Limited autolysis reduces the Ca2+ requirement of a smooth muscle Ca2+-activated protease.

Chicken gizzard smooth muscle contains large amounts of Ca2+-activated protease activity. Approximately 15 mg of purified enzyme can be obtained from 1 kg of fresh muscle. The enzyme consists of two subunits (Mr = 80,000 and 30,000) present in a 1:1 molar ratio. In the presence of CaCl2, the 80,000/30,000-dalton heterodimer (form I) is rapidly converted by limited autolysis to a 76,000/18,000-dalton species (form II). Both the 80,000- and 30,000-dalton subunits are degraded simultaneously. Moreover, the Ca2+ dependence for autolysis (K0.5 = 300 microM) is identical for both subunits. Neither the time course nor the Ca2+ dependence of the autolytic conversion reaction is altered by 10- and 20-fold molar excesses of substrate. Limited autolysis markedly reduces the Ca2+ requirement for substrate degradation. Using N-[ethyl-2-3H]maleimide-labeled 27,000-dalton cardiac myosin light chains as substrate, the Ca2+ requirement of form I was found to be quite high (K0.5 = 150 microM). Under similar conditions, the Ca2+ requirement of form II was 30-fold lower (K0.5 = 5 microM). Limited autolysis did not alter the specific activity of the enzyme. Our results demonstrate that smooth muscle contains an abundant amount of Ca2+-activated protease. Moreover, autolysis of this enzyme may play an important regulatory role by converting the native form to a species that is fully active at physiological levels of intracellular calcium ion.

Ca2+-requiring proteases have been isolated from several tissue sources (1)(2)(3)(4)(5)(6)(7)(8). Most of these enzymes appear to have a neutral pH optimum and are potently inhibited by reagents that react with sulfhydryl groups (1)(2)(3)(4)(5)(6)(7)(8). Until recently, there was much skepticism about any important role for these enzymes in normal cell function because of the excessively high concentrations of ea2+ required for activation, generally in the millimolar range. However, several recent reports have described forms which are fully active at micromolar concentrations of Ca2+ (9-12). The relationship between the high and low Ca2+-requiring enzymes has not been established with certainty. At least one group has purified these two forms from skeletal muscle and shown that while both are heterodimers consisting of 80,000-and 30,000-dalton subunits, they differ only in ea2+ requirement and elution profile from DEAE-cellulose (12). Most recently, Imahori and colleagues have proposed that a ea"-activated neutral protease isolated from chicken skeletal muscle (M, = 80,000) can be converted by limited autolysis from a form that requires millimolar levels of ea2+ to one that is sensitive to ea2' in the micromolar range (10,11).
Recently, we have purified a Ca2+-activated protease from smooth muscle. Large amounts of this enzyme can be prepared from either chicken gizzard or bovine aortic smooth muscle. Like the protease originally purified from skeletal muscle by Dayton et al. (2), the native form of the smooth muscle enzyme is a heterodimer consisting of 80,000-and 30,000dalton subunits bound with a molar ratio of 1:l. In this study we report the purification of the ea2+-activated protease from chicken gizzard smooth muscle. In addition, we demonstrate that the native enzyme can be converted by limited autolysis from a high to a low Ca2+-requiring form.

Materials
All chemical reagents were purchased from Sigma, as were the chromatographic media, DEAE-Sephacel, and Reactive Red-I20 agarose. Sephacryl S300 was obtained from Pharmacia Fine Chemicals.

Methods
Electrophoresis-Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed in 7.5% slab gels according to the procedure of Porzio and Pearson (13) with modifications to improve gel stability (14). Nondenaturing gel electrophoresis was performed in 5% tube gels at 4 "C using a Fairbanks buffer system as described previously (15). Electrophoresis grade reagents were obtained from Bio-Rad. Tube gels stained with Coomassie blue were scanned with an ISCO model 1310 gel scanner and slab gels in a Beckman DU-8 spectrophotometer using the gel-scanning accessory and peak integrator. the horizontal mode using the LKB Multiphor as previously described Characterization Studies-Isoelectric focusing was performed in (16). Gels contained 2% Ampholines, pH 3-10 (Pharmacia), and 10% glycerol. Approximately 30 pg of enzyme was applied to the surface of the gels, and focusing was conducted at 4 "C with 10 watts, constant power, for 6 h. Sedimentation coefficients were determined by sedimentation of 50 pg of purified enzyme in 530% glycerol gradients as previously described (17). Markers included 100 pg each of catalase (11.3 S ) , alcohol dehydrogenase (7.4 S), and bovine serum albumin (4.3 S). Determinations of Stokes radius were obtained by gel filtration in a Sephacryl S200 column (0.9 X 60 cm) equilibrated with 20 mM MOPS,' pH 7.0, 1 m~ EGTA, 1 mM dithiothreitol, and 0.5 M NaCI. Markers included 1 mg each of catalase (5.2 n m ) , aldolase (4.2 m), bovine serum albumin (3.5 n m ) , ovalbumin (2.5 nm), and cytochrome c (1.4 nm). The sedimentation coefficient and Stokes radius were used to calculate molecular weight and frictional ratio according to the procedures described by Siege1 and Monty (18).
Protease Assays-Ca'+-activated protease activity was determined by using N-[ethyl-2-3H]maleimide-labeled 27,000-dalton bovine myosin light chains as substrate. The light chains were partially purified from cardiac myosin by a modification of the method of Pires and Perry (19). The 20,000-and 27,000-dalton light chains were separated by covalent chromatography onp-hydroxymercuribenzoate agarose (20). The 27,000-dalton light chains eluted from the column with 0.1 M cysteine were dialyzed free of thiol and labeled stoichiometrically with N-[etl1yl-2-~H]maleimide (3 mol of " S H per mol of light chain) in the presence of 6 M urea. After treatment with 5 mM dithiothreitol and dialysis to remove all reactants, approximately 10 @ light chains were added to a typical incubation mixture consisting of 50 m MOPS, pH 7.0, 2 m~ dithiothreitol, 2 rn EGTA, and various concentrations of CaC12 in a total volume of 50 pl. Assays were conducted under linear reaction conditions at 25 "C and were terminated at 30 s by the sequential addition of 10 pl of bovine serum albumin (10 mg/ml) in 50 m EGTA and then 100 pl of 20% trichloroacetic acid. Following sedimentation at 10, OOO X g for 15 mi n, a 100-p1 aliquot of the supernatant was counted in a Beckman Beta liquid scintillation counter. Activity is expressed as nanomoles of N-ethylmaleimide-peptide(s) released per min per mg of protease.
Ca2'/EGTA buffers were prepared in 20 m~ MOPS, pH 7.0, and 1 mM dithiothreitol with 2 m EGTA and varying amounts of CaC12. An association constant of 2.3 x IO6 M" was used for calculations of ionized calcium concentrations, and adjustments for pH and ionic strength were made according to procedures described by Harafugi and Ogawa (21).

RESULTS
Enzyme Purification-Ca2+-activated protease was typically prepared from 1 kg (trimmed weight) of fresh chicken gizzards. Gizzards were trimmed to remove mucosal and fibrous tissue and homogenized in a Waring blender in 3 volumes of 40 rrm MOPS, pH 7.2, 2 mM EGTA, 1 lll~ EDTA, and 1 m~ dithiothreitol. The homogenate was sedimented at 10,000 X g for 30 min and then subjected to ammonium sulfate fractionation. The precipitate obtained between 30 to 60% saturation was collected by sedimentation and resuspended in approximately 600 ml of 20 m~ MOPS, pH 7.0,l mM EDTA, 1 m EGTA, and 1 mM dithiothreitol (Buffer A). This material was dialyzed for 24 h against 15 liters of Buffer A and then applied to a column (5 X 18 cm) of DEAE-Sephacel equilibrated in the same buffer. A single peak of activity was resolved from this step (data not shown), and the active fractions were pooled (abcut 200 ml), concentrated to 100 ml (Amicon ultrafiltration cell, PM-10 filter), and applied to a column (5 X 100 cm) of Sephacryl S300 equilibrated in Buffer A plus 0.5 M NaCl. Gel filtration also resolved a single peak of protease activity (data not shown). The active fractions from this step were pooled (100 ml) and applied to a column (1.5 X 20 cm) of Reactive Red-120 agarose equilibrated with Buffer A. As shown in Fig. 1, the entire sample was loaded onto the column and then washed with approximately 150 ml of Buffer A plus 0.5 M NaCl.
protease activity was adsorbed to the column. Specific elution was achieved by washing the Reactive Red-120 agarose next with Buffer A alone. A sharp peak of protein with some tailing was eluted, and approximately 80% of the protease activity applied to the column could be recovered. A l l fractions containing protease activity were pooled (180 ml), concentrated (Amicon ultrafiltration cell, PM-10 fdter) to approximately 10 ml, and then dialyzed against Buffer A. This concentrated protease served as the source of enzyme for all subsequent studies described in this report. Table I summarizes the purification results. Approximately 15 mg of purified enzyme was obtained from this representative preparation. The protease was purified approximately 1429-fold. The true yield of activity could not be determined, since the initial extract had no detectable protease activity. This was due to the presence of an uncharacterized inhibitor(s). Similar difficulties in assaying for Ca2+-activated protease activity in crude muscle extracts have been described by other investigators (22).
The purified enzyme is shown in Fig. 2. Two components were present: an 80,000-dalton subunit and a 30,000-dalton subunit. Densitometric scanning of the gel shown in Fig. 2 gave a ratio of peak area of 2.7:l (80,000 to 30,000) yielding a molar ratio of approximately 1:l.
Effects ofAutolysis on Subunit Composition-The protease was found to undergo autolysis in the presence of Ca2+. In preliminary experiments with the enzyme it was found that the rate of autolysis and the kinds of fragments produced were strongly affected both by the Ca2+ concentration and the temperature at which experiments were conducted. It became apparent, however, that certain discreet intermediate autolytic products could be identified. As shown in Fig. 3, both the 80,000-and 30,000-dalton subunits of the protease were degraded. Limited proteolysis of the 80,000-dalton subunit yielded a 76,000-dalton polypeptide, while the 30,000-dalton subunit was converted to a peptide with a molecular weight of 18,000. To distinguish between the precursor and its autolytic product, we have referred to them as form I (i.e. the native enzyme consisting of 80,000-and 30,000-dalton subunits) and form I1 (i.e. the autolytic product consisting of 76,000-and 18,000-dalton subunits). Prolonged incubation of the protease in the presence of CaC12 resulted in the generation of other fragments and ultimately in loss of enzyme activity.  T o determine if the presence of substrate could alter either the rate or Ca" dependence of the conversion of form I to form 11, 0-, IO-, and 20-fold molar excesses of purified 27,000dalton cardiac myosin light chains were incubated with the protease, and autolysis was monitored as described in Figs. 4 and 5. As summarized in Table 11, neither the rate nor Ca" dependence of the conversion was altered by the presence of these concentrations of substrate. Since it was noted that degradation of both the 80,000-and 30,000-dalton subunits occurred very rapidly a t 25 "C, a time course was conducted on ice. Fig. 4 summarizes results of this study. Even at 0 "C, limited autolysis of the 80,000-and 30,000-dalton subunits was rapid at high concentrations of CaC12 (1 mM) and, moreover, degradation of both occurred simultaneously. Identical results were obtained over a protein concentration range of 0.05 to 2.5 mg/ml (approximately 0.5 to 23 PM enzyme). While the rate of autolysis was reduced a t lower concentrations of Ca2+ (i.e. 0.5 mM), as shown in Fig. 5 the 80,000-and 30,000-dalton subunits were degraded to the same extent a t corresponding concentrations of Ca". The concentration of Ca2+ necessary for half-maximal autolytic Purified enzyme (0.4 mg/ml) was incubated at 0 "C under conditions similar to those described in Fig. 3 except that the CaCI? concentration was 1 mM. Aliquots of the enzyme (5 pg) were obtained at the indicated time points, and autolysis was stopped by adding the aliquot directly to 10 pI of a sodium dodecyl sulfate-containing buffer (13). Following electrophoresis, gels were stained with Coomassie blue, and lanes were scanned for protein bands. The percentage of conversion (80,000 to 76,000 daltons and 30,000 to 18,000 daltons) was calculated from the relative peak ratios obtained by densitometric scans of the gels using the peak integrator of the Beckman DU-8 spectrophotometer. M , 80,000to 76,000dalton conversion; 0---0,30,000-to 18,000-dalton conversion. FIG. 5. Calcium dependence of autolysis. Buffer conditions were similar to those described in Fig. 3 except that several different calcium concentrations were used. The enzyme concentration was 0.4 mg/ml, and the temperature was 25 "C. Assays were terminated at 15 s in the manner described in Fig. 4. The percentage of conversion was determined by densitometric scanning as described in Fig. 4  both are simple heterodimers. The sedimentation coefficient of form I1 was found to be smaller than that of form I (Table   III), while the Stokes radii of both were identical. As indicated by the increase in frictional ratio of form I1 ( f / f ; , ) , autolysis enhanced asymmetric shape. The calculated molecular weights for both forms were in good agreement with molecular weights obtained from sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Table IV). Moreover, the calculated mo-

1'AHI.E 11
Effect of suhstrnte on the time course and Ca" dependence of ntrtoly.sis For both the time course and Ca"-dependence studies, the protease concentration was 1 p~. and autolysis was monitored and quantitated by gel electrophoretic methods as described in Figs. 4 and 5. l'he calcium concentration required for half-maximal autolytic conversion (K,,:, Ca") was determined at 0. 10. and 20 p~ substrate concentrations as described in Fig. 5. l'he time course of autolytic conversion at 0 "C in the presence of 0.5 mM CaCI:! was followed at 0, 10, and 20 p~ substrate concentration and the time required for 50'7 conversion (tl ?) determined from densitometric scans as described in Fig. 4

activated protease
Approximately 5 pg of purified enzyme obtained from each separation method was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The areas under each subunit determined by densitometric scanning were expressed as a peak ratio (80,000/ 30,000 for form I and 76,000/18,000 for form 11). Molar ratios were determined by correction of peak ratios for differences in subunit molecular weight.  FIG. 7 . Polyacrylamide gels of the two forms of Ca*+-activated protease. A, nondenaturing polyacrylamide tube gels stained with Coomassie blue as described in Fig. 6. a, form I; b, form 11. B, sodium dodecyl sulfate-polyacrylamide gel electrophoresis of protease obtained from the nondenaturing gels shown in A. a, form I; h, form

11.
lecular weights (form I, 108,OOO; form 11,92,400) were found to be consistent with the predicted subunit composition and suggested that the enzymes are simple heterodimers when studied in the presence of 1 m~ EGTA. These results do not exclude the possibility of oligimerization in the presence of calcium. The pH optimum was essentially identical for both forms. On the other hand, form I was found to be substantially more acidic (PI, 4.9) by isoelectric focusing when compared to form I1 (PI, 5.4).
Additional data confirming the subunit composition of the Ca2+-activated protease were obtained. When subjected to nondenaturing polyacrylamide gel electrophoresis as shown in Fig. 6, a single peak of protein corresponding to the protease activity could be resolved for both forms of the enzyme. of a Ca'+-activated Proteuse Despite its larger size, form I was found to migrate slightly faster than form 11, possibly as a result of its more acidic composition (Table 111). Fig. 7 shows the nondenaturing gels stained with Coomassie blue The Ca2' requirement for substrate degradation for both forms of the protease was determined as shown in Fig. 9. Form I showed a rather broad activation curve with half-maximal activity measured a t 150 PM calcium ion concentration. In contrast, form I1 was fully active at 10 PM calcium ion, and the estimated calcium ion concentration required for halfmaximal activation was 5 PM. When these data were analyzed in accordance with the Hill equation (23), a coefficient of 2.3 was obtained for form I, while a coefficient of 5.75 was determined for form I1 (Fig. 10). The Hill coefficient for form I is difficult to interpret in view of the likelihood of two events occumng simultaneously: autolysis and substrate degradation. On the other hand, the coefficient of 5.75 for form I1 suggests the possibility of multiple calcium ion-binding sites exhibiting positive cooperativity.

DISCUSSION
The role of Ca2+-activated proteases in cell function is uncertain. The protease purified from rabbit skeletal muscle may be important in dissolution of myofibrils by degrading proteins of the Z line (24)(25)(26)(27). Of particular concern is how the various enzymes purified to date can be active at physiological intracellular concentrations of Ca2+, since most seem to require millimolar levels of Ca2+ for full activity (1)(2)(3)(4)(5)(6)(7)(8).
Several recent studies have reported the presence of Ca2+activated protease activities which are active at micromolar levels of Ca2' (9)(10)(11)(12). In one study two forms with identical subunit composition were purified and found to require either high or low concentrations of Ca2+ for activity (12). Most recently, Imahori and colleagues, working with an apparently unique form of the Ca"-activated protease from chicken skeletal muscle, have reported that limited proteolysis converts a high Ca2'-requiring enzyme to a low calcium-requiring form (10,11). Their enzyme, which consists of a single 80,000dalton polypeptide that requires Ca2' in the millimolar range for activity, was converted to a 76,000-polypeptide that was active at micromolar levels of Ca2+ (10, 11).
Our results are similar to those of Imahori and colleagues.
Limited AutoZysZs of a Some important differences are notable. Firstly, the chicken gizzard smooth muscle Ca2+-activated protease, like that from cardiac muscle (7), skeletal muscle (2), and platelets (8) is a heterodimer. As detailed in this report, the 80,000-and 30,000dalton subunits maintain in a 1:l molar ratio through a variety of separation methods. In addition, the molecular weight of the holoenzyme calculated from the Stokes radius and sedimentation coefficient ( i e . 108,000) is consistent with the subunit composition proposed. Because the chicken gizzard protease is a heterodimer and both subunits were autolyticdy degraded simultaneously and at comparable Ca2+ concentrations, it is not clear whether limited proteolysis of only one of the subunits or both is prerequisite for the change in Ca2+ requirement for substrate degradation. A second major difference is the Ca2+ requirement for substrate degradation. In general, both forms I and I1 of the chicken gizzard enzyme require lower levels of Ca2+ for activity than do the high and low Ca2+-requiring forms of the chicken skeletal muscle protease described by Imahori and colleagues (10,11). For example, while the low Ca2+-requiring skeletal muscle enzyme is half-maximally active at a calcium ion concentraton of 40-50 PM, form I1 of the gizzard enzyme is fully active at this concentration and half-maximally active at 5 PM. This disparity may result from the intrinsic difference in subunit composition between the chicken skeletal muscle and smooth muscle enzymes. On the other hand, it may also reflect differences in assay methods or in calculation of calcium ion concentration.
Neither the calcium ion dependence nor the time course of autolysis was altered by the presence of 10-and 20-fold molar excess of substrate. Moreover, the same parameters were unaffected over a 50-fold range in protease concentration incubated in the absence of substrate. Recently, Mellgren et al. have shown that autolytic inactivation of the rabbit skeletal muscle Ca2'-activated protease is unaffected by a 20,000fold molar excess of substrate (28). Collectlvely, these results suggest that autolysis of the Ca2+-activated protease may result from an intramolecular process.
Form I1 of the smooth muscle Ca2+-activated protease Like the precursor, form I, is a heterodimer. The 76,000 and 18,000 subunits maintain a 1:l molar ratio through a variety of separation steps (Table IV) and can be cross-linked in the presence of dimethylsuberimidate. The low Ca2+ requirement of this form of the protease suggests that it could be the active species at physiological concentrations of intracellular Ca2+. At present, however, we have not identified form I1 occurring spontaneously in smooth muscle. In addition, while form I1 is quite sensitive to low concentrations of Ca2+, the conversion process &e. I + 11) requires relatively high levels of CaZ+ (& = 300 p~) .
Thus, if Ca2+-dependent autolysis is an important step which regulates the amount of micromolar Ca2+-sensitive protease present in smooth muscle, the autolytic process must occur in a localized region of the cell where transient calcium ion concentrations are high. On the other hand, our findings do not exclude the possibilities that other factors may alter the CaZ+ dependence of autolysis or that other proteases within the cell may catalyze the conversion.
It seems likely that the Ca2+-activated protease plays an important role in smooth muscle function. A large amount of this enzyme is present in both chicken gizzard and bovine aorta (29). The yield of protease from 1 kg of chicken gizzard 7u2+-actzuated Protease 9077 smooth muscle is 3 to 20 times greater than from comparable amounts of cardiac (7) and skeletal (2,4,5 ) muscles. This difference may be due to the variety of purification methods used to obtain the enzyme, differences in the amount or kinds of endogenous inhibitors present, or intrinsic content of the protease in different types of muscle. In any case, the precise role of the Ca2+-activated protease in smooth muscle function is not yet established. Most recently, we have found that when added to a contracted chemically skinned strip of smooth muscle, form I1 of the protease causes immediate and permanent loss of tension (30). This implies a potential role for the protease in smooth muscle contractile or cytoskeletal protein degradation.