Autolysis, Ca2+ requirement, and heterodimer stability in m-calpain.

The roles of N-terminal autolysis of the large (80 kDa) and small (28 kDa) subunits in activation of rat m-calpain, in lowering its Ca2+ requirement, and in reducing its stability have been investigated with heterodimeric recombinant calpains containing modified subunits. Both autolysis and [Ca2+]0.5 were influenced by the ionic strength of the buffers, which accounts for the wide variations in previous reports. Autolysis of the small subunit (from 28 to 20 kDa) was complete within 1 min but did not alter either the Ca2+ requirement ([Ca2+]0.5) or the stability of the enzyme. Autolysis of the NHis10-80k large subunit at Ala9-Lys10 is visible on gels, was complete within 1 min, and caused a drop in [Ca2+]0.5 from 364 to 187 μM. The lower value of [Ca2+]0.5 is therefore a property of the Δ9-80k large subunit. Autolysis at Ala9-Lys10 of the unmodified 80-kDa large subunit is not detectable on gels but was assayed by means of the fall in [Ca2+]0.5. This autolysis was complete in 3.5 min and was inhibited by high [NaCl]. The autolysis product of these calpains, which is essentially identical to that of natural m-calpain, was unstable in buffers of high ionic strength. Calpain in which the large subunit autolysis site had been mutated was fully active but did not undergo a drop in [Ca2+]0.5, showing that m-calpain is active prior to autolysis. The main physiological importance of autolysis of calpain is probably to generate an active but unstable enzyme, thus limiting the in vivo duration of calpain activity.

The calpains consist of an 80-kDa catalytic subunit (the large subunit), containing four domains, I-IV, and a 28-kDa regulatory subunit (the small subunit), containing two domains, V and VI (9,10). Domains I and V are involved in autolysis; domain II contains the most obvious active site residues (11); the function of domain III is not yet known, although it must also take part in the conformational changes induced by Ca 2ϩ ; and the C-terminal domains of both subunits, IV and VI, contain putative E-F hand motifs, some of which bind Ca 2ϩ (12)(13)(14)(15). No structural information is available for the complete calpain heterodimer, but the crystal structures of domain VI, with and without bound Ca 2ϩ , have been solved recently. 1 On exposure to sufficient Ca 2ϩ , the calpains are assumed to undergo a conformational change which permits the following four events: limited autolysis of the small subunit from 28 to 20 kDa; limited N-terminal autolysis of the large subunit; proteolysis of a substrate such as casein if it is present; and further inactivating proteolysis of the large subunit. However, the order of these events and the Ca 2ϩ concentrations required have been difficult to establish. The Ca 2ϩ requirements ([Ca 2ϩ ] 0.5 ) for casein hydrolysis by (initially) non-autolyzed calpains are usually reported to lie in the ranges of 5-50 M Ca 2ϩ for -calpain and 250 -1000 M Ca 2ϩ for m-calpain. These ranges are lowered by prior autolysis to 1-5 M Ca 2ϩ for -calpain and 100 -200 M Ca 2ϩ for m-calpain (16 -25). The reported values vary widely for at least two reasons, first because the casein assay reflects the net effects of activation, autolysis, and inactivation, and is not well suited for kinetic studies (26); and second because, as shown here, the observed values are highly dependent on experimental conditions. While it is fairly clear that autolysis lowers [Ca 2ϩ ] 0.5 of both and m-calpain, it has not been clear whether large and/or small subunit cleavage is required for, or responsible for, the fall in [Ca 2ϩ ] 0.5 and whether the non-autolyzed calpains are intrinsically active against substrates such as casein. In this work, we have answered some of these questions about autolysis and Ca 2ϩ requirement in m-calpain, using a variety of recombinant large and small subunits with N-and C-terminal modifications.
N-terminal Amino Acid Sequencing-All of the subunit modifications described were checked by DNA sequencing, and in some cases at the amino acid level. Amino acid sequences were obtained by automated Edman degradation, either from individual subunits following gel electrophoresis and blotting onto polyvinylidene difluoride membrane or from calpain solutions containing both subunits following concentration on a ProSpin cartridge (Applied Biosystems).
Calpain Large Subunit-cDNA for the rat m-calpain 80-kDa subunit and for this subunit with a C-terminal His-tag (80k-CHis 6 ), in the plasmid pET-24d(ϩ), has been described previously (7). Natural calpain large subunits are N-terminally blocked, presumably with an acetyl group, but in E. coli the subunits are expressed unblocked, with or without an initiating methionine residue, according to the nature of the second residue (28). The novel constructs described below are listed in Tables I and II. NHis 10 -80k and NHis 10 -80k-CHis 6 -To provide an N-terminal Histag, the NcoI site at the initiation codon of 80 kDa was converted to an NdeI site by site-directed mutagenesis. The 888-bp 2 MluI-NdeI fragment from pET-16b(ϩ) was then ligated into these sites in NdeI-80k. The resultant cDNA (NHis 10 -80k) codes for an m-calpain large subunit with an N-terminal leader peptide of 21 amino acids (MGH 10 SSGHIEGRH-). NHis 10 -80k-CHis 6 was cloned by ligating the 1030-bp MluI-BamHI 5Ј-fragment from NHis 10 -80k into the same sites in 80k-CHis 6 .
L8F, A9F-80k-CHis 6 -DNA coding for the N-terminal amino acid sequence of 80k-CHis 6 , MAGIAMKLAKDRE . . . , was converted to sequence coding for MAGIAMKFFKDRE . . . by means of site-directed mutagenesis, which introduced a DraI restriction site (tttaaa) into the cDNA at the new Phe 9 -Lys 10 position. The initial purpose of this construct was to abolish the expected autolysis site at Ala 9 -Lys 10 (23), but the construct was also used to make deletion mutants. The construct L8S, A9F-80k-CHis 6 was generated in the same way.
Large Subunit N-terminal Truncations-The construct L8F, A9F-80k-CHis 6 was digested with NcoI and DraI, removing a (28 ϩ 24nucleotide) fragment, followed by blunt ending with the Klenow fragment of E. coli DNA polymerase and religation. The product codes for a large subunit N-terminal sequence of MKDRE (M⌬9 -80k-CHis 6 ) in which the initiating methionine residue was retained on expression (as shown by amino acid sequencing). This truncation was designed to approximate the first autolysis product of m-calpain, ⌬9 -80k, which has the N-terminal sequence KDRE (23). Several other N-terminally truncated forms were made by digestion of L8F, A9F-80k-CHis 6 with NcoI and DraI, followed by insertion of pairs of complementary oligonucleotides, giving rise to the N-terminal amino acid sequences shown in Table II. A second NcoI site was introduced into 80k-CHis 6 at Gly 19 (converting the expressed amino acid sequence GLGSHERAIK to the sequence GLGSMAHERAIK) by site-directed mutagenesis. A 60-bp fragment was removed by NcoI digestion, and the plasmid was religated, generating a large subunit N-terminal amino acid sequence expressed as AHERAIK (A⌬20 -80k-CHis 6 ). This also was designed to approximate a second reported autolysis product, ⌬19 -80k, which has the N-terminal sequence SHERAIK (16,23). All of these constructs had C-terminal His-tags but some of them were cloned also without a C-terminal His-tag.
Rat 21-and 28-kDa Subunits, (28k and NHis 10 -28k)-Isolation of cDNA for the full-length rat calpain small subunit and its expression both as an unmodified 28-kDa subunit and as an NHis 10 -28-kDa subunit have been described elsewhere. 3 The C-terminal domain VI fragment of the rat calpain small subunit (Met 87 -Ser 270 ), previously described as the 21-kDa subunit, is now referred to as ⌬86.
Rabbit NHis 10 -28k-The rabbit calpain small subunit was expressed with and without an NHis 10 group, using a cDNA clone kindly provided by Dr. C. Crawford (University of Oxford).
Expression and Assay-Expression, purification, assay, electrophoresis, and immunoblotting of calpain were performed as described previously (30). 3 Titration of Ca 2ϩ Requirement-For measurement of the Ca 2ϩ dependence of calpain activity, a modification of the standard casein assay was used. The duplicate assays contained 2 mg of casein in a final volume of 0.5 ml of 0.2 M NaCl, 10 mM ␤-mercaptoethanol, 50 mM Tris-HCl, pH 7.6, net final CaCl 2 concentrations from zero M to 3.8 mM, and 20 l of enzyme sample (containing in most cases 5-7 units of activity). The mixtures were incubated at 25°C for 30 min; reaction was terminated by addition of 0.35 ml of ice-cold 10% trichloroacetic acid, and the A 280 nm values of the supernatants were recorded. For Ca 2ϩ titration, the enzymes were freshly re-purified by chromatography on MonoQ, in buffers containing 50 mM Tris-HCl and 2 mM EDTA. Precise comparisons were valid only between enzymes purified on the same day and titrated in the same buffers. The dissociation constant for Ca 2ϩ -EDTA at pH 7.6 was taken to be 1.22 ϫ 10 Ϫ8 M. 4 The Ca 2ϩ concentration required for half-maximal activity with casein as substrate is given as [Ca 2ϩ ] 0.5 . This value was calculated by fitting the normalized activity data to the equation y ϭ x n /(k n ϩ x n ), where y is the fraction of maximum activity, k ϭ [Ca 2ϩ ] 0.5 , n is the Hill constant, and x is [Ca 2ϩ ].
Conditions of Autolysis-Calpain samples (0.5-2 M, 50 -200 g/ml) in 2 mM EDTA, 10 mM ␤-mercaptoethanol, 0.05-0.5 M NaCl, 50 mM Tris-HCl, pH 7.6, at 20°C were adjusted with CaCl 2 to a net final concentration of 2.5 mM Ca 2ϩ , and shaken, normally for 30 -120 s, followed by addition of EDTA to a net final concentration of 2 mM free EDTA. In some cases, casein (5 mg/ml) was also present. Control samples received the same volumes of EDTA first and CaCl 2 second. Samples were taken for Ca 2ϩ titration, gel electrophoresis, and blotting, or re-chromatography on a MonoQ column. Autolysis conditions were chosen so that not more than 25% of the original activity was lost.
Stability of Autolysis Products-Stability of calpain is defined here in a special sense, in terms of recovery of activity from the MonoQ column. Control and autolyzed calpain samples were assayed to determine the loss of total activity due to autolysis and then chromatographed on the MonoQ column in the presence of 2 mM EDTA. Non-autolyzed mcalpains were eluted in high yield from this column at approximately 0.4 M NaCl, but for some constructs the activity remaining after brief autolysis was recovered only in poor yield from the column.

Effects of Subunit Modifications on Formation of Active Heterodimeric Calpains
Many combinations of modified m-calpain large and small subunits have been co-expressed. The yields of activity derived from 4 liters of E. coli, as observed after the first purification step, are given in Table I. The table also shows the specific activities of the purified calpains, which with one or two exceptions were all close to 1800 units/mg protein. The subunit compositions of the purified calpains were checked by immunoblotting and by Coomassie staining of SDS/Tris-Tricine gels, and specific activities of the purified products The initial yield is given as the total number of units of activity in 4 liters of culture, measured after DEAE-Sephacel chromatography, the first step of purification. The results of a single experiment are shown, but most expressions were repeated with consistent results. The specific activity is given for the enzyme after three purification steps. Construction of the various large subunits is described under "Experimental Procedures," and the small subunits ⌬86 and NHis 10 -28k have been described previously. 3 For comparison, an initial yield of 11,350 units of natural rat m-calpain was obtained from 400 g of rat carcass. to establish their heterodimeric character. The differing yields of activity in each co-expression experiment appear to reflect primarily the relative levels of expression of the two subunits, but the fairly constant specific activities of the purified products show that the subunit modifications in most cases did not affect heterodimer assembly or catalytic function. However, some of the active heterodimers had properties that were useful in the study of autolysis and [Ca 2ϩ ] 0.5 . Much of the work has been done with calpains having a C-terminal His-tag, since they were more highly expressed, but the results have been duplicated in several cases with calpains lacking the C-terminal His-tag. Table I also includes the fully active construct L8F, A9F-80k-CHis 6 /⌬86, in which the autolysis site at Ala 9 -Lys 10 has been abolished, for comparison with L8S, A9F-80k-CHis 6 / ⌬86, which was surprisingly inactive.

Steric Factors in Co-expression
As described elsewhere, 3 both the unmodified 28-kDa or NHis 10 -28k were less well expressed than ⌬86 but were able to form active enzymes with 80 kDa. Conversely, NHis 10 -80k formed active enzyme with ⌬86. However when both large and full-length small subunits contained N-terminal extensions (NHis 10 -80k/NHis 10 -28k or TRX-80k-CHis 6 /NHis 10 -28k), very little activity was formed and was not stable to further purification (Table I). Immunoblots of the crude extracts showed that the subunits were sufficiently expressed, and particularly the TRX-construct was highly expressed as predicted (not shown) (27). These results indicate that the two N-terminal extensions caused a steric interference to heterodimer formation and suggest therefore that the N termini of both subunits must be close to each other in normal m-calpain. It may be noted here also that the structure of domain VI 1 strongly suggests that the C termini of the large and small subunits are also close together. A very similar dependence of [Ca 2ϩ ] 0.5 on ionic strength was observed with several other recombinant calpains that were tested and also with partially purified natural rat m-calpain. Variation of pH between 7.2 and 7.8 (around the pH optimum for casein hydrolysis (30)) had little effect, but the value of [Ca 2ϩ ] 0.5 was greatly increased at pH 8.5 (not shown). The Ca 2ϩ concentrations of the solutions were checked by flame photometry, and we confirmed the observation (19) that dialysis of the casein substrate against EDTA had no effect on [Ca 2ϩ ] 0.5 values. All subsequent Ca 2ϩ titrations were therefore performed in 50 mM Tris-HCl, 0.2 M NaCl, pH 7.6.

Small Subunit Autolysis, [Ca 2ϩ ] 0.5 , and Heterodimer Stability
Expression of m-calpain with an NHis 10 -28k small subunit (in fact a mixture of partially degraded small subunits) and autolysis of this heterogeneous small subunit to a homogeneous 20-kDa small subunit have been described previously. 3 In 80k-CHis 6 /NHis 10 -28k, small subunit autolysis was complete within 1 min (as shown later, Fig. 4) and did not cause a fall in [Ca 2ϩ ] 0.5 (Fig. 2a). Since this autolysis was done in 0.4 M NaCl, as discussed later, large subunit autolysis can be neglected in this experiment; the result therefore demonstrates that small subunit autolysis alone does not contribute to the fall in [Ca 2ϩ ] 0.5 in this m-calpain.

Intermolecular Autolysis of the Small Subunit
To investigate the relative contributions of inter-and intramolecular reaction to autolysis of the small subunit, autolysis experiments were performed with mixtures of active and inactive calpains. In this experiment only, the full-length small subunits were from rabbit, and it should be noted that the monoclonal antibody (kindly provided by Dr. R. Mellgren, University of Toledo) used to observe the small subunit does not react with the rat small subunit. No other difference has been detected between calpains composed of the rat m-calpain large subunit and rabbit or rat calpain small subunits. 5 As shown in Fig. 3, in a mixture of equal molar amounts of the inactive mutant C105S-80k-CHis 6 /rabbit-NHis 10 -28k with active 80k-CHis 6 /⌬86, the heterogeneous small subunits of the inactive calpain were converted to a homogeneous 20-kDa form within 1 min, and this reaction could only be intermolecular (tracks 1-3). Autolysis of the small subunit in active 80k-CHis 6 /rabbit-NHis 10 -28k was also complete within 1 min (tracks 4 -6), but in this case it is not possible to distinguish between inter-and intramolecular reaction. In the control experiment, C105S-80k-CHis 6 /rabbit-NHis 10 -28k alone was not autolyzed (tracks 7-9). and (b) the heterogeneous small subunit in 80k-CHis 6 /NHis 10 -28k. In each case it is evident that autolysis was slightly retarded by the presence of 5 mg/ml casein (but was still essentially complete after 1 min), which is consistent with some contribution by intermolecular reaction to autolysis of both subunits. This effect of casein on autolysis of the large subunit was evident also in the Ca 2ϩ titrations described below.

Large Subunit Autolysis and [Ca 2ϩ ] 0.5
In the study of large subunit autolysis, conditions were chosen so that the calpains did not lose more than 25% of their starting activity, since it was felt that normalization of Ca 2ϩ titration data obtained after very extensive autolysis might be misleading. It was found that the effects of autolysis of the various calpains on their [Ca 2ϩ ] 0.5 for casein hydrolysis depended both on the experimental conditions during autolysis and also on the differences in the recombinant large subunits.
Effect of High [NaCl]-Autolysis of 80k/⌬86 in 0.4 M NaCl (as eluted from the Q-Sepharose column, the final step in purification) caused loss of activity, showing that the calpain was active, but did not change [Ca 2ϩ ] 0.5 . The titration curves are not shown since they were identical to those in Fig. 2. This result was unexpected but was highly reproducible. The activity remaining after autolysis of 80k-CHis 6 /⌬86 in 0.4 M NaCl was recovered in high yield from re-chromatography on MonoQ, and this material had clear N-terminal sequence showing that both subunits were intact. This result shows that m-calpain is active without large subunit autolysis, a point that was established even more rigorously with the L8F, A9F mutant (see below). Partially purified natural rat m-calpain also showed no change in [Ca 2ϩ ] 0.5 on autolysis in 0.4 M NaCl (data not shown).
When 80k/⌬86 was autolyzed in 0.18 M NaCl, the value of [Ca 2ϩ ] 0.5 showed the expected fall; in the presence of 5 mg/ml casein, about 270 s of autolysis were required to reach the lowest value of [Ca 2ϩ ] 0.5 , falling from 324 to 186 M (Fig. 5a). A comparison of autolysis of 80k/⌬86 for 120 s in 0.18 M NaCl with and without casein showed that casein slightly retarded the conversion to a low-Ca 2ϩ requiring form, from 338 M at time 0 to 292 M in the presence of casein and to 246 M in the absence of casein (Fig. 5b). N-terminal sequencing of 80k- Calpain samples were shaken gently at 20°C in the presence of 2 mM Ca 2ϩ , followed by addition of excess EDTA to prevent further reaction. Portions were run on several gels and analyzed by immunoblotting. Tracks 1-3, equimolar amounts of C105S-80k-CHis 6 /NHis 10 -28k and 80k-CHis 6 /⌬86; tracks 4 -6, 80k-CHis 6 /NHis 10 -28k; tracks 7-9, C105S-80k-CHis 6 /NHis 10 -28k; the three tracks in each case represent incubation times of 0, 0.5, and 1 min. The upper and lower portions of the blot were treated separately with appropriate antibodies as described. 3 In this experiment, the 28-kDa small subunits were from rabbit. The monoclonal antibody to the rabbit small subunit does not react with the rat small subunit. CHis 6 /⌬86 after incubation with Ca 2ϩ for 10 min gave a mixture of three sequences, representing approximately 33% intact large subunit (AGIAMKL), 66% autolyzed large subunit (KDREAA), and 100% intact small subunit. This cleavage at Ala 9 -Lys 10 is in agreement with a previous report (23).
Effect of the N-terminal His-tag-In contrast to the behavior of 80k/⌬86, the fall in [Ca 2ϩ ] 0.5 on autolysis of the N-terminally extended NHis 10 -80k/⌬86 at Ala 9 -Lys 10 was equally rapid in high or low [NaCl], or with or without casein (Fig. 6), and the small drop in molecular mass of the large subunit could be readily detected on gels (see above, Fig. 4). The large subunit of autolyzed NHis 10 -80k/⌬86 was N-terminally sequenced after gel electrophoresis and blotting, giving the sequence KDREAA which results from cleavage at Ala 9 -Lys 10 (strictly Ala 30 -Lys 31 in this construct). Over longer periods of autolysis, the remaining activity correlated with densitometry of the large subunit (Fig. 7), and since the large subunit after 30 -60 s is 100% cleaved at Ala 9 -Lys 10 , the activity after this time was therefore clearly a property of the autolyzed calpain. These results show that the lower [Ca 2ϩ ] 0.5 in m-calpain is a property of the ⌬9 -80k large subunit.
Effect of Abolishing the Autolysis Site-Calpain does not possess a strong consensus cleavage sequence to direct our choice of mutation, but there are very few examples of cleavage with a phenylalanine residue in the P 1 position (1). In accordance with this, the autolysis site mutant L8F, A9F-80k/⌬86 showed no significant change in [Ca 2ϩ ] 0.5 under any autolysis conditions (Fig. 8). This result demonstrates clearly, as also suggested earlier, that calpain with an intact large subunit is active in casein hydrolysis without autolysis.

Large Subunit Autolysis and Calpain Stability
After quenching of the autolysis reaction with excess EDTA, the remaining activity of the various calpains was stable for some hours at 0°C. The results in Table III suggest, however, that the autolysis product ⌬9 -80k/⌬86 is relatively unstable during purification on MonoQ, which includes elution at approximately 0.4 M NaCl. Incubation of 80k-CHis 6 /⌬86 with 2.5 mM Ca 2ϩ in 0.4 M NaCl causes loss of activity with time but does not involve cleavage at Ala 9 -Lys 10 , and the subsequent MonoQ recovery was high (84%); autolysis of several constructs in conditions where a relatively slow fall in [Ca 2ϩ ] 0.5 occurs and where the conversion to ⌬9 -80k/⌬86 was therefore incomplete was followed by intermediate recoveries from MonoQ (40 -60%); and autolysis of NHis 10 -80k-CHis 6 /⌬86, which was complete, gave much lower recoveries (20%). In the latter case, it was shown that the active material recovered in low yield from MonoQ was ⌬9 -80k-CHis 6 /⌬86 (Fig. 9).

N-Terminal Truncation of the Large Subunit in m-Calpain
The treatment of the initiating methionine residue in E. coli does not permit expression of the previously suggested large subunit autolysis products, ⌬9 -80k and ⌬19 -80k, with the desired free N-terminal amino acids (28). As a close approximation, the constructs M⌬9 -80k-CHis 6 /⌬86, ⌬8 -80k-CHis 6 / ⌬86, and A⌬20 -80k-CHis 6 /⌬86 were all expressed but were inactive (Table II), although the proteins could be isolated as heterodimers in average amounts (not shown). The excised N-terminal residues were then progressively reintroduced but did not restore activity until ⌬2-80k-CHis 6 /⌬86. The presence or absence of the C-terminal His-tag had no effect on these observations. At present we have no explanation for these results.

DISCUSSION
At the beginning of this work we found it difficult to reconcile the extensive but sometimes inconsistent literature on [Ca 2ϩ ] 0.5 and autolysis in m-calpain. It has now become clear that the observed values of [Ca 2ϩ ] 0.5 are highly dependent on experimental conditions. It should be noted also in the following discussion that autolysis of calpain involves not only the generation of the ⌬93-small subunit and the ⌬9-large subunit, in which form the enzyme is active and has a lower [Ca 2ϩ ] 0.5 , but also involves cleavage of the large subunit to smaller fragments that are no longer active (32,33). The relative rates of limited N-terminal autolysis and of inactivating autolysis are also dependent on the construct and on the ionic strength. Together these observations appear to explain much of the wide variation in [Ca 2ϩ ] 0.5 reported over the last 15 years from different laboratories.
Conditions for Measurement of [Ca 2ϩ ] 0.5 -With 80k-CHis 6 / ⌬86,in the absence of any prior autolysis, it was found that [Ca 2ϩ ] 0.5 was highly dependent on the concentration of Tris buffer and of NaCl (Fig. 1). This was true also for several different calpain constructs and for partially purified natural rat m-calpain. Within the range of pH 7.2-7.8, the effect of pH was not significant, but at higher pH values [Ca 2ϩ ] 0.5 increased significantly. These effects of ionic strength and pH on [Ca 2ϩ ] 0.5 for m-calpain have not, to our knowledge, been reported previously.

TABLE II N-terminal truncations of the large subunit
The N-terminal amino acid sequence of the natural rat m-calpain large subunit is assumed to be Ac-AGIAMKLAKDREAAEGLG-SHERAIK (29). The table shows the N-terminal amino acid sequences of various constructs as co-expressed in E. coli with ⌬86, together with the abbreviated name of the large subunit, and the yield of activity from 4 liters of E. coli. Sequences underlined were confirmed by amino acid sequencing. a These constructs were prepared with or without a C-terminal Histag, which affected the yield of expressed protein but did not affect whether activity was observed. had identical [Ca 2ϩ ] 0.5 (Fig. 2b). These results demonstrate unequivocally that autolytic removal of domain V does not affect [Ca 2ϩ ] 0.5 of m-calpain. Domain V of the small subunit is clearly not required for calpain activity or for formation of active calpain as a foreign protein in E. coli, although it presumably has some role in calpain regulation in eukaryotic cells. It may be noted also that removal of domain V does not by itself cause instability in m-calpain, since 80k(ϮCHis 6 )/⌬86 is stable through several steps of purification.
We have not studied the question of whether small subunit autolysis is strictly required, as seems likely, before large subunit autolysis or before casein hydrolysis. It would be interesting to attempt to generate mutant full-length small subunits that are resistant to autolysis to prove that removal of domain V is the essential first step. This may be difficult in practice, since domain V appears to be prone to hydrolysis at several sites. 3 Small subunit autolysis has been described both as intramolecular, on the basis of kinetic observations (19,34), and also as intermolecular (22,35). The experiments reported here with inactive calpain showed that intermolecular autolysis of the small subunit can occur at least as rapidly as any other autolysis step in this work (Fig. 3). In active m-calpain, small subunit autolysis was slightly retarded in the presence of casein, which also suggests some contribution by an intermolecular mechanism (Fig. 4). While the intermolecular reaction is clearly feasible in vitro, intramolecular reaction may be more likely in vivo, where the calpains are "relatively dilute and surrounded by potential substrates" (35).
Large Subunit Autolysis in m-Calpain-It has frequently been reported that incubation of natural m-calpain with Ca 2ϩ causes a fall in Ca 2ϩ requirement for hydrolysis of casein or other substrate, without any apparent change in the large subunit that can be detected on gels (2,23,24). As a result of this difficulty, the fall in [Ca 2ϩ ] 0.5 was at first attributed to small subunit autolysis alone (34) and later to cleavage of the large subunit either at Ala 19 -Ser 20 (16) or at Ala 9 -Lys 10 (23).
The construct NHis 10 -80k/⌬86 has the advantage that its autolysis can be easily seen on gels by Coomassie staining or by immunoblotting (Figs. 4,9), in a manner highly reminiscent of autolysis in -calpain (22,26,36). It should be noted that the pre-and post-autolysis values of [Ca 2ϩ ] 0.5 are the same for NHis 10 -80k/⌬86 as for 80k/⌬86 and that the autolysis product ⌬9 -80k/⌬86 is identical to the autolysis product of natural m-calpain, except for 7 N-terminal amino acids in the small subunit that do not affect its properties. 3 These results demonstrate that the lower [Ca 2ϩ ] 0.5 in m-calpain is a property of the ⌬9 -80k large subunit, so that the value of [Ca 2ϩ ] 0.5 can be used as a reliable index of large subunit autolysis at Ala 9 -Lys 10 .
Autolysis at Ala 9 -Lys 10 in NHis 10 -80k/⌬86 is essentially complete in about 50 s in any conditions (data not shown), although the autolysis shown in Fig. 6 was for 2 min to permit comparison with other experiments. This is significantly faster than autolysis of 80k/⌬86, as measured by the fall in [Ca 2ϩ ] 0.5 , which in the presence of 0.18 M NaCl and 5 mg/ml casein required approximately 270 s for completion (Fig. 5a). It has been reported elsewhere that large subunit autolysis in m-calpain was extremely slow, but the measurements depended on antipeptide antibodies that may not have been sufficiently diagnostic of autolysis (22).
The delay in autolysis in the presence of casein (Fig. 5b) suggested that large subunit autolysis in m-calpain could also be at least to some extent an intermolecular reaction, just as for the small subunit. In keeping with this idea, it has recently been shown that -calpain could autolyze m-calpain in vitro, suggesting a cascade mechanism for in vivo activation of m-calpain (37). In that report, m-calpain activation was stated to be due to small subunit hydrolysis, and a fall in [Ca 2ϩ ] 0.5 of m-calpain was observed from 160 to 64 M (measured in 10 mM HEPES, compare the value of 175 M in 50 mM Tris in Fig. 1). We have shown, however, that small subunit autolysis does not affect [Ca 2ϩ ] 0.5 , and in our view this fall in [Ca 2ϩ ] 0.5 was undoubtedly due to large subunit autolysis that could not be detected on gels.
Activity of Intact m-Calpain-It is evident that intact mcalpain must be active in the first steps of its own autolysis, and it has been suggested previously that intact m-calpain (at least with an intact large subunit) could hydrolyze substrates such as casein (17,20,24). This point has now been rigorously proven with the mutant L8F, A9F-80k/⌬86, which hydrolyzes casein with the same specific activity and [Ca 2ϩ ] 0.5 as the other calpains described; on exposure to Ca 2ϩ it loses activity (by inactivating cleavage of the large subunit) but does not undergo a fall in [Ca 2ϩ ] 0.5 and therefore does not undergo N-terminal autolysis (Fig. 8). The same conclusion could be drawn from the activity of 80k/⌬86 in 0.4 M NaCl, where N-terminal autolysis also does not occur.
Stability of Autolyzed m-Calpain-It was reported previously that autolyzed m-calpain was less stable to dialysis at high ionic strength than at low ionic strength (23). This is reflected in our finding that ⌬9 -80k/⌬86 formed by autolysis of various constructs was very poorly recovered from the MonoQ column, which requires elution at 0.4 M NaCl (Table III). We have made a related observation that the stability of the heterodimer containing an intact large subunit and an N-terminally truncated small subunit (80k/NHis-⌬116) was also reduced at high ionic strength. 3 Both these observations imply that salt links contribute to subunit binding. Autolyzed calpain is clearly not exposed to 0.4 M NaCl in vivo, but the loss of stability may still have in vivo implications that remain to be explored. Loss of ⌬9 -80k/⌬86 in the presence of EDTA suggests that it is prone to processes such as subunit dissociation (3) and aggregation. The constructs M⌬9 -80k/⌬86 and ⌬8 -80k/⌬86 were designed as models of the natural autolysis product and were expressed at average levels as heterodimers. They were, however, inactive and stable to purification and therefore differ in some way from the natural autolysis product. They may be in some way improperly folded, or the free N-terminal lysine residue at position 10 may be especially important, but this point requires further work.
Summary-These studies have clarified a number of aspects a Percentage of activity remaining in the incubation mixture after brief autolysis, with respect to the control sample incubated in the same conditions but without Ca 2ϩ .
b Percentage of activity recovered after application of the autolysis incubation mixture to the MonoQ column, with respect to the control sample also applied to the MonoQ column.
c This incubation was carried out in 0.5 M NaCl without casein for 1 min. of m-calpain biochemistry. For purified m-calpain in vitro, the pathways and rates of autolysis, and also the observed values of [Ca 2ϩ ] 0.5 , are highly dependent on the experimental conditions. Autolysis of the small subunit in m-calpain in the presence of Ca 2ϩ is rapid, is slightly retarded in the presence of casein, and does not affect either [Ca 2ϩ ] 0.5 or the stability of calpain. m-Calpain is fully active prior to autolysis of the large subunit, but in vitro at an approximately physiological ionic strength the intact large subunit has a short half-life in the presence of Ca 2ϩ , and it is assumed that the same is true in vivo. Autolysis of the large subunit at Ala 9 -Lys 10 may be a little slower than that of the small subunit and is also slightly retarded by the presence of casein, showing that at least part of the large subunit autolysis reaction is intermolecular. Although autolysis of both subunits can occur by intermolecular reaction, there are some grounds for believing that the reaction in vivo may be mainly intramolecular. Cleavage at Ala 9 -Lys 10 is responsible both for the fall in [Ca 2ϩ ] 0.5 and for the reduction in stability of the autolysis product. It is clear, however, that factors either in addition to, or other than, autolysis are involved in lowering [Ca 2ϩ ] 0.5 in vivo (1,2), and calpain autolysis products cannot be detected in normal tissue extracts and must, therefore, be very rapidly cleared after activation (20). In our view, the physiological importance of autolysis is probably that it generates forms of active calpain with short half-lives, resulting from their lower [Ca 2ϩ ] 0.5 , their instability, and their further auto-degradation, thus limiting the duration of calpain activity in vivo.