs encoding full-length mouse nCL-2, human and mouse 30K, and human and mouse 30K-2/CAPNS2 were amplified by PCR using Pfu turbo DNA polymerase (Stratagene Inc.), and the sequences were verified after being subcloned into appropriate vectors. For bacterial expression, the amplified mouse nCL-2 cDNA was ligated into the pCold I vector (TaKaRa) to produce proteins fused with an NH 2 -terminal His tag. For COS7 cell expression, the amplified cDNAs were ligated into a modified pSRD vector (31) to produce proteins fused with an NH 2 -terminal FLAG, HA, or MYC tag. The nCL-2 deletion mutants were constructed by introducing termination codons at the appropriate positions using a PCR-mediated site-directed mutagenesis method as previously described (10).
Protein Expression in Bacterial Cells and Purification-The constructed pCold I expression plasmids were transformed into E. coli BL21/pG-Tf2 (TaKaRa), and the transformants were cultured at 37°C in 3 liters of LB medium containing 0.1 mg/ml ampicillin until A 600 ϭ 0.6. Protein expression was induced by the addition of 0.5 mM isopropylthiogalactoside and 5 ng/ml tetracycline (tet) for 24 h at 15°C. Harvested cells were washed once with phosphate-buffered saline, and resuspended in 150 ml of lysis buffer (20 mM Tris-Cl (pH 7.5), 150 mM NaCl, 5 mM imidazole, and 1.4 mM 2-mercaptoethanol), and lysed with a French Press (American Instruments Inc.). The cell lysate was spun at 55,000 ϫ g for 20 min, and the recovered supernatant was filtered through a 0.22-m pore filter (Millipore). The supernatant was then applied by gravity to a 2-ml Ni 2ϩ -chelating agarose column (Qiagen) equilibrated with lysis buffer. The column was washed with 10 column volumes of lysis buffer, and then six column volumes of wash buffer (20 mM Tris-Cl (pH 7.5), 150 mM NaCl, 50 mM imidazole, and 1.4 mM 2-mercaptoethanol). Protein was then eluted with 8 ml of elution buffer (20 mM Tris-Cl (pH 7.5), 150 mM NaCl, 250 mM imidazole, and 1.4 mM 2-mercaptoethanol). The eluate was immediately dialyzed against buffer A (20 mM Tris-Cl (pH 7.5), 1 mM EDTA, and 1 mM dithiothreitol) at 4°C. The dialyzed sample was further purified by a Mono Q HR10/10 anion-exchange column (GE Healthcare) with a linear gradient of 0 -0.5 M NaCl in buffer A. For gel filtration, the sample was applied to a HiLoad 16/10 Superdex 200 column (GE Healthcare) equilibrated with buffer A containing 150 mM NaCl. The purity of the protein was examined by SDS-PAGE analysis. The peak fraction was stored at 4°C until use.
Protein Expression in Culture Cells and Immunoprecipitation-COS7 cells were transfected with 5 g of the expression plasmids by electroporation, as previously described (31). For immunoprecipitation, the COS7 cells were harvested 48 h after transfection and lysed with lysis buffer (20 mM Tris-Cl (pH 7.5), 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.2 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 10 M pepstatin A). The lysed cells were spun at 100,000 ϫ g for 20 min and incubated with a 50% slurry of FLAG (M2)-agarose (Sigma) for 2 h, then the protein-bound agarose was washed five times with 1 ml of the lysis buffer. The immunoprecipitated proteins were eluted with 0.1 mg/ml FLAG peptide (Sigma) and mixed with SDS sample buffer. For the in vitro proteolysis assay, 10 l of the eluted immunoprecipitated proteins were incubated with 2 l of 6ϫ Ca buffer (20 mM Tris-Cl (pH 7.5), 100 mM CaCl 2 , and 6 mM dithiothreitol) or 2 l of 6ϫ EDTA buffer (20 mM Tris-Cl (pH 7.5), 100 mM EDTA, and 6 mM dithiothreitol) at 37°C. The reactions were stopped by the addition of 5ϫ SDS sample buffer.
Protein Sequencing-After SDS-PAGE, the proteins were blotted onto a Pro-Blot membrane (Applied Biosystems, Tokyo, Japan). Target protein bands on the membrane were visualized by Coomassie Brilliant Blue G-250 staining, excised, and washed three times with 0.1% (v/v) trifluoroacetic acid in 50% (v/v) methanol and then with absolute methanol. Sequencing was performed using the ABI 491cLC protein sequencer (Applied Biosystems), according to the manufacturer's instructions.

Preparation of Recombinant Mouse nCL-2 Using an E. coli
Expression System-To investigate the solubility and expression efficiency of nCL-2, full-length mouse nCL-2 was expressed with or without 30K in E. coli BL21(DE3). nCL-2 was abundantly expressed in both cases, and no significant difference was observed in its amount or solubility (data not shown). Therefore, nCL-2 was expressed without 30K for the purpose of purification. To improve the solubility of the nCL-2 in E. coli, a tet-on chaperon-inducible BL21/pG-Tf2 strain was used, and active nCL-2 was successfully recovered in a soluble fraction that represented one-third to one-half of the total nCL-2 ( Fig.  1A, lanes 1 and 2).
Purification of Recombinant nCL-2-As described under "Experimental Procedures," NH 2 terminally His-tagged recombinant nCL-2 was expressed and purified to homogeneity by Ni 2ϩ -chelating agarose and Mono Q anion-exchange column chromatographies in succession (Fig. 1A, lanes 3 and 4). The peak elution of nCL-2 from the Mono Q column was around 0.2 M NaCl. Approximately 1 mg of recombinant nCL-2 was purified from 3 liters of E. coli culture, and subjected to further analyses.
Enzymatic Properties of Recombinant nCL-2-Using the nCL-2 prepared above, the enzymatic characterization of recombinant nCL-2 was performed, in comparison with recombinant humanand m-calpains. A fluorogenic substrate, Suc-LLVY-MCA, was used to detect the proteolytic activities. The activity of nCL-2 was Ca 2ϩ dependent, and was blocked by known inhibitors ofand m-calpains but not by other inhibitors (Fig. 2, A and B), demonstrating that nCL-2 possesses characteristics typical of conventional calpains. The nCL-2 was expressed in E. coli and purified on Ni 2ϩ -chelating and Mono Q columns. Samples were separated by SDS-PAGE and stained by Coomassie Brilliant Blue (CBB). nCL-2 bands are indicated by arrowheads. Lanes: 1, total lysate of nCL-2-expressing E. coli BL21/pG-Tf2 cells; 2, its supernatant after centrifugation; 3, an nCL-2 fraction (5 g) eluted from the Ni 2ϩ -chelating column; 4, the final nCL-2 fraction (1.5 g) after Mono Q column chromatography. An asterisk indicates a host protein still included in the final nCL-2 fraction. B, Suc-LLVY-MCA hydrolyzing activities of total lysate supernatants (100 g) of BL21/pG-Tf2 cells transformed with empty pCold vector (mock), or expression vector for inactive nCL-2:C105S (nCL-2CS) or WT nCL-2 (nCL-2). Fluorescence units in the vertical axis represent an increase in emission at 460 nm Ϯ Ca 2ϩ at 20°C for 30 min. C, "mock purification" using mock-transformed BL21/pG-Tf2 cell lysate. Samples were separated by SDS-PAGE and stained by Coomassie Brilliant Blue. Lanes: S1 and S2, nCL-2-and mock-transformed cell lysate supernatants, respectively; E1-E3, the final Mono Q fractions eluted around 0.2 M NaCl. The nCL-2 band is indicated by an arrowhead. concentration of Ca 2ϩ required for half-maximal activity was about 0.3 mM, almost identical to that of m-calpain ( Fig. 2A). The calculated specific activity was a 40th and a 25th of that of and m-calpain, respectively. The specific activity of the caseinolysis of nCL-2 was also lower than those ofand m-calpain (data not shown).
To exclude a possibility that the proteolytic activity observed is due to host protease(s) contaminating in the final nCL-2 fraction, the Suc-LLVY-MCA assay was performed for the lysates of BL21/pG-Tf2 cells transformed with mock (pCold empty vector), or expression vectors for inactive nCL-2 (nCL-2: C105S, in which the active site Cys-105 was mutated to Ser), or WT nCL-2. As a result, the mock-and nCL-2:C105S-transformed cell lysates showed virtually no activity (0.25 Ϯ 0.01 and 1.97 Ϯ 0.02 fluorescence units, respectively) in the presence of Ca 2ϩ , whereas WT nCL-2-transformed cell lysate showed Ͼ200 times their activities (403.07 Ϯ 0.81 fluorescence units) (Fig. 1B). Furthermore, "mock purification" was also performed using the mock-transformed cell lysate (Fig. 1C, S2) in the same manner as that for WT nCL-2 ( Fig. 1C, S1). As a result, the final Mono Q fractions (Fig. 1C, lanes E1-E3) around 0.2 M NaCl, where WT nCL-2 was mainly eluted, showed no activity. Thus, the possibility of host protease activities was excluded.
The optimal temperature for nCL-2 activity was 20°C (Fig.  2C), which was similar to that forand m-calpain under our assay conditions, and to that reported previously for nCL-4 (32,33). The time courses of nCL-2 activity at various temperatures showed that the initial rates were maximal at 30 to 37°C, whereas maximal activity was achieved at 15°C (Fig. 2D). The low optimal temperature for nCL-2 indicates that it is unstable at higher temperatures in the presence of Ca 2ϩ . The optimal pH for nCL-2 activity was 6.0, which is lower than the optimal pH values for conventional calpains (Fig. 2E).
Autolytic Properties of nCL-2-The autolysis of nCL-2 was examined by SDS-PAGE analysis, as shown in Fig. 3A. Various breakdown products appeared in a time-dependent manner, and were confirmed to be autolytic fragments of nCL-2 by Western blot analysis (Fig. 3B). By NH 2 -terminal sequencing of a 43-kDa fragment (Fig. 3A, arrowhead e), one of the autolysis sites was determined to be between Ala-5 and Ala-6. Given its molecular weight, the COOH terminus of this fragment was in or near the acidic loop in domain III (Fig. 3C, e).
The autolytic fragment just below the 81-kDa full-length band (Fig. 3, A and B, arrowhead a) reacted with the anti-nCL2 but not the anti-His antibody. Based on this and its molecular mass (about 79 kDa), this autolytic fragment probably spanned from Ala-6 to the COOH terminus (Fig. 3C, arrowhead a).
Autolytic fragments of 63 kDa (Fig. 3, A and B, arrowhead b) and 41 kDa (arrowhead f ) reacted with the anti-nCL2 and RP2calpain 8 (anti-domain III) antibodies, and with only the anti-nCL2 antibody, respectively. This indicates that the 63-and 41-kDa fragments extend from Ala-6 to the second EF-hand motif (EF-2) in domain IV and to the beginning of domain III, respectively (Fig. 3C, arrowheads b and f ). Autolytic fragments of 45 and 44 kDa (Fig. 3, A and B, arrowheads c and d) reacted with the anti-nCL2 and anti-His, but not the RP-calpain 8 antibodies, indicating that these fragments extend from the NH 2 terminus to the acidic loop in domain III and to the beginning of domain III, respectively (Fig. 3C, arrowheads c and d). In summary, the autolytic sites of nCL-2 include the NH 2 terminus, the NH 2 -terminal parts of domain III, and EF-2 in domain IV.
nCL-2 Exists as a Monomer and Homo-oligomer-The finding that 30K was not required for nCL-2 activity prompted us to investigate the tertiary structure of nCL-2. To examine whether nCL-2 exists as a monomer or homo-oligomer, the purified nCL-2 was subjected to gel filtration column chromatography. As shown in Fig. 4A, nCL-2 was eluted with peaks at positions corresponding to about 80 and 160 kDa. This result indicates that nCL-2 is present as both a monomer and homo-oligomers, and that the monomer and homodimer forms predominate. Suc-LLVY-MCA hydrolyzing activities of both fractions were measured, and the calculated specific activity of the peak fraction at about 160 kDa was about 1/15 of that at about 80 kDa.
To confirm the homo-oligomerization of nCL-2, immunoprecipitation analysis was performed. Full-length mouse nCL-2 fused with an NH 2 -terminal FLAG, HA, or MYC tag were co- expressed in COS7 cells; the cell lysate was spun in an ultracentrifuge and then immunoprecipitated using the appropriate antibodies. The m-calpain subunits, FLAG-30K and HA-mCL, were used as controls. As shown in Fig. 4C, HA-mCL was coprecipitated with FLAG-30K (lane 2), whereas HA-nCL-2 was not (lane 5), consistent with the yeast two-hybrid results described above. These results indicated that this method was applicable for detecting the protein interactions of interest.
nCL-2 Oligomerization Is Not through Domain IV but through Domain III-To investigate which domain is required for the homooligomer formation of nCL-2, several truncation mutants of nCL-2 ( Fig. 4B) were used in the immunoprecipitation experiment. Because the COOH-terminal fifth EF-hand motif (EF-5) in domains IV and VI is critical for the heterodimer formation ofand m-calpains, we first tested whether the EF-5 of nCL-2 was necessary for oligomerization. When a tagged EF-5 deletion mutant (Fig. 4B, FLAG-or HA-⌬EF5) was co-expressed with full-length nCL-2 or its ⌬EF5 mutant, the ⌬EF5 mutant was coprecipitated with the ⌬EF5 mutant as well as with full-length nCL-2 (Fig. 4F, lanes 6-8; Table 1), indicating that EF-5 is not required for the homo-oligomerization of nCL-2.
Therefore, to identify the novel domain(s) critical for the homo-oligomer formation, domain IV-and domain III/IV-truncated mutants (Fig. 4B, ⌬dIV and ⌬dIII ϩ IV, respectively) were tested. The ⌬dIV mutant was co-precipitated with the ⌬dIV mutant (Fig. 5A, lane 3) as well as with full-length nCL-2 (Fig.  5B, lanes 3 and 8; Table 1). On the other hand, the ⌬dIII ϩ IV mutant showed almost no interaction with the ⌬dIII ϩ IV mutant or with fulllength nCL-2 (Fig. 5, A, lane 6; and B, lanes 5 and 10; Table 1). Consistent with these results, HA-domain IV of nCL-2 (Fig. 4B, dIV) was co-precipitated with neither FLAG-full-length nCL-2 nor FLAG-dIV of nCL-2 (Fig. 5C, lanes 3 and 5; Table 1). These results indicate that domain IV is dispensable and domain III is necessary for the oligomer formation of nCL-2.
To confirm the involvement of domain III in nCL-2 oligomerization, FLAG-domain III of nCL-2 (Fig. 4B, dIII) was co-expressed with HA-full-length nCL-2 or HA-dIII of nCL-2. The dIII expression was very faint (Fig. 5C, lysate, lanes 6 and 7), probably because domain III alone is unstable in nature, as previously reported for rat mCL domain III (34). As shown in Fig.  5C, lanes 8 and 10, FLAG-full-length nCL-2 and FLAG-dIII co-precipitated HA-dIII, although the signal was not as clear as in the co-precipitation experiments using full-length nCL-2 or the ⌬dIV mutants (Fig. 5, A and B; Table 1). These results strongly suggest that nCL-2 oligomerizes through domain III,  SEPTEMBER 21, 2007 • VOLUME 282 • NUMBER 38

JOURNAL OF BIOLOGICAL CHEMISTRY 27851
and that domains I and/or II may also be required for the proper conformation for binding ( Table 1).
Effect of Oligomerization on nCL-2 Activity-To examine the effect of homo-oligomer formation on nCL-2 autolysis, HA-nCL-2 was co-expressed with FLAG-nCL-2, immunoprecipitated using an anti-FLAG antibody, and incubated in the absence or presence of Ca 2ϩ , followed by Western blot analysis using anti-FLAG or anti-HA antibodies. As shown in Fig. 6A, both FLAG-nCL-2 and HA-nCL-2 showed Ca 2ϩ -dependent autolysis in the same manner.
Next, to investigate whether or not Ca 2ϩ induces dissociation of the nCL-2 homo-oligomer, the HAand FLAG-nCL-2 oligomers were bound to anti-FLAG-agarose, incubated with or without Ca 2ϩ , washed, and eluted with FLAG-peptide to see if HA-nCL-2 was associated with or dissociated from the bound FLAG-nCL-2 protein. In this experiment, to avoid the autolysis of nCL-2 during incubation with Ca 2ϩ , inactive nCL-2:C105S (nCL-2CS) was used. As shown in Fig. 6B, lanes 3 and 6, HA-nCL-2CS was detected at the same level in the absence or presence of Ca 2ϩ . These results indicate that the nCL-2 homo-oligomer possesses protease activity without dissociating in the presence of Ca 2ϩ .
No Interaction of nCL-2 with 30K-2-Recently, 30K-2/CAPNS2, a novel small subunit that shares about 70% amino acid identity with 30K, was identified, and E. coli-expressed 30K-2 was shown to form a heterodimer with mCL that had enzymatic properties similar to m-calpain (mCL ϩ 30K) (35,36). Therefore, it was possible that 30K-2 interacts with and functions as a regulator of nCL-2. When FLAG-30K-2 was expressed in combination with HA-nCL-2 or HA-mCL, HA-mCL but not HA-nCL-2 was co-precipitated with FLAG-30K-2 (Fig. 7, lanes 3 and 6). The interaction of 30K-2 and mCL was weaker than that of 30K and mCL, consistent with previously reported results (35).

DISCUSSION
In this study, we purified E. coliexpressed recombinant stomachspecific calpain, nCL-2, and demonstrated that it can be present as a monomer or form a homo-oligomer through domain III. This is the first description of the enzymatic and biochemical properties of nCL-2.
As expected from the high sequence similarity between nCL-2 and mCL, several enzymatic properties of nCL-2 resembled those of m-calpain, such as Ca 2ϩ dependence and the inhibitor profile of its enzyme activity. However, several other properties were unique to nCL-2, which might reflect its specific expression and functions in the stomach that cannot be compensated for by theand m-calpains. First, nCL-2 showed . Right, the band intensity of full-length nCL-2 (closed arrowhead) was quantified by densitometry. An asterisk indicates the impurity as also shown in Fig. 1A. B, Western blot analyses with anti-nCL2 (left), anti-His (middle), and RP2-calpain 8 (right) antibodies. Closed arrowheads and open arrowheads (a-f) indicate full-length and autolytic fragments of nCL-2, respectively. C, schematic illustration of the recognition sites for the anti-nCL2, anti-His, and RP2-calpain 8 antibodies, the autolytic sites, and positions of the autolytic fragments obtained above (a-f).
a low optimal temperature in a long incubation assay, which was similar to that of nCL-4 (33) but different from that of p94⌬ (37). Second, the optimal pH for nCL-2 activity was lower than that for theand m-calpains. These properties suggest that nCL-2 evolved to adapt to stomach-specific conditions. Alter-natively, these findings may suggest the existence of certain as-yet unidentified stabilizing factor(s) specific for nCL-2 in the stomach.
The specific activity of nCL-2 was much lower than the specific activities of theand m-calpains. The strong autolytic activity (see Fig. 3A), however, indicated that the low specific activity was not due to partial denaturation of the purified recombinant nCL-2. Moreover, the nCL-2 autolysis was very rapid and extensive; i.e. full-length nCL-2 decayed with a t1 ⁄ 2 of less than 10 s at 25°C, and it was extensively degraded within 3 min (see Fig. 3A). These properties are distinct from theand m-calpains and from nCL-4 ϩ 30K (33,38,39), but similar to p94, although p94 apparently does not require Ca 2ϩ for autolysis (31). Taken together, these findings suggest that 30K suppresses the autolysis of calpains with which it forms a heterodimer, and may indicate that nCL-2 and p94 need to be self down-regulated very rapidly after their activation, probably to play specific roles in highly differentiated cells like pit cells and muscle cells, respectively.
Alternatively, the low specific activity of nCL-2 for Suc-LLVY-MCA and casein, despite its strong autolytic activity, may indicate that the substrate specificity of nCL-2 is rather different from that of theand m-calpains. In this regard, the identification of in vivo substrates for nCL-2, including the can-   1 and 4) was co-expressed with FLAG-nCL-2CS (lanes 3 and 6) or mock (lanes 2 and 5) in COS7 cells. The cell lysates were immunoprecipitated with anti-FLAG-agarose, and the protein-bound anti-FLAG-agarose was incubated with (lanes 4-6) or without (lanes 1-3) Ca 2ϩ , then washed with the same buffer. The protein remaining on the agarose was subjected to Western blot analysis using anti-FLAG and anti-HA antibodies.  didate molecule ␤-COP (28), is a very important and emerging future issue. It was shown that nCL-2 is not co-precipitated with 30K or 30K-2, and that it is stable and has full proteolytic activity without 30K, strongly suggesting that no interaction occurs between nCL-2 and 30K or 30K-2 in vivo. As mentioned above, although it cannot be excluded that a regulatory subunit(s) other than 30K and 30K-2 may exist, these results strongly suggest that nCL-2 functions without a regulatory subunit in the stomach. Considering that the penta-EF-hand domain of mCL is more similar to nCL-2 in its primary sequence than to that of nCL-4, but that nCL-4, but not nCL-2, forms a heterodimer with 30K, the dimer formation between penta-EF-hand domains appears to be governed by complicated structural properties that cannot be predicted from the primary sequence.
Importantly, nCL-2 exists not only as a monomer, but also as a homo-oligomer, and the oligomerization is most likely through domain III. The suppressive effect of oligomerization on nCL-2 activity is possible to be related to regulation mechanisms of nCL-2 activity. In other words, as 30K and 30K-2 act as activity suppressors for conventional calpains and nCL-4, oligomerization of nCL-2 may function for the same purpose without 30K or 30K-2. This property has never been reported for any kind of calpain under any experimental conditions. However, although the oligomerization was observed even in cells exogenously expressing nCL-2 near the physiological levels, further analysis is required to conclude whether or not native nCL-2 forms oligomers in the stomach.
The homo-oligomer formation through domain III may be supported by the fact that the three-dimensional structure of domain III resembles tumor necrosis factor ␣ (20), which forms a homotrimer (40), although the trimerizing C2-domain has not been reported.
Very recently, it was reported that minicalpains 1, 2, and 9 (domain II) in the presence of Ca 2ϩ are structurally different despite their highly conserved sequences, suggesting that typi-cal calpains have rather divergent mechanisms for their activation and regulation (41). It is possible that there are structural modifications specific to nCL-2 oligomerization involving inter/intra-domain interactions among domains I, II, and III of nCL-2.
Our recent study highlighted a role of nCL-2 in membrane trafficking via its location at the Golgi and interaction with a coatomer subunit of vesicles derived from the Golgi (28). This function implies a possible physiological function of the nCL-2 oligomerization: one nCL-2 molecule could bind directly to scaffold proteins at the membrane, whereas another nCL-2 in the oligomer, could appropriately proteolyze its substrates at the membrane, such as ␤-COP. One of the most important issues to be resolved is whether the monomer-oligomer transition is constitutive or inducible in vivo, and, if inducible, when it is induced. The answers to such questions will shed light on the physiological functions of nCL-2 and are emerging future issues.