Functional Properties of Recombinant Calpain I and of Mutants Lacking Domains III and IV of the Catalytic Subunit*

The catalytic subunit (L-μCANP) of human calpain I (μCANP, the high Ca2+ affinity form) and two of its mutants were expressed in Escherichia coli or using the baculovirus Sf9 system. The mutants lacked domain III (L-μCANPΔ3) and the calmodulin-like domain IV (L-μCANPΔ4), respectively. The bacterially expressed proteins were solubilized from the inclusion bodies and refolded with polyethylene glycol. In Sf9 cells, co-expression of the inhibitor calpastatin was necessary to prevent autolysis of L-μCANP, whereas co-expression of the regulatory subunit enhanced it. Only very low levels of mRNA of the truncated form L-μCANPΔ4 were found in bacmid-transfected Sf9 cells, and it proved impossible to isolate this mutant using the baculovirus expression system. While the apparentK m (Ca2+) of freshly isolated human erythrocyte μCANP was about 60 μm, the recombinant monomeric forms L-μCANP and L-μCANPΔ3 required 65–215 and 400–530 μm Ca2+, respectively. Bacterially expressed L-μCANPΔ4 was Ca2+-independent; the presence of inhibitors during its renaturation was necessary to prevent its autolysis. A chimeric form (L-μmCANP) composed by domains I–III of μCANP and domain IV of calpain II (mCANP, the low Ca2+affinity form) was also expressed in Sf9 cells. This mutant required less Ca2+ (about 50 μm) than native erythrocyte calpain for half-maximal activity and had the highest specific activity of all calpains tested. Domain III proved unnecessary for the activity of the recombinant catalytic subunit, but its absence raised the K m (Ca2+) and removed its inactivation at high Ca2+ concentrations. All recombinant proteins were active as monomers in polyethylene glycol-containing buffers; the in vitro association with the regulatory subunit enhanced only slightly the V max and the Ca2+ dependence of the expressed proteins. Activation by Ca2+ promoted the separation of the two subunits of the expressed recombinant proteins.

The calcium-activated neutral cysteine proteinase calpain (CANP) 1 is a heterodimer of a catalytic 80-kDa subunit and a regulatory 30-kDa subunit (1)(2)(3). The physiological role of CANP is not fully established, but its participation in events such as cell division, signal transduction, and long term potentiation has been suggested (4 -7). Calpains have also been proposed to play a role in various pathological processes associated with altered protein metabolism and/or altered calcium homeostasis (8,9). Two major isoforms of the protease are known, CANP (or calpain I) and mCANP (or calpain II), which require about 50 and 300 M Ca 2ϩ for half-maximal activity in vitro, respectively (10,11). More recently, a third calpain isoform, termed p94, which is specific for skeletal muscle and appears to be partly located in the cell nucleus, has also been identified (12,13). The large subunit of CANP is composed of four domains (I-IV) as follows: domain I probably acts as a "repressor" of the catalytic activity and is cleaved at its N-terminal portion during the autolytic activation of the protease (3,14); domain II contains the essential cysteine and histidine residues of the active site (15); the role of domain III, which has no obvious sequence homology to other proteins, is still obscure; domain IV is a calmodulin-like domain that contains four putative Ca 2ϩ -binding sites corresponding to the helixloop-helix Ca 2ϩ -binding motif typical of calmodulin-like proteins (15)(16)(17). The small subunit consists of an N-terminal glycine-rich hydrophobic domain V and of a calmodulin-like domain IVЈ that is homologous to domain IV of the catalytic subunit. The association of the two subunits to form the heterodimer results from the interaction of domains IV and IVЈ (8). The function of the 30-kDa subunit is not established, but Ca 2ϩ appears to dissociate it from the catalytic subunit, activating it (18). Ca 2ϩ also promotes the autoproteolysis of the enzyme, cleaving amino acids from both subunits (11,19). The autoproteolytic processing is linked to the activation of CANP (two degradation products of about 78 and 76 kDa are produced seconds after the incubation of the enzyme with Ca 2ϩ ), but recent work has shown that nonautoproteolyzed CANP may also be active (20).
Reports have appeared describing the purification of both heterodimeric CANP isoforms from several tissues, e.g. muscle, liver, and kidney (2), but the isolation of large amounts of pure and active protein is laborious. cDNAs for the large and small subunits of calpains I and II have been cloned (21)(22)(23)(24), and the expression of the large subunit of calpain II (22) and of an N-terminally truncated 21-kDa variant of the small subunit (24) has been achieved in Escherichia coli. Domain IV of the large subunit of calpain I has also been expressed (21). In a recent publication the expression of active human calpain I using the baculovirus system has been described (25).
The work described in this contribution reports the expression of the catalytic subunit of calpain I (L-CANP) and of two mutants lacking domains III or IV (L-CANP⌬3 and L-CANP⌬4) in E. coli and using the baculovirus expression system. A chimeric form (L-mCANP), in which domain IV of CANP was replaced by that of mCANP, was also expressed to study the role of domain IV in the Ca 2ϩ requirement for the activation of the enzyme.

EXPERIMENTAL PROCEDURES
Materials and General Methods-Restriction enzymes were obtained from New England Biolabs (Schwalbach, Germany). For the PCR purification, midi-plasmid purification and gel extraction kits from Qiagen AG (Basel, Switzerland) were used. The Expand TM Long Template PCR System kit was from Boehringer Mannheim (Rotkreuz, Switzerland). The oligonucleotides were purchased from Microsynth (Balgach, Switzerland). The cDNAs for the large subunit of rat m-calpain and for the truncated small subunit of rat calpain were kindly donated by Dr. J. S. Elce (Queen's University, Kingston, Ontario, Canada). The cDNA for the large subunit of human -calpain was obtained as described previously (21). The pGEM-T vector was from Promega (Madison, WI), and the E. coli host cell XL-1 Blue was from Stratagene (Zurich, Switzerland). The plasmids used for bacterial expression (pET-3d and pET-9d) and the E. coli strain BL21(DE3)pLysE were from Novagen (Madison, WI). The Bac-to-Bac TM System of Life Technologies, Inc. and Spodoptera frugiperda cells (Sf9) from Pharmingen (San Diego, CA) were used for the expression with baculovirus. The fluorogenic substrate Suc-Leu-Tyr-AMC and calpeptin (Cbz-Leu-nLeu-H) were from Novabiochem (Lä ufelfingen, Switzerland). Cbz-Leu-Leu-Tyr-CHN 2 was synthesized as described previously (26). Protein concentrations were determined using the bicinchoninic acid Protein Assay (Pierce, Lausanne, Switzerland) with bovine serum albumin as standard. SDS-PAGE was carried out according to Laemmli (27). The polyclonal goat anti-(platelet mCANP) serum was kindly provided by Dr. A. H. Schmaier (University of Michigan, Ann Arbor), and the monoclonal anti-(bovine calpastatin) antibodies were from SWant (Bellinzona, Switzerland). Expression of the 21-kDa C-terminal domain of the small subunit of rat calpain was carried out according to Graham-Siegenthaler et al. (24). Human erythrocyte calpain I was isolated by affinity chromatography on immobilized calpastatin peptides as previously described (28).
Cloning and Expression of Human L-CANP, L-CANP⌬3, and L-CANP⌬4 in E. coli-DNA fragments corresponding to the fulllength L-CANP (domains I-IV), L-CANP⌬4 (domains I-III), domains I-II, and domain IV were prepared by PCR of the cDNA for the large subunit of human -calpain subcloned in pUC118, using specific primers with restriction site overhangs (29) (Fig. 1A). The PCR products were inserted into the pGEM-T vector and transformed in XL-1 Blue. After plasmid purification and digestion of the subcloned fragments with the desired restriction enzymes, the DNA fragments were inserted into pET-3d. The resulting clones for L-CANP, L-CANP⌬3, and L-CANP⌬4 were expressed in BL21(DE3)pLysE as described by Ma et al. (21). Cells were harvested after induction at 30°C, resuspended in lysis buffer (20 mM sodium phosphate buffer, pH 7.3, 500 mM NaCl, 1% (v/v) Triton X-100, 0.2 mM PMSF, and 25 mM dithiothreitol), and then lysed by mild sonication for 1 min at 20% pulse (Branson sonifier B-30, Branson Sonic Power Co., Danbury, CT). After clearing the lysate by centrifugation at 5000 ϫ g for 15 min at 4°C, both soluble and pelleted protein fractions were analyzed by SDS-PAGE.
Protein Extraction from the Inclusion Bodies-Inclusion bodies, containing practically all of the expressed proteins, were treated according to a modified method of Russel-Harde et al. (30). Briefly, they were suspended at 25 mg of protein/ml in washing buffer (100 mM Tris-HCl, pH 8.0, 6 M urea, and 25 mM dithiothreitol), mixed for 30 min at room temperature, and centrifuged at 500 ϫ g for 15 min at 4°C. The pellet was resuspended at 50 mg of protein/ml in a solution of 2% (w/v) Zwittergent 3-14 (Calbiochem, Lucerne, Switzerland) in deionized water, mixed for 30 min at room temperature, and centrifuged at 5000 ϫ g for 15 min at 4°C. The washed inclusion bodies were solubilized with denaturing buffer (20 mM Tris-HCl, pH 7.5, 6 M GdnHCl, 1 mM EDTA, 1 mM EGTA, 5 mM 2-mercaptoethanol, 150 mM NaCl) at a concentration of 10 -50 g of protein/ml. The renaturation of each recombinant protein was carried out by dialysis, according to a modification of the method described by Yoshizawa et al. (31). The proteins denatured as described above were dialyzed at room temperature first against 200 volumes of buffer A (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 5 mM 2-mercaptoethanol, 100 mM NaCl) containing 5% (v/v) glycerol and 3 M GdnHCl for 6 h and then against the same buffer with 10% (v/v) glycerol, 0.1% (v/v) Triton X-100, 0.05 M GdnHCl, and PEG 4000 for an additional 6 h. The polyethylene glycol was added to the dialysis buffer at a molar ratio to the recombinant protein of about 10 to 1 (assuming that the inclusion bodies only contain the recombinant proteins). For L-CANP⌬4, 10 M calpeptin was always added to the dialysis buffers. Refolding of the dialyzed proteins was then continued by keeping them at room temperature.
Co-expression of the Large and the Truncated Small Subunits of Calpain in E. coli-The clone for the wild-type L-CANP was transferred from the ampicillin-resistant expression vector pET-3d into the kanamycin-resistant pET-9d. The resulting plasmid was transformed together with the cDNA for the truncated small subunit (in an ampicillin-resistant vector) in BL21(DE3)pLysE, and the cells were plated onto LB/agar containing both ampicillin and kanamycin. The expression at 30°C and the separation of the soluble and pelleted protein fractions were performed as described above.
Baculovirus Production-PCR amplification was performed as described above, with the exception that EcoRI restriction sites were inserted instead of NcoI sites (Fig. 1B). Moreover, the DNA for the large subunit of rat m-calpain was used to amplify the fragment coding for its domain IV, and another amplification of the region encoding domains I-III of CANP was carried out to add a HindIII site necessary for ligation with the cDNA for the domain IV of mCANP (Fig. 1B). The rat small subunit was used to amplify the portion coding for the 21-kDa protein. A full-length human calpastatin cDNA was prepared by fusing two previously isolated partial cDNA clones pWICS46 and cs19 (32,33) at the XhoI site in domain 2. The cDNA was modified by adding the Kozak's consensus translation initiation sequence (34) and the HA1-tag sequence (YPYDVPDYASL) at the N terminus of its coding region for monoclonal antibody recognition (35). The 2.3-kilobase pair SphI/KpnI fragment of the resulting calpastatin cDNA, designated pTicHACS, and the amplified products (cloned in pGEM-T) were inserted into the plasmid pFASTBAC1 to construct the six different baculovirus transfer vectors. The vectors were then used to generate the bacmids containing the individual recombinants in Max Efficiency DH10Bac cells by following the protocol of the manufacturer of the Bac-to-Bac TM system. Recombinant baculovirus particles were produced by transfecting Sf9 insect cells in monolayer with each bacmid DNA using the Cellfectin reagent. They were collected and used for infection of other Sf9 cells as described in the manufacturer's protocol.
Characterization of the Proteins Expressed in Sf9 Cells-Sf9 cells were seeded in 6-well plates at about 1.0 ϫ 10 5 cells/cm 2 , baculoviruses were added (separately or in pairs) after cell attachment at an m.o.i. of 5.0, and the cells were harvested at 350 ϫ g for 10 min 42 h after infection. The cell pellets were resuspended in hypotonic buffer (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 5 mM 2-mercaptoethanol, 0.1 mM PMSF, 0.1% (v/v) Triton X-100) and centrifuged at 15,000 ϫ g for 10 min at 4°C. The supernatants were used for the determination of protein concentrations and immunoblot analysis with antibodies against calpain and calpastatin.
Preparation of Crude Protein from Sf9 Cells-To produce recombinant proteins for purification, monolayer cultures of 1.2-1.4 ϫ 10 7 insect cells in 150-cm 2 flasks were infected (separately or in pairs) at an m.o.i. of 5.0 and harvested 42 h after infection by centrifugation at 350 ϫ g for 10 min. Pelleted cells were treated according to a modification of the method described by Meyer et al. (25). They were resuspended in homogenizing buffer (20 mM Tris-HCl, pH 7.5, 250 mM sucrose, 2 mM EDTA, 2 mM EGTA, 5 mM 2-mercaptoethanol, 0.1 mM PMSF, 0.1% (v/v) Triton X-100), broken with 10 strokes in a Dounce homogenizer (Wheaton, Millville, NJ), and centrifuged at 2,000 ϫ g for 10 min at 4°C. The supernatants were incubated for 1 h with final concentrations of 100 mM NaCl and 5 M PEG 4,000. Then they were re-centrifuged at 38,700 ϫ g for 1 h at 4°C, and the resulting new supernatants were filtered through glass wool to obtain crude extracts. These were subjected to a 40% ammonium sulfate precipitation and centrifuged at 13,600 ϫ g for 30 min at 4°C. The resulting protein pellets were resuspended in resuspension buffer (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 0.02% (w/v) NaN 3 , 5 mM 2-mercaptoethanol, 85 mM NaCl, 5 M PEG 4,000, 0.1% (v/v) Triton X-100) and dialyzed overnight against the same buffer. The samples were then loaded onto a DEAE-Sepharose CL-6B column (Pharmacia, Uppsala, Sweden) pre-equilibrated with resuspension buffer. The recombinant proteins were eluted from the column with resuspension buffer containing 300 mM NaCl (final). Fractions containing the expressed proteins were pooled on the basis of SDS-PAGE and dialyzed against buffer B (50 mM Tris acetate, pH 7.5, 0.02% (w/v) NaN 3 , 100 mM NaCl, 5 M PEG 4,000, 0.1% (v/v) Triton X-100) containing 1 mM EGTA.
Purification of the Recombinant Proteins Expressed in Insect Cells-When Sf9 cells were co-infected with viruses for calpastatin and calpain recombinants, the expressed inhibitor (HA1-CALST) was removed using a modification of the procedure of Wadzinski et al. (35). Briefly, 5 g of anti-HA (12CA5) antibodies (Boehringer Mannheim, Rotkreuz, Switzerland) were added to these preparations, and the mixtures were rotated end-over-end overnight at 4°C. About 100 l of protein A-Sepharose (Pharmacia, Uppsala, Sweden) were added, and the incubation at 4°C was continued for 2 h. The beads were washed with a low salt buffer (50 mM Tris-HCl, pH 7.5, 1 mM EGTA) to elute the calpain recombinants, and HA1-CALST remained bound to the matrix. To the released proteins, final concentrations of 100 mM NaCl, 5 M PEG 4,000, and 0.1% (v/v) Triton X-100 were added. The mixtures were then dialyzed overnight against buffer B containing 1 mM EGTA. These samples and the crude enzyme preparations, obtained as described above, were incubated with 100 M final concentration of calpeptin for 30 min at 4°C. Then an equal volume of buffer B containing CaCl 2 was added dropwise with stirring to obtain a total free calcium concentration of 0.8 mM, and the incubation was continued for 1 additional h. The inhibitor-modified enzyme preparations were loaded onto a B27-Sepharose calpain affinity column (28) pre-equilibrated with buffer B containing 2 mM CaCl 2 . The column was washed extensively with the same buffer to remove unbound proteins, and the calpain recombinants were eluted with 5 mM EGTA in buffer B. The purified proteins were concentrated, and the residual inhibitor was removed with several buffer changes (buffer B containing 1 mM EGTA, 1 mM EDTA, 5 mM 2mercaptoethanol, 5 mM dithiothreitol) in Ultrafree filters (Millipore, Bedford, MA) prior to the determination of protein concentrations and the assay of calpain activity.
In Vitro Association of the Subunits to Form Heterodimeric Calpain-The truncated 21-kDa small subunit was added to the refolding mixtures for L-CANP or L-CANP⌬3 purified from the inclusion bodies as described above, or to L-CANP, L-CANP⌬3, or L-mCANP produced with the baculovirus expression system. The small subunit was in 10 -20-fold molar excess of the recombinant catalytic subunits. The protein mixtures were kept at room temperature, and their activities were measured.
Dissociation of the Subunits-The dissociation of subunits by Ca 2ϩ was studied using a chromatographic procedure. Heterodimeric calpain (L-CANP with the 21-kDa protein) was incubated with calpeptin, loaded onto the B27-Sepharose calpain affinity column (28) in the presence of Ca 2ϩ , and eluted with EGTA as described above.
Calpain Assay-Calpain activity was assayed fluorimetrically according to a modified method of Sasaki et al. (36), with a Spex Fluorolog 1680 (Spex Industries, Inc., Edison, NJ). In this assay, 40-l portions of the samples (300 -400 nM) were added to 500 l of total volume of assay buffer (50 mM Tris acetate, pH 7.5, 0.25 mM Suc-Leu-Tyr-AMC, 5 mM cysteine, 0.02% (v/v) 2-mercaptoethanol). Ca 2ϩ was added in a 10-l volume to initiate the hydrolysis of the fluorogenic substrate at 22°C. Excitation was at 370 nm, and the emission of the released AMC was measured at 460 nm. In the case of L-CANP⌬4 (produced in E. coli), it was necessary to reduce the amount of calpeptin in the 40-l portions; the reversible inhibitor was removed by ultrafiltration with several buffer changes in Ultrafree filters. The volume of the retained samples was adjusted to 40 l.

Preparation of the Recombinants of L-CANP in E. coli-The
cDNA fragments for the wild-type L-CANP and the mutants L-CANP⌬3 and L-CANP⌬4 were prepared as illustrated in Fig. 1A, by subcloning PCR products first in the pGEM-T vector and then in the expression vector pET-3d. All three recombinant proteins accumulated as inactive insoluble aggregates. The inclusion bodies containing the proteins expressed at 30°C were purified through a combination of centrifugation steps. At the end of these washing steps, highly enriched (about 95-97%) preparations of the recombinant proteins were obtained (Fig.  2). To achieve solubilization, in the final step of the purification procedure (30) the pH was initially shifted to 12.0 and after some minutes brought back rapidly to 8.0. However, pH 12.0 irreversibly inactivated calpain, as established in control experiments on the purified erythrocyte CANP (EC CANP). Therefore, the solubilization step at pH 12.0 was omitted, and the isolated inclusion bodies were instead solubilized with guanidine hydrochloride (GdnHCl).
Denaturation and Refolding of the Bacterially Expressed Proteins-During the treatment with 6 M GdnHCl and prior to the refolding attempts by dialysis, the recombinant proteins were diluted to 10 -50 g/ml (37,38). As already reported for other proteins (39), a non-denaturing concentration of GdnHCl (0.05 M) in the second dialysis buffer had a beneficial effect on the folding efficiency. Further purification of the solubilized proteins, e.g. by chromatography on DEAE-Sephacel, was unnecessary, since the procedure for the isolation of the inclusion bodies yielded a rather homogeneous preparation (Fig. 2). However, the activity of the recombinant proteins was less stable than that of native erythrocyte CANP; the activity increased to a maximum at day 2 after the two dialysis steps, after which the protein started to aggregate and to lose activity (not shown). The renaturation procedure was also applied to the two mutants, but in the case of L-CANP⌬4 it was necessary to perform the dialysis in the presence of the reversible inhibitor calpeptin to prevent the autoproteolysis of the protein and its disappearance from the dialysis tube (Fig. 2, lane 15). Attempts to increase the specific activity of the recombinant wild-type L-CANP by performing the refolding assay in the presence of the 21-kDa N-terminally truncated version of the regulatory subunit yielded only minor improvements (Table I). The recombinant truncated regulatory subunit used in these experiments was from rat liver, since the two subunits refold properly and recover activity after denaturation also when isolated from different species (40).
Simultaneous Expression of L-CANP with the N-terminal Portion of the Small Subunit in E. coli-An isolated colony of BL21(DE3)pLysE cells containing plasmids encoding both the large and the 21-kDa subunit was submitted to expression with IPTG. As shown in Fig. 3, the truncated small subunit was expressed in soluble form, whereas essentially all of the large subunit was recovered in the insoluble fraction.
Production of Recombinant Baculoviruses-The cDNA constructs used to prepare recombinant viruses expressing L-CANP, L-CANP⌬4, L-CANP⌬3, L-mCANP, CALST, and the 21-kDa subunit were obtained after transposition of the clones into bacmids. These were used to transfect Sf9 cells that subsequently released the recombinant baculoviruses.
Only the viral titer for the truncated form L-CANP⌬4 was nearly zero. Northern hybridization showed, however, that transcription of the mRNA encoding the truncated enzyme occurred but to a much lower extent than for the other proteins (not shown).
Characterization of the Proteins Expressed Using the Bacu-lovirus System-The expression of the recombinant catalytic subunits was examined in immunoblots of proteins of infected Sf9 cells (Fig. 4A). An antiserum against platelet mCANP detected L-CANP, L-CANP⌬3, and L-mCANP but failed to detect L-CANP⌬4. The latter evidently was not expressed, as expected from the low level of baculovirus production. Antibodies against calpastatin detected HA1-CALST (Fig. 4B). As reported by Meyer et al. (25), heterogeneity of the large subunit was observed; in addition to the 80-kDa subunit, two polypeptides of about 78 and 76 kDa were also present, the latter in smaller amounts (Fig. 4A, lane 1). When L-CANP was co-  a The maximal specific activity is defined as the pmol of Suc-Leu-Tyr-AMC hydrolyzed by 12.5 pmol of the protease within 1 min at room temperature.
b To calculate the relative maximal activity of every sample, the maximal specific activity of native EC CANP was considered as 100%.
c Native EC CANP not subjected to the denaturation/renaturation procedure. d EC CANP denatured with GdnHCl and renatured without PEG, NaCl, and Triton X-100. e PEG-EC CANP, native EC CANP denatured with GdnHCl and renatured with PEG. NaCl, and Triton X-100.
g The recombinant small subunit was added to the refolding buffer containing PEG, NaCl, and Triton X-100.
h L-CANP, L-CANP⌬3, and L-mCANP co-expressed with calpastatin (except for the case of L-CANP⌬3) in Sf9 cells. Calpastatin was then removed during the purification procedure as described under "Experimental Procedures." expressed with the 21-kDa subunit, the 78-kDa protein was not detected, and the 76-kDa product became the most evident (Fig. 4A, lane 2). The expressed chimeric form L-mCANP had obvious bands at 80, 78, and 76 kDa (Fig. 4A, lane 3). Also in this case co-expression with the 21-kDa subunit induced the disappearance of the 78-kDa polypeptide (Fig. 4A, lane 4). The intensity of the two remaining bands was weak, indicating autolysis of the chimeric form. At variance with this, the expressed L-CANP⌬3 only showed one immunoreactive polypeptide of 55 kDa, also when co-expressed with the small subunit (Fig. 4A, lanes 5 and 6). The 78-and 76-kDa polypeptides failed to form when L-CANP and L-mCANP were co-expressed with CALST (Fig. 4A, lanes 7 and 8). Thus, coexpression with calpastatin apparently prevented autolysis of recombinant calpains in the Sf9 cells. The expression of CALST was monitored with antibodies against bovine calpastatin (Fig.  4B, lanes 2 and 3).
Purification of Recombinant Proteins from Sf9 -The recombinant proteins were produced by 1.4 ϫ 10 7 infected Sf9 cells grown in monolayer, and purification of L-CANP, L-mCANP, and L-CANP⌬3, or of co-expressed L-CANP-CALST, L-mCANP-CALST, and L-CANP⌬3-21-kDa was performed as described under "Experimental Procedures." The final purified products are shown in Fig. 5. The extracts containing co-expressed CALST were treated to remove the CALST (see the "Experimental Procedures") prior to the purification step on the B27 affinity column. The purification yielded 25-35 g of 90 -95% pure recombinant enzymes from 1.4 ϫ 10 7 cells.
Ca 2ϩ -dependent Activity-The Ca 2ϩ dependence of the recombinant calpains was studied fluorimetrically with the synthetic substrate Suc-Leu-Tyr-AMC using samples of purified erythrocyte CANP as controls (Table I). The recombinant proteins used were produced under conditions that yielded no autoproteolysis products. Thus, as shown in Fig. 4A for the experiment on calpains produced in Sf9 cells, L-CANP, L-CANP⌬3, and L-mCANP were co-expressed with CALST (except for the case of L-CANP⌬3) and were subjected to activity measurements after removal of CALST. As mentioned above, the level of activity for the bacterially expressed proteins peaked 2 days after dialysis against PEG, NaCl, and Triton X-100. Half-maximal activity of the native erythrocyte CANP and of the same enzyme submitted to the denaturation/renaturation treatment required 60 -65 and 100 -105 M Ca 2ϩ , respectively (Fig. 6). The V max of the purified erythrocyte CANP reached about 80% that of the untreated native enzyme when denatured and refolded in the presence of PEG, NaCl, and Triton X-100 and decreased to about 60% if the GdnHCl treatment followed by refolding was performed in the absence of PEG. The V max of L-CANP expressed in E. coli and renatured under optimal conditions reached 65-70% that of native erythrocyte CANP and 90 -93% when expressed using the baculovirus system. The apparent K m (Ca 2ϩ ) was 65-70 M in the case of the virally produced protein and 205-220 M for the protein expressed in bacteria. Higher Ca 2ϩ concentrations dramatically decreased the activity, as also observed for erythrocyte CANP. The V max of mutant L-CANP⌬3 produced in E. coli was 52-56% that of the native erythrocyte enzyme (K m (Ca 2ϩ ) of 510 -540 M) and increased to 85-90% for L-CANP⌬3 from Sf9 cells (K m (Ca 2ϩ ) of 380 -420 M). At variance with erythrocyte CANP and the expressed L-CANP, the activity of mutant L-CANP⌬3 did not decline at high Ca 2ϩ concentrations. In vitro association of L-CANP and L-CANP⌬3 with the purified recombinant small subunit slightly increased the activity and calcium dependence of the recombinant catalytic subunits produced in insect cells (Table  I). The maximal activity of mutant L-CANP⌬4 expressed in E. coli and renatured in the PEG-containing buffer corresponded to only 42-45% that of the native erythrocyte calpain. The mutant was Ca 2ϩ -insensitive (range tested, 0 -20 mM Ca 2ϩ ). Calpeptin had to be present in the dialysis buffer when purifying L-CANP⌬4. Removal of the inhibitor by dialysis prior to performing the activity assay led to the rapid autoproteolysis of this mutant and to its disappearance from the dialysis tube (not shown). The activity was thus measured after diluting the reversible inhibitor in the assay buffer by ultrafiltration. Possibly, then, the lower activity of this mutant could have been due to its partial degradation during the ultrafiltration step. The V max of the chimera L-mCANP expressed using the baculovirus was 92-95% that of native erythrocyte CANP, and in this case the addition of the 21-kDa subunit brought the activity to 100 -105%. Unexpectedly, the Ca 2ϩ requirement for halfmaximal activity of L-mCANP was even lower than that of erythrocyte CANP, i.e. 49 -53 M, and was decreased to 42-45 M by in vitro association with the small subunit.
Dissociation of the Subunits upon Ca 2ϩ Activation-The B27 peptide (the product of exon 1B of the CALST gene) is known to bind only the catalytic subunit. As already observed (28), the small subunit failed to bind to the B27 peptide column. Heterodimeric calpain in buffer B containing Ca 2ϩ and 100 M calpeptin was loaded onto the B27 column (Fig. 7, lane 1) which was then washed thoroughly with a Ca 2ϩ -containing buffer. The large subunit was retained, whereas the small subunit was lost in the flow-through (Fig. 7, lane 2). The large subunit was eluted with an EGTA-containing buffer (Fig. 7, lane 3). The isolated catalytic subunit showed full enzyme activity (not shown).
Autoproteolytic Processing of the Recombinant Calpains-The four recombinant monomers (L-CANP, L-mCANP, and L-CANP⌬3 expressed in Sf9 cells and L-CANP⌬4 produced in E. coli) underwent autoproteolysis when incubated for 2 min with 1 mM Ca 2ϩ , L-CANP⌬4 was rapidly converted to species with lower molecular weight also in its absence. The degradation of the four proteins was completely inhibited by Cbz-Leu-Leu-Tyr-CHN 2 (Fig. 8). In all cases, the typical three-band pattern of limited autoproteolysis of the catalytic subunit, resulting from cleavage at two N-terminal sites (3,14), was observed.

DISCUSSION
This study describes the expression and properties of the recombinant catalytic subunit (L-CANP) of -calpain, of two mutants lacking domain III (L-CANP⌬3) or domain IV (L-CANP⌬4), and of a chimeric form of theand m-calpain (L-mCANP). When expressed in E. coli, the proteins were always produced as inactive aggregates in the inclusion bodies, possibly a useful device to prevent cell damage, especially in the case of the Ca 2ϩ -independent L-CANP⌬4. Co-expression of the large CANP subunit with the regulatory subunit was also attempted in E. coli to improve the production of soluble and active catalytic subunit (24,25). The attempt was unsuccessful, i.e. the small subunit was expressed in soluble form but failed to prevent the aggregation of L-CANP and did not improve its expression.
The recently reported successful expression of biologically active and stable monomeric and heterodimeric human -calpains in eukaryotic cells (25) prompted the extension of the work to the baculovirus Sf9 cell expression system. However, the truncated mutant L-CANP⌬4 could not be expressed in Sf9 cells, and baculovirus particles for this mutant were not produced. The mRNA for this mutant was present after transfection of the insect cells, but its amount was much lower than that for L-CANP or L-CANP⌬3. The reasons for the very low level of transcription of L-CANP⌬4 is not understood at the moment, except that one could speculate that the production of constitutively active L-CANP⌬4 could be lethal to eukaryotic cells. Possibly, insect cells activate a down-regulating system to prevent the formation of dangerous viral DNA.
The autoproteolytic process leading to the production of the 78-and 76-kDa forms of expressed L-CANP was enhanced in the baculovirus system by co-expressing the 21-kDa subunit. At variance with Meyer et al. (25), who observed increased accumulation of the 80-kDa protein upon co-expression with the small subunit, in the present work the 21-kDa subunit was found to improve the solubility of the catalytic subunit but not its stability. In fact, in co-expressing cells, L-CANP underwent more pronounced autolysis with the appearance of signif- FIG. 8. Autoproteolytic processing of the recombinant proteins. SDS-PAGE was carried out on the four recombinant calpains after incubation at room temperature for 2 min in the presence (ϩ) or absence (Ϫ) of 1 mM Ca 2ϩ . Incubation with 2 M Cbz-Leu-Leu-Tyr-CHN 2 prior to addition of Ca 2ϩ is indicated by an asterisk. L-CANP⌬4 (E. coli) was previously subjected to an ultrafiltration step to dilute calpeptin, whereas the other three catalytic monomers (Sf9) were subjected to in vitro association with the rat regulatory 21-kDa subunit as described under "Experimental Procedures." icant amounts of the 76-kDa form. On the contrary, co-expression with calpastatin prevented the autoproteolytic process.
The chimeric variant L-mCANP underwent an autoproteolytic pattern similar to that of L-CANP, i.e. approximately equivalent amounts of the 80-, 78-, and 76-kDa products were observed. Also the expressed chimera was less stable when co-expressed with 21 kDa; the disappearance of the 78-kDa polypeptide could be taken as an indication that the small subunit somehow destabilizes this intermediate. The presence of degradation products of L-CANP and L-mCANP in the insect cells could be due to the fact that the overexpressed calpain monomers, exceeding the amount of endogenous calpastatin, escape its control and therefore undergo autoproteolysis in the cytoplasm; the presence of CALST in Sf9 cells was confirmed by reverse transcription-PCR work (not shown). The more pronounced autolysis of the large subunit, when the recombinant small subunit was co-expressed, may be caused by the small subunit-mediated translocation of the protease to membranes, where calpain activation is enhanced.
When exploring favorable conditions for the refolding of the bacterially expressed proteins, the latter were allowed to interact in vitro with the 21-kDa subunit (except for L-CANP⌬4, since the association between calpain subunits occurs via domains IV and IVЈ (41)). Although this had no significant effect on the renaturation of the proteins produced in E. coli, a slight improvement was observed for the proteins expressed using the baculovirus system. Perhaps the proteins produced in eukaryotic cells may be in a different conformational state with respect to those expressed in E. coli. In agreement with the finding of Yoshizawa et al. (31), the large subunit alone was fully active. Therefore, the small subunit is important but not necessary for calpain activation; it could perhaps contribute to the proper folding of the catalytic subunit. Since Ca 2ϩ activation of heterodimeric calpain is strictly correlated to subunit dissociation, one could suggest that the heterodimeric form is forced to undergo dissociation upon Ca 2ϩ binding to produce the active monomer. The suggestion would be compatible with previously published studies (18,28). A recent report, however, argues for the continued association of calpain subunits in the presence of Ca 2ϩ and during the proteolysis of substrates (42).
Despite the large literature on the structure and function of calpains, little is known about the roles of domains III and IV of their native 80-kDa subunits. The experiments presented in this contribution have shown that the removal of domain III did not induce significant loss of calpain activity but altered the sensitivity of the protein to Ca 2ϩ , i.e. it increased very substantially the apparent K m (Ca 2ϩ ) of the mutant. It would thus be attractive to speculate that domain III mediates the Ca 2ϩactivating signal to calpain, i.e. upon binding of Ca 2ϩ to domain IV, domain III would amplify the activating message to the active site in domain II. Possibly a change in the tertiary structure of domain III could increase the accessibility of the active site.
As for domain IV, its removal has fortified the proposal that this domain is the Ca 2ϩ receptor in the molecule. The finding that the replacement of domain IV of CANP with that of mCANP improved the affinity of the protein for Ca 2ϩ was unexpected, given the considerably lower Ca 2ϩ affinity of mCANP. This suggests that domain IV may be important in the interaction of the molecule with Ca 2ϩ but not as the only domain in the large subunit that mediates the Ca 2ϩ response. As for the finding that L-CANP, L-mCANP, and L-CANP⌬3 were all active and Ca 2ϩ -dependent in the absence of the small subunit, it clearly indicates that Ca 2ϩ , in addition to its function in dissociating the two calpain subunits, also has other roles; it could, for example, induce conformational changes of the large subunit that may be essential for its activation.