Characterization of Two Recombinant Drosophila Calpains CALPA AND A NOVEL HOMOLOG, CALPB*

We have sequenced the cDNA of a novel Ca 2 1 -acti-vated cysteine proteinase (calpain) from the fruit fly, Drosophila melanogaster . The predicted protein, designated as CALPB, shows high similarity to the previously identified Drosophila calpain, CALPA. The two proteins were expressed in Escherichia coli and purified to homogeneity by metal-chelate affinity chromatography either from inclusion bodies or from the bacterial cytosol. Both enzymes were Ca 2 1 -dependent proteinases and attained half-maximal activation in the presence of milli- molar Ca 2 1 . The activity and the rate of activation of CALPA, but not CALPB, could be activated by phos- phatidylinositol 4,5-diphosphate, phosphatidylinositol 4-monophosphate, phosphatidylinositol, and phospha- tidic acid. A truncated form of CALPA, lacking the CALPA-specific unique insertion region, has also been expressed and characterized. Although it lacked the 16-amino acid long putative membrane-anchoring seg- ment, its activation by phospholipids was similar to that of the full-length CALPA protein. The enzymes undergo N-terminal autolysis in a Ca 2 1 -dependent manner which was shown with CALPB to run parallel with enzyme activation. Moreover, fully autolyzed CALPB lacked the characteristic activation phase indicating the require- ment for autolysis upon activation of this calpain form in vitro . The analysis of the mechanism of activation in Drosophila calpains seems to corroborate the autolysis model of calpain activation. The set up in a final volume of 50 m l at room temperature in calpain puffer. Calpains were added in a final concentration of 0.1–0.3 mg/ml, the concentration of the fluorescent substrate was 1 m M and, where needed, phospholipids were added at a final concentration of 50 m g/ml. The reaction was started by the addition of CaCl 2 . Autolysis of Calpains— Autolysis of calpains was followed under conditions identical to those of the flurometric assay. The reaction was started by CaCl 2 and stopped by the addition of SDS sample buffer and immediate boiling for 2 min. The samples were analyzed on SDS- polyacrylamide gel electrophoresis using Coomassie Brilliant Blue staining. Band intensities were determined by densitometry in a Bio- Rad GelDoc 1000 video densitometer. The linear relationship of optical densities and protein quantity was checked by calibration.

The Ca 2ϩ -dependent intracellular cysteine proteinases, calpains, are important regulators of many cellular pathways (1,2). Through limited proteolysis of key cellular components they are involved in processes as distinct as nerve growth (3), muscle homeostasis (4), development (5), and memory (6). Several calpain isoforms are known in vertebrates including the ubiquitousand m-calpain and some tissue-specific isoforms, e.g. p94 (CAPN3, nCL-1) 1 (7), nCL-2 (8), and nCL-4 (9). While tissue-specific isoforms seem to be monomeric, ubiquitous calpains are associated with a small or regulatory subunit. Inver-tebrate calpain homologs have also been described, including a Schistosoma mansoni calpain (10,11), an atypical form in Caenorhabditis elegans (12) and a Drosophila melanogaster calpain, CALPA (5,13). The latter organism is particularly important in the study of Ca 2ϩ -regulated calpains, since C. elegans, which is equally accessible to genetic manipulations, seems to lack the typical, Ca 2ϩ -dependent calpains (14). Therefore the fruit fly as a model organism is a prime candidate in the attempt to clarify the role of calpains in vivo.
The finding of Drosophila CALPA was preceded by biochemical studies which revealed the existence of at least two calpain isoforms in the fruit fly (15). Another calpain isoform, CANP (probably distinct from the two partially purified ones), has been purified to homogeneity (16). However, none of the genes encoding these proteins have yet been found. The only Drosophila calpain cDNA cloned so far is coding for the CALPA protein (5,13). The protein is expressed in a few tissues, including some cells in the brain, midgut, and embryo. It carries a unique insertion sequence in the calmodulin-like domain with a 16-amino acid long putative membrane-anchoring region in it. Otherwise, its domain structure (domain I, regulatory propeptide; domain II, Cys-protease; domain III, putative regulatory; domain IV, Ca 2ϩ binding) and sequence are similar to that of conventional vertebrate calpains except for the Nterminal region of domain I which shows no homology to any other protein sequence. No small subunit, associated with CALPA or the purified CANP protein has been found in line with evolutionary studies which suggest the lack of the small subunit from protostomes (17). The CALPA protein has been expressed in E. coli and purified partially, and some of its possible in vivo substrates have been identified (18). However, apart from these studies, no biochemical data are available on the mechanism of action of Drosophila calpains with known sequences.
Here we describe the sequence of a novel fruit fly calpain, CALPB, and the bacterial expression, purification, and enzymological characterization of Drosophila CALPA and CALPB. The recombinant proteins worked as active, Ca 2ϩ -dependent proteinases. A detailed study is presented on the activation, autolysis, and lipid dependence of these proteins. The parallel nature of the enzymes' autolysis and activation and the lack of the activation period in the fully autolyzed enzyme provide insights into the mechanism of calpain activation.

EXPERIMENTAL PROCEDURES
Materials and General Methods-All chemicals were obtained from Sigma. Restriction enzymes were purchased from Promega and Sigma. Site-directed mutagenesis was carried out using a Unique Site Elimination kit from Amersham Pharmacia Biotech, following the instructions of the supplier. The pET-22b(ϩ) vector and Escherichia coli BL21 (DE3) were obtained from Novagen. The Ni-NTA resin was purchased from Qiagen. SDS-polyacrylamide gel electrophoresis was carried out according to Laemmli (19). Recombinant m-calpain was prepared using the pET-24d-80k-CHis 6 and the pACpET-24 expression vectors gener-ously provided by John S. Elce (Department of Biochemistry, Queen's University, Kingston, Ontario, Canada) as described in Ref. 20. The Drosophila CalpA cDNAs, corresponding to the long and the short alternative transcript, were the same as described in Ref. 13. f28 is coding for the short, and f32 for a partial long CALPA form. The 5Ј-end is identical in the two in vivo transcripts, the f32 clone is, however, a 5Ј-truncated partial clone.
DNA Sequencing-The partial sequence (525 base pairs) of a D. melanogaster clone from the EST data base showed similarity to the 5Ј-end of Drosophila CalpA (GenBank accession number Z46891). The clone was purchased from the I.M.A.G.E. Consortium (Clone ID LD18261; GenBank accession number AA538677) (21), and its complete nucleotide sequence has been determined using partial clones generated by restriction digestions and unique primers designed to cover the remaining gaps. The cDNA was sequenced on both strands using an ABI 373 DNA Sequencer (Applied Biosystems). The sequence has been deposited in the GenBank data base under GenBank accession number AF062404. The 5Ј-end of the cDNA was assembled from 16 further EST clones found in the GenBank data base (the GenBank accessions numbers are: AA538677, AA820523, AA541102, AA803400, AA950981, AA940775, AI239004, AA951075, AA951990, AA820314, AA979265, AA820721, AA202961, AA536318, AA539598, AA539124). The identity of the starting Met was confirmed by all of the EST clones overlapping in this region (AA820523, AA950981, AI239004, AA940775, AA951990, AA202961, AA539598, AA820721, and AA820314). These all carried an in-frame stop codon before the starting Met. The identified open reading frame coded for a novel Drosophila calpain, CALPB.
N-terminal Sequence Analysis-The samples were run on SDS gels and blotted onto polyvinylidene difluoride membranes (Sigma) for 120 min, at 250 mA in a buffer containing 10 mM CAPS, pH 11, and 10% methanol. Sequence analysis was performed on a Perkin-Elmer ABI 471A pulsed-liquid phase sequencer equipped with on-line Perkin-Elmer ABI 120 PTH analyzer and Kipp and Zonen dual-pen strip chart recorder (type BD 41). The polyvinylidene difluoride-bound proteins were sequenced employing a modified Edman degradation sequencer program.
Construction of Expression Vectors-The f28 clone, containing the short transcript of the CalpA gene, and the f32 clone, containing the long transcript truncated at the 5Ј-end, were ligated at a BamHI site to obtain the f28-32 clone with a full-length CALPA open reading frame. A frameshift mutation, present in the f32 clone was replaced with the corresponding region in f28 using PvuI sites. The open reading frame from the f28-32 clone was amplified with the CATGAATTCACATATG-GACGACTTGAGGG and CATAGTCGACCGAATATATTGTGCGCTC primers using proofreading Pfu polymerase (Stratagene). The amplified fragment was digested with NdeI-SalI and ligated into the NdeI-XhoI sites of the pET-22b(ϩ) expression vector. The ligation region of the resulting CALPA-pET-22b vector ( Fig. 1A) was checked by sequencing.
The region containing the CALPB open reading frame in the LD18261 clone was amplified using the GCCTATTCTAGACATATGC-CCTATCCAACTGGCATG and ATATCTAGATCTCGAGAGAGTAAAT-TGTTCGCTCCAGCC primers. Since Pfu polymerase yielded insufficient product, this clone was amplified using Taq polymerase. The polymerase chain reaction product was digested with NdeI-XhoI and ligated into the NdeI-XhoI sites of the pET-22b(ϩ) expression vector. The ligation region of the resulting CALPB-pET-22b (Fig. 1B) vector was checked by sequencing.
Site-directed Mutagenesis of CALPA-pET-22b-A mutant form of CALPA, lacking the inserted region, has also been created via sitedirected mutagenesis. The primers TTGGAGAAACGGTTGGGTAC-CACTGGTTTGGGTAAATC and GAGTCTCATCCACAGGTACCTTT-GCTGTTACAGG were used in a Unique Site Elimination (U.S.E.) strategy (Amersham Pharmacia Biotech), carried out according to the manufacturer's instruction. KpnI sites were introduced at the 5Ј-and 3Ј-end of the inserted region. The mutated plasmid was cut with KpnI and ligated. The identity of the mutations and the ligation site have been verified by sequencing. The resulting plasmid is referred to as CALPA-⌬Ins-pET-22b. This construct allowed expression of the truncated, 87-kDa CALPA-⌬Ins protein, having only two residues (Val and Pro, introduced with the KpnI sites) instead of a 72-amino acid-long region between Val 620 and Val 693 .
Calpain Preparation-The E. coli strain BL21(DE3) was transformed with the expression vectors using conventional techniques. Cells were grown in NZYM media containing ampicillin (100 g/ml) at 37°C, 250 rpm, until OD 600 0.7-1.0. The expression of recombinant CALPA was induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.2 mM. Incubation was continued for 3 h at 37°C, 250 rpm. The culture was cooled on ice and centrifuged at 3,000 ϫ g for 20 min at 4°C. Cells were suspended in 100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 20 mM 2-ME, 1 mM phenylmethylsulfonyl fluoride, 5 mM benzamidine, 1% Triton X-100, and sonicated for 6 ϫ 15 min at 16 m on ice. The lysate was centrifuged at 3,000 ϫ g for 20 min at 4°C. The supernatant was discarded, and the pellet was washed with 100 mM Tris-HCl, pH 7.5, 2 M urea, 20 mM 2-ME. Inclusion bodies were solubilized in 100 mM NaH 2 PO 4 /Na 2 HPO 4 , pH 7.5, 8 M urea, 20 mM 2-ME, 20 mM imidazole for 30 min at room temperature. Solubilized inclusion bodies were diluted in 10 volumes of Ni-NTA buffer I (100 mM NaH 2 PO 4 /Na 2 HPO 4 , pH 7.5, 6 M urea, 20 mM 2-ME, 20 mM imidazole) and incubated overnight at 4°C. Insoluble inclusion bodies were removed by centrifugation at 100,000 ϫ g for 60 min at 4°C. The supernatant was applied to a 10 ml of Ni-NTA resin (Qiagen) equilibrated with Ni-NTA buffer I. The resin was washed with 10 volumes of Ni-NTA buffer I. CALPA was eluted with a linear gradient of imidazole from 20 to 250 mM in Ni-NTA buffer I. The calpaincontaining fractions were pooled, and the protein was refolded in calpain buffer (10 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 0.5 mM benzamidine, 5 mM 2-ME) containing 4, 3, 2, 1, and 0 M urea, at least 8 h each.
The CALPA-⌬Ins and CALPB proteins were purified from the soluble fraction of E. coli. Cells were grown in the same conditions as described above. The proteins were expressed in the presence of 0.05 mM isopropyl-1-thio-␤-D-galactopyranoside at 20°C for 3 h, 250 rpm. Cells were suspended in 100 mM NaH 2 PO 4 /Na 2 HPO 4 , pH 7.5, 150 mM NaCl, 1 mM EDTA, 20 mM 2-ME, 1 mM phenylmethylsulfonyl fluoride, 5 mM benzamidine, and sonicated for 6 ϫ 15 min at 16 m on ice. The lysate was centrifuged at 100,000 ϫ g for 60 min at 4°C and the supernatant was applied to the Ni-NTA column. After washing with 10 column volumes of Ni-NTA buffer II (100 mM NaH 2 PO 4 /Na 2 HPO 4 , pH 7.5, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 5 mM benzamidine, 20 mM 2-ME) containing 20 mM imidazole, native calpains were eluted with a linear gradient of 20 -250 mM imidazole in Ni-NTA buffer II. To prepare CALPB from inclusion bodies, isopropyl-1-thio-␤-D-galactopyranoside was added to a final concentration of 1 mM and protein was expressed for 3 h at 37°C, 250 rpm. The purification of CALPB from inclusion bodies was done in the same way as preparation of CALPA. Unless otherwise indicated, CALPB refers to the enzyme prepared from the bacterial cytosol.
Caseinolytic Assay-The activity of calpain was measured in a caseinolytic assay as described in Ref. 16.
Fluorometric Assay-Calpain activity was also measured in a continuous fluorometric assay in a JASCO FP 777 spectrofluorometer. The cleavage of the fluorescent substrate, N-succinyl-Leu-Tyr-7-amino-4methyl-coumarin (Sigma), was recorded. The excitation wavelength was 380 nm and light emitted at 460 nm was monitored. Reactions were Autolysis of Calpains-Autolysis of calpains was followed under conditions identical to those of the flurometric assay. The reaction was started by CaCl 2 and stopped by the addition of SDS sample buffer and immediate boiling for 2 min. The samples were analyzed on SDSpolyacrylamide gel electrophoresis using Coomassie Brilliant Blue staining. Band intensities were determined by densitometry in a Bio-Rad GelDoc 1000 video densitometer. The linear relationship of optical densities and protein quantity was checked by calibration.

RESULTS
Sequence of Drosophila CALPB-The LD18261 clone contained an open reading frame encoding a 791-amino acid polypeptide with a predicted molecular mass of 90,134 daltons. The first ATG was assigned to be the initiation codon. Although we cannot yet exclude the possibility that the natural protein has further N-terminal amino acids, the identity of the start site was corroborated by the occurrence of a 5Ј in-frame stop codon in all of the overlapping EST clones. A polyadenylation signal was found within the 3Ј-untranslated region. The new predicted Drosophila calpain homolog, designated as CALPB, has a domain structure similar to that of vertebrate calpains and shows highest similarity to the CALPA protein (13). However, the N-terminal ends of the two proteins are unrelated, and the unique insertion sequence present in domain IV of CALPA is absent from CALPB (Fig. 2). The homologous region of the sequences shows 78% amino acid identity. CALPB carries the conserved Cys, His, and Asn active site residues, characteristic of papain-like cysteine proteases, and six putative Ca 2ϩ -binding EF-hand motifs (Fig. 2).
To study the relationship of CALPB to vertebrate calpains, a phylogenetic tree was constructed from some representative calpain homologs and Drosophila calpains (Fig. 3). The position of CALPA and CALPB in the tree suggests that separation of these isoforms occurred later than the divergence of the ances- tors of vertebrates and insects, therefore these two fruit fly paralogs have no direct counterpart in vertebrates.
Purification and Specific Activity of CALPA, CALPA-⌬Ins, and CALPB-The plasmid constructs allowed the production of recombinant CALPA, CALPA-⌬Ins, and CALPB fused at their C terminus with a 6-His affinity tag. While most of recombinant CALPA was found in insoluble aggregates, CALPA-⌬Ins and CALPB could be purified from the bacterial cytosol in sufficient quantities. CALPA was purified from inclusion bodies using a denaturation step in 8 M urea. When expressed at higher temperature in the presence of 1 mM isopropyl-1-thio-␤-D-galactopyranoside, CALPB also formed inclusion bodies which were treated similarly to be able to compare the effect of the different protocols. All three enzymes were purified on a metal-chelate affinity resin which yielded highly pure proteins (see below, Fig. 5). Only the eluted fractions containing CALPA-⌬Ins contained two contaminating proteins. Chromatography was carried out in either denaturing or nondenaturing conditions. Denatured proteins were refolded using different dialysis protocols (addition of PEG 8000, 0.1% Triton X-100, enzyme diluted to 50 g/ml, dialysis at room temperature), however, none of these yielded enzymes with high specific activity (data not shown). The yield and specific activity of the different enzymes are shown in Table I. CALPB prepared from the soluble fraction has the highest specific activity (similar to that of recombinant m-calpain). Renatured CALPB is less active, indicating the lack of perfect refolding. CALPA has the lowest specific activity probably due to imperfect refolding (similarly to renatured CALPB). However, the specific activity of CALPA-⌬Ins is also low suggesting that the low activity of CALPA is not only due to the denaturation-renaturation procedure but also is intrinsic to CALPA. The low specific activity of CALPA and CALPA-⌬Ins could be increased 4 -5-fold by the addition of certain phospholipids (see below).
Ca 2ϩ Dependence and the Effect of Other Divalent Cations-Both CALPA and CALPB attained maximal activity in the millimolar calcium range (Fig. 4). The calcium concentration required for half-maximal activity was 2.18 Ϯ 0.46 for CALPA, 3.23 Ϯ 0.23 mM for CALPA-⌬Ins, 4.38 Ϯ 0.63 mM for renatured CALPB, and 3.28 Ϯ 0.46 mM for soluble CALPB. Given that the soluble forms have millimolar Ca 2ϩ need as well, it is unlikely that only renaturation is responsible for the unusually high values. Ba 2ϩ , Mg 2ϩ , Cu 2ϩ , and Ni 2ϩ ions did not activate either CALPA or CALPB, however, CALPB, but not CALPA, was slightly activated by Mn 2ϩ ions (data not shown).
Autolysis of Calpains- Fig. 5 shows the autolytic conversion of CALPA, CALPA-⌬Ins, and CALPB in the presence of calcium. Autolysis of CALPA-⌬Ins and CALPB resulted in a sin-  gle, 81-kDa product, while multiple autolytic products of different lengths were generated from CALPA. CALPA-⌬Ins could not be fully autolyzed, because of its parallel degradation, which led to the enzyme's disappearance upon longer incubation times. The difference in the autolytic products of CALPA and CALPA-⌬Ins is very likely due to the differences in the preparation procedures. CALPA is prepared from inclusion bodies, and has lower specific activity, probably due to partially misfolded proteins in the preparation which fraction could be degraded in an aspecific manner, giving multiple bands on the SDS gel. Multiple autolytic fragments were also observed when renatured CALPB was analyzed (data not shown).
The 81-kDa autolytic fragment of CALPA-⌬Ins and CALPB was N-terminal sequenced after electrophoresis and blotting, giving the sequence NMRVL and NMFW for CALPA-⌬Ins and CALPB, respectively. Self-cleavage was thus shown to have occurred between Lys 54 -Asn 55 in CALPA-⌬Ins and between Gln 90 -Asn 91 in CALPB (Fig. 1). The cleavage sites are at identical positions, 88 (CALPA-⌬Ins) and 89 (CALPB) amino acids N-terminal from the active site Cys. Interestingly, the second cleavage site of mammalian -calpain before Leu 28 (giving the 76-kDa active form (22)) is 87 amino acids from Cys, whereas m-calpain is cleaved before Ser 20 , i.e. 85 amino acids from Cys (23). This conservation of the autolytic site suggests a common evolutionary origin and the ancient nature of the self-regulation of calpain by autolysis.
Vertebrateand m-calpains were shown to become more sensitive to calcium after autolysis (23, 24). This, however, was not the case for CALPB. Fully autolyzed CALPB showed halfmaximal activity in the millimolar range, at 4.54 Ϯ 0.56 mM free Ca 2ϩ . CALPA and CALPA-⌬Ins were not studied in this respect because of their imperfect autolysis.
Activation of Calpains-The activation of CALPA, CALPA-⌬Ins, and CALPB was studied in a fluorometric assay. The continuous monitoring of the cleavage of a fluorogenic substrate allowed detailed kinetic analyses. After the addition of calcium activity continuously increased until it reached a plateau of maximal activity. As seen in Fig. 6 there is a lag phase which corresponds to activation. The fluorometric progress curves were differentiated to better visualize the activation period (Fig. 7). The enzymes are thus inactive at the beginning of the reaction, and become progressively activated as the reaction proceeds.
The Relation of Autolysis and Activation in CALPB-Activation of CALPB ran parallel with its autolysis as shown in Fig.  8. This parallelism and the unusually long activation periods hint at the necessity of autoprocessing before conversion of the substrate is made possible. To test the hypothesis that autolysis is a prerequisite for, and not a consequence of, activation, CALPB was fully autolyzed, dialyzed against calpain puffer, and assayed in the fluorometric test (Fig. 9). In this case no activation period could be observed, even at low Ca 2ϩ concentrations, indicating that, in contrast to the intact protein, the autolyzed form becomes fully active right after the addition of calcium.
Similar measurements could not be carried out with CALPA and CALPA-⌬Ins, since these enzymes underwent only imperfect autolysis and were gradually degraded. However, based on the identical cleavage site in CALPA-⌬Ins and CALPB and the similar kinetics of activation, we propose that CALPA is also activated through N-terminal autolytic processing.
The Effect of Phospholipids-Vertebrate calpains have been shown to be activated by certain phospholipids. The occurrence of CALPA immunoreactivity in the 100,000 ϫ g pellet in cellfractionation experiments (data not shown) and the presence of a 16-amino acid long hydrophobic segment in the primary structure of CALPA pointed to its possible regulation by membrane lipids. These led us to measure the effect of various phospholipids on the activity and activation rate constant (first-order rate constant of activation) of our recombinant enzymes. The activation rate constant is the reciprocal of transit time if we assume that activation can be described by a firstorder reaction (25).
The activity and activation rate constant of CALPA and CALPA-⌬Ins were greatly enhanced by the addition of certain phospholipids (Fig. 10). PIP 2 , PIP, PI, and phosphatidic acid increased the activation rate constant and the maximal rate of the hydrolysis of the fluorescent substrate. Phosphatidylethanolamine and phosphatidylcholine had only minor effects. The only difference that could be observed between CALPA and   FIG. 5. Autolysis of CALPA (A), CALPA-⌬Ins (B), and CALPB (C). The enzymes were incubated in the presence of 10 mM Ca 2ϩ at 24°C for the times indicated above the lanes. Since autolysis and activation of CALPA and CALPA-⌬Ins was very slow in the absence of phospholipids, these proteins were autolyzed in the presence of 50 g/ml PIP 2 . The autolytic conversion of CALPA-⌬Ins and CALPB into one 81-kDa fragment contrasts the heterogeneous autolytic profile of CALPA. The two lower molecular weight bands in the CALPA-⌬Ins sample are not autolytic products but contaminating proteins present in the original preparation. No autolysis was observed in the absence of Ca 2ϩ (not shown).
FIG. 6. Fluorometric measurement of calpain activity. Progress curves of soluble CALPB in the presence of 1, 5, and 10 mM free Ca 2ϩ . The lag phase is clearly observable. The linear phase of the curves was extrapolated to zero substrate conversion (dashed lines) to determine the transit time of activation. Renatured CALPB was activated similarly, although its specific activity was lower (not shown).
CALPA-⌬Ins was the much greater increase of the activation rate constant of CALPA in the presence of PIP 2 and phosphatidic acid. In contrast to CALPA and CALPA-⌬Ins, soluble and renatured CALPB could barely be activated by these phospholipids. PIP 2 enhanced CALPB activity 2-fold and PA increased the rate of CALPB autolysis 4-fold, but the latter had no effect on the activity of the fully activated enzyme. To see whether phospholipids alter the Ca 2ϩ sensitivity of CALPA, the caseinolytic activity of the enzyme was determined in the presence of either PIP or phosphatidic acid and varying amounts of calcium. Neither of these lipids altered significantly the Ca 2ϩ concentration needed for half-maximal activity (2.23 Ϯ 0.28 and 2.08 Ϯ 0.03, respectively). DISCUSSION

CALPB, a New Drosophila Calpain-
The new homolog, reported here, is the second known protein in Drosophila which belongs to the canonical, Ca 2ϩ -regulated calpains. It has been expressed, along with CALPA and a mutant form of CALPA, in E. coli, and purified to homogeneity. Active CALPA could be produced using a denaturation step, similarly to recombinant human -calpain (26), while CALPB and CALPA-⌬Ins were isolated from the soluble fraction as reported for recombinant rat m-calpain (20,27).
Calpain Activation-The study of these enzymes can contribute to the elucidation of regulatory mechanisms in calpains which has been controversial for a long time (1, 28 -30). One of the widely accepted models claims that calpains become active through N-terminal autolysis (31, 32). The alternative view holds that intact calpain is the active form (33,34), and autolysis is only responsible for the dissociation of the enzyme from the plasma membrane (35,36), for fine-tuning substrate specificity (37) or subunit dissociation (30,38). The parallel nature of autolysis and activation in Drosophila CALPB and the lack of an activation phase in the fully autolyzed enzyme argue for the necessity of autolysis in Drosophila calpain activation in vitro. The almost identical position of the self-cleavage site in vertebrate and the fruit fly calpains reflects conserved structural features even in the sequentially unrelated N-terminal region. This conservation indicates functional constraints and therefore the characteristics of autolysis in Drosophila calpains point to similar regulation in the vertebrate enzymes. The apparent difference is that the Ca 2ϩ requirement of CALPB is not reduced upon autolysis. In fact, it may not be autolysis that alters Ca 2ϩ sensitivity in vertebrate calpains either; in the model proposed by Suzuki and co-workers it is dissociation of the heterodimer (30,38) that causes the increase in the enzyme's Ca 2ϩ sensitivity.
Ca 2ϩ Sensitivity-A considerable difficulty in understanding calpain activation in vivo stems from the unphysiologically high in vitro calcium requirement. The high Ca 2ϩ need of Drosophila enzymes cannot be attributed to imperfect renaturation because soluble CALPA-⌬Ins and CALPB also required Ca 2ϩ in the millimolar range. The in vivo Ca 2ϩ concentrations, however, are at least an order of magnitude lower than those in the present assays, which raises concerns about the physiological significance of the results. Vertebrate calpains become sensitized to calcium either upon phospholipid binding (39), through autolysis, dissociation, or upon binding to an activator protein (40 -42). Autolysis and phospholipids had minor effect on the Ca 2ϩ requirement of CALPB and CALPA, respectively, arguing for the existence of other regulatory factors, potentially similar to the vertebrate activator proteins.
Phospholipid Action-The effect of phospholipids on calpain activity and autolysis has been studied extensively. According to one view calpain migrates to the plasma membrane upon Ca 2ϩ signals (35,43) and becomes activated by membrane lipids (39) through a decrease in the Ca 2ϩ concentration needed for autolysis. The effect of phospholipids on the activity and activation rate constant of CALPA suggests a similar regulatory mechanism. In the case of CALPA certain phospholipids seem to cause an increase in enzyme activity. This leads to enhanced autolysis which in turn results in the faster activation observed. It can be shown that a change in the Ca 2ϩ sensitivity of autolysis (indirect kinetic parameter) can result from increased specific activity and activation rate constant (direct kinetic parameters) assuming similar activating effect of phospholipids at all Ca 2ϩ concentrations. Again it is not likely that renaturation has much altered the lipid dependence of CALPA since soluble CALPA-⌬Ins was affected similarly, moreover, the behavior of renatured and soluble CALPB was also indistinguishable.
The 16-amino acid long hydrophobic stretch in the CALPAspecific inserted region is the first obvious candidate to mediate lipid action. However, the removal of it had only slight effect on the lipid dependence of enzyme activity. This hydrophobic stretch is therefore not solely responsible for lipid activation, but may probably serve as an anchor bringing CALPA close to the membrane, and other lipid-binding sites should exist to mediate lipid action in CALPA-⌬Ins.
Despite the phylogenetic distance of Drosophila and vertebrate calpains, many features of calpain activation seem to be evolutionary conserved. The similarities in the activation, autolysis, and lipid dependence substantiate our efforts with CALPA and CALPB as useful models in the analysis of the calpain system, and urge us to exploit the genetic accessibility of the fruit fly.  Fig. 6). To generate the 81-kDa fragment, CALPB was incubated in the presence of 50 mM Ca 2ϩ for 1 min at room temperature. The reaction was stopped with EDTA and the autolyzed sample was dialyzed against calpain buffer overnight. The completion of autolysis was checked by SDS-polyacrylamide gel electrophoresis.  Fig. 6). Activity was determined from the plateau of the differentiated fluorometric progress curves (see Fig. 7). The fluorometric assay was set up in a final volume of 50 l, in the presence of 20 mM Ca 2ϩ and 50 g/ml lipid. Protein concentration was between 0.1 and 0.3 mg/ml. Activity was taken as 100% in the absence of lipids (Control).