ATP-dependent Protease in Bovine Adrenal Cortex TISSUE SPECIFICITY, SUBCELLULAR LOCALIZATION, AND PARTIAL CHARACTERIZATON*

Proteolytic activities in bovine adrenocortical mitochondria were investigated using [‘*C-methyZ]casein as a substrate. Washed mitochondria showed a low proteolytic activity at pH 7.5 or 8.2. ATP (5 mM) plus MgCl, (7.5 mM) stimulated the proteolysis 9 times at pH 8.2. It was further demonstrated unequivocally by various approaches that the ATP-dependent proteolytic activity localizes in mitochondrial matrix. The activity of the solubilized protease was sensitive to N-ethylmaleimide, mersalyl acid, phenylmethylsulfonyl fluoride, o-vanadate, rn-vanadate, vanadyl sulfate, and quercetin but not by oligomycin and ouabain. The ATP-dependent proteolytic activity was eluted at the position of 650,000 daltons on an Ultrogel AcA 22 column as a single symmetrical peak. The gel-filtered enzyme showed high specificity to ATP. GTP and UTP partially substituted ATP. ADP, AMP, tripolyphos- phate, a,&methylene ATP, and B,y-methylene ATP had little or no stimulating activity. ATP did not stimulate the activity in the absence of MgCl,. We measured ATP-dependent proteolytic activities in mitochondrial fractions from several tissues in rat and bovine. Adrenal cortex was one of the tissues of highest activity. In addition, we investigated the effect of adrenal atrophy on the ATP-dependent protease activity in rat adrenal. The

IesteroI to pregnenolone in adrenal cortex mitochondria (4)(5)(6)(7) . Although the reason for the lability of the ribosomal product is not understood at present, the involvement of mitochondrial proteolytic enzymes is a possibility, namely in view of the action of the protein factor on the mitochondria (6, 7 ) .
In addition, to the tropic effect, ACTH stimulates cell proliferation and differentiation (1,2). Accordingly, hypophysectomy decreases tissue weight, number of mitochondria, and many enzyme activities (8)(9)(10)(11). Proteases should participate on these atrophic processes of adrenal cortex after withdrawal of ACTH. Our interest is focused into the proteolytic digestion of mitochondrial proteins for two reasons. One is the fact that mitochondrial disintegration is regulated by  and the other is to understand the reason for the lability of so-called labile protein.
If adrenal cortex mitochondrial proteins are digested by an autophagic mechanism of lysosomes as seen by the presence of autophagic vacuoles containing mitochondria (ll), all mitochondrial proteins and enzymes should have the same halflives. However, Kimura (8), Purvis et al. (9), and Pfeiffer et al. (10) observed that in rat adrenal cortex mitochondrial P-450,, and P-45OIl8 had shorter half-lives than other enzymes. This fact suggests the involvement of mitochondrial protease in the proteolytic degradation of selected mitochondrial proteins and enzymes. At present, studies on adrenal cortex mitochondrial protease are completely lacking, although the lysosomal proteases have been investigated (15,16). In this study, we wish to describe the presence of a strong activity of an ATP-dependent protease in adrenal cortex mitochondria, together with its partial enzymic properties.

Proteolytic Activity in Mitochondrial Fraction from Bovine
Adrenal Cortex-We first examined the proteolytic activity of bovine adrenal cortex mitochondria by using casein as a substrate. The mitochondrial fraction was sonicated and incubated with 14C-labeled casein together with some reagents known to activate or inhibit proteases at two different pH values. As shown in Table 1 (Miniprint), the hydrolytic activities without any addition were low at pH 7.5 and 8.2. The activity was slightly inhibited by EDTA, an inhibitor of metaloproteases. Dithiothreitol and CaC12, which are cofac-Portions of this paper (including "Materials and Methods," part of "Results," Tables I and 11, and Figs. 2 and 3) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 84 "2365, cite the authors, and include a check or money order for $3.20 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press. 5511 tors for thiolproteases and Ca2+-dependent protease, respectively, did not show any activation. At pH 8.2, ATP-Mg activated the proteolysis about 9 times in the presence of oligomycin, however. The ATP-dependent activity was higher at pH 8.2 than at pH 7.4. These results strongly suggest the presence of ATP-dependent protease in bovine adrenal cortex mitochondria.
Standard Assay Conditions- Fig.  1 shows the proteolytic activity changes in the sonicated mitochondrial sample as a function of time. In the absence of ATP-Mg, the rate of proteolysis was very slow (curue A). When 2 mM ATP and 7.5 mM MgClz were added, activation in proteolysis was observed (curue C). The rate was linear for 30 min but it was drastically slowed down at the later phase. At a high concentration of ATP (8 mM) the reaction was slightly stimulated (curue D), compared with that at 2 mM ATP. In the presence of either 0.1 mM oligomycin (curue E ) or 0.25% Triton X-100 (curue F) with 2 mM ATP the rate of proteolysis proceeded linearly for 60 min. These reagents inhibited the liberation of inorganic phosphate from ATP (data not shown). When 100,000 x g supernatant solution of sonicated mitochondria was used as an enzyme sample, the reaction rate was linear at least for 120 min at 2 mM ATP, even in the absence of oligomycin and Triton X-100. The addition of membrane fraction (100,000 x g precipitate) greatly inhibited the reaction (data not shown). These results indicate the inhibition of the ATP-dependent protease activity by membrane-bound ATPase and other ATP-degrading enzymes such as alkaline phosphatase. The decrease in activity after prolonged incubation can be explained by decrease in ATP concentration due to membrane-bound ATP-consuming enzymes. On this basis, the addition of Triton X-100 or oligomycin into the reaction mixture is essential to measure the full activity of ATP-dependent protease in whole mitochondria. Subcellular Distribution of ATP-dependent Proteolytic Activity-We have investigated then the subcellular distribution of the ATP-dependent proteolytic activity in bovine adrenal cortex. The cortex was fractionated by differential centrifugation into five fractions, i.e. nuclear, heavy mitochondrial, light mitochondrial, microsomal, and soluble fractions. The ATPdependent proteolytic activity was measured in the presence of 0.25% Triton X-100 ( cific activity of the ATP-dependent proteolytic activity was highest in the heavy mitochondrial fraction and its profile was quite similar to those of the mitochondrial marker enzymes, citrate synthase and succinate dehydrogenase. On the other hand, the relative specific activities of both lysosomal marker enzymes, ,&galactosidase and cathepsin D, and a peroxisomal marker, catalase, were higher in the light mitochondrial fraction than in the heavy mitochondrial fraction. A considerable amount of catalase was found in soluble fraction in agreement with observation by Pazoles et al. (34). The profiles of glucose-6-phosphatase and 5'-nucleotidase, both of which were highest in the microsomal fraction, were quite different from that of the proteolytic activity. These results strongly suggest that the ATP-dependent proteolytic activity locates on mitochondria in bovine adrenal cortex.
We further performed sucrose density gradient centrifugation of the mitochondrial fraction and found that the modal density of ATP-dependent protease activity (1.150) was the same as those of mitochondrial markers but different from those of lysosomal (1.160) or peroxisomal (1.155) markers (data not shown).
Submitochondrial Localization of ATP-dependent Proteolytic Actiuity-We next examined submitochondrial distribution of ATP-dependent protease in bovine adrenal cortex. Mitochondria were disrupted with a French press in a hypertonic sucrose solution. After isolation of mitoplast and outer membrane fractions by differential centrifugation, the inner membranes were disrupted by Lubrol WX treatment, and we obtained five submitochondrial fractions. As seen in Fig. 3 (Miniprint), Fraction 1 had a high relative specific activity of monoamine oxidase and seemed to contain mainly mitochondrial outer membrane. Fraction 2 (soluble fraction after pressing) was rich in adenylate kinase activity, a marker of intermembrane space. Relative specific activity of monoamine oxidase was also high in this fraction. This might be due to incomplete sedimentation of outer membrane fragment even after centrifugation at 150,000 x g for 2.4 h as suggested by the previous investigators (18, 36). Fraction 3 is a mixed membrane fraction which contains both succinate dehydrogenase and monoamine oxidase with relatively low specific activities. Those of succinate dehydrogenase and citrate synthase were highest in Fractions 4 and 5, respectively. The profile of ATP-dependent proteolytic activity in these fractions was similar to that of citrate synthase activity, indicating this proteolytic system locates in the mitochondrial matrix. There is, however, a minor difference between two profiles; the relative specific activity of ATP-dependent protease in Fraction 4 (inner membrane fraction) was slightly higher than that of citrate synthase. This may suggest a weak interaction of the protease with inner membranes.
Tissue Distribution of ATP-dependent Proteolytic Actiuity-We measured ATP-dependent proteolytic activity in mitochondrial fractions from various tissues of bovine and rats.
In these experiments we used oligomycin to protect ATP from its degradation by ATPase. After incubation an aliquot of the trichloroacetic acid-soluble supernatant solution was subjected to analyses of inorganic phosphate for monitoring ATPdegrading activity as well as of radioactivity for protease assay. In all measurements, more than 1 mM ATP was left at the end of incubation, assuming that all inorganic phosphate was liberated from y-phosphate of ATP.
Rat adrenal, muscle, and lung had high ATP-dependent proteolytic activities (Fig. 4). In rat muscle and lung, ATPindependent activities were higher than the ATP-dependent ones. Testis, another steroidogenic tissue, had a much lower ATP-dependent activity than adrenal. ATP-dependent pro- tease activity in rat liver mitochondria was very low in our assay system, though we could detect significant activity in some preparations. In bovine adrenal cortex, the ATP-dependent proteolytic activity was much higher than that in adrenal medulla and liver. Stability of the Enzyme Activity and pH Dependency-We have partially characterized ATP-dependent proteolytic activity from bovine adrenocortical mitochondria. After extracting with 50 mM Tris-C1 buffer, pH 7.8, containing 10% glycerol and 10 mM mercaptoethanol, the enzyme activity was not stable. 20-30% and 90% of the activity were lost after 22 h at 4 "C at pH 7.8 and 6.0, respectively. Heat treatment at 50 "C for 2 min inactivated the activity to about 20% of the original activity at pH 7.8. Maximum activity was obtained at pH [8][9] and little activity was detected at pH 6. Imidazole and Tris buffers gave the same activity at the same pH.
Effects of Inhibitors-Effects of various inhibitors on the solubilized enzyme activity are shown in Table I1 (Miniprint). N-Ethylmaleimide and mersalyl acid, inhibitors of thiol proteinases, inhibited the ATP-dependent protease activity. PMSF, an inhibitor of serine proteases, also inhibited the activity, while TLCK and TPCK, which are inhibitors for trypsin and chymotrypsin, respectively, were not inhibitory. Pepstatin and o-phenanthroline, which inhibit carboxyl proteinases and metaloproteinases, respectively, had no effects. o-Vanadate, m-vanadate, vanadyl sulfate, and quercetin, inhibitors of various ATPases, inhibited the ATP-dependent protease in bovine adrenocortical mitochondria, while oligomycin and ouabain, which inhibit mitochondrial Fl-ATPase and Na,K-ATPase, respectively, had no effects.
Partial Purification of the ATP-dependent Protease-The extracted ATP-dependent protease was precipitated by ammonium sulfate at a concentration between 37 and 57% of saturation with a recovery of 82%. The proteins in the active fraction after ammonium sulfate precipitation were fractionated through a gel filtration column as described under "Materials and Methods" As shown in Fig. 5, a single symmetrical peak of ATP-dependent proteolytic activity was observed. The recovery of the activity was 67%. The specific activity of the ATP-dependent protease of the combined peak fractions was about 10 times higher than that of the original mitochondrial fraction. The molecular weight was about 650,000 as determined by the use of a calibrated column.
The ATP-dependent proteolytic system in reticulocytes is composed of several components including ubiquitin, which is present in all tissues including adrenal (37). To see whether the ATP-dependent proteolytic system in bovine adrenocortical mitochondria could be resolved into two components, we tried DEAE-cellulose chromatography under the conditions where ubiquitin is removed (38). By DEAE chromatography we obtained two fractions, Fraction I (nonabsorbed to DEAEcellulose) and Fraction I1 (0.5 M KC1 eluate). Fraction I contained no ATP-dependent proteolytic activity, while 60% of the applied enzyme activity was detected in Fraction 11. Addition of Fraction I to Fraction I1 had no further increase in the ATP-dependent proteolytic activity of Fraction 11. This result suggests that our system appears to be independent of ubiquitin.
Nucleotide Specificity-Finally, we examined the effects of various nucleotides and calcium ion using the partially purified enzyme preparation (Table 111).
The protease required both ATP and M%+ for its maximum activity. Ca2+ was about half as active as M%+. GTP and UTP could replace ATP only partially. Tripolyphosphate and a nonhydrolyzable analogue of ATP, @,y-methylene ATP, failed to stimulate the proteolytic activity. Another ATP analogue, a,B-methylene ATP stimulated the activity only slightly. This slight activity was not due to contamination of ATP in a,@methylene ATP preparation, since the activity was not affected by the addition of hexokinase and glucose. AMP had practically no stimulating activity while ADP had considerable activity. However, t.his effect of ADP seemed to be due to ATP formation from ADP by contaminating adenylate ki- . . . . . , absorbance at 280 nm. nase. Bovine adrenocortical mitochondria are rich in adenylate kinase in its intermembrane space (see Fig. 3) and the gel-fdtered protease preparation still contained significant adenylate kinase activity, though the molecular weight of the kinase was much smaller than that of the protease. The rate of ATP formation from ADP was about 8 pM/min under our protease assay conditions, which seemed to be sufficient to exert the observed proteolytic activity in the presence of ADP.

Adrenal ATP-dependent Protease
(Half-maximal stimulation was attained at 250 p~ ATP under present experimental conditions.) Furthermore, the addition of hexokinase and glucose (0.4 mM) inhibited the "ADPdependent" activity but not the activity in the presence of excess of ATP (2 mM). From these data we concluded that ADP per se had no stimulating activity. Effect of ACTH on ATP-dependent Proteolytic Activity in Adrenals from Dexamethasone-treated Rat-Some years ago, it was reported that half-life of mitochondria in rat adrenal increased 2-&fold by ACTH treatment and decreased by dexamethazone treatment (12)(13)(14). Therefore, we have examined whether this proteolytic system decays similarly to other mitochondrial enzymes during atrophic process.
One group of rats received dexamethasone and saline to induce atrophy and another group received dexamethasone and ACTH to maintain adrenal function for four days. Adrenal ATP-dependent proteolytic activity in each rat was determined using a cell homogenate. Tissue weight and protein content were reduced to 60% in dexamethasone/salinetreated rats, showing considerable atrophy by this treatment. All mitochondrial enzyme activities tested decayed more than tissue weight (about 38-54% of the control value). The decay in ATP-dependent proteolytic activity was similar to that of succinate dehydrogenase and cytochrome oxidase (45%). The decrease in cytochrome P-450 (38%) was more prominent than other mitochondrial enzymes. On the other hand, lysosomal enzyme activities decreased less than tissue weight (68-87%). Glucose-6-phosphate dehydrogenase, an enzyme in soluble fraction, also decreased to a similar extent to the mitochondrial enzymes (43%), while decrease in glucose-6-phosphatase activity, a microsomal enzyme, was the same as the decrease in tissue weight (57%).

DISCUSSION
In this study the presence of ATP-dependent protease was demonstrated in bovine adrenocortical mitochondria; ATP stimulated the proteolytic activity nearly 9 times in the presence of oligomycin and MgCl,. The ATP-dependent protease seems to be a major protease in bovine adrenocortical mitochondria using casein as a substrate. EDTA-sensitive protease and DTTor Ca'+-stimulated protease activities in adrenal mitochondrial samples were very low or not detected. In mitochondria from other tissues than adrenal ATP-independent proteolytic activity was high whereas the dependent activity low. The adrenocortical ATP-dependent protease activity localizes in the mitochondrial matrix as judged by the profiles of differential centrifugation, sucrose density gradient centrifugation, and submitochondrial fractionation.
Some intracellular protein degradative processes have been known to be energy-dependent in a variety of organisms (39) despite the exergonic nature of the proteolytic reaction. One explanation for this energy dependency is intralysosomal protein degradation. Lysosomal enzymes have acidic pH optimum, and intralysosomal pH has to be maintained at 4.5-5.0 by an ATP-driven proton pump (40,41) in order to degrade efficiently all substances incorporated into lysosomes through endocytosis or autophagy. Sequestration of cellular components into autophagic vacuoles is likely to be also energydependent (39). Clearly, our enzyme does not belong to the class of lysosomal proteases although it requires ATP for activity.
In recent years, two non-lysosomal ATP-dependent proteolytic systems were found in mammalian cells. One is a ubiquitin-dependent proteolytic system in reticulocytes (38,(42)(43)(44) and the other is a vanadate-sensitive, ATP-dependent and ubiquitin-independent protease in rat liver mitochondria (45). The former system is composed of several protein components including ubiquitin, while the latter behaved as a singl-protein entity as judged by gel filtration and ionexch. .Ige chromatographies and seemed to be similar to the bacterial ATP-dependent protease (46,47). The mitochondrial system, which was first discovered in liver by Desautels and Goldberg (45), is implicated in the involvement in the degradation of proteins synthesized by mitochondria. More recent studies showed that rabbit reticulocytes also contained ubiquitin-independent, vanadate-sensitive, and ATP-dependent proteolytic system (48). Furthermore, in spinach leaves (49) and pea chloroplasts (50,51) ATP-dependent proteolytic activities were observed.
ATP-dependent proteolytic activity in bovine adrenocortical mitochondria is different from the ubiquitin-dependent proteolytic system of reticulocytes in the following aspects: 1) adrenocortical ATP-dependent proteolytic activity is localized in mitochondria; 2) the reticulocyte system is sensitive to inhibition by TPCK, TLCK, ando-phenanthroline (421, while the adrenocortical protease is not; 3) the adrenocortical protease behaved as a single protein entity in gel filtration chromatography, whereas the ubiquitin system is composed of several components; and 4) the adrenocortical protease failed to be resolved into two components under the conditions where ubiquitin was separated from the other components. The adrenocortical ATP-dependent protease also differs from the other ATP-stimulated proteases such as cathepsin D (52), histone H1 degrading enzyme in lymphocyte (531, and pyrophosphate-stimulated protease in rat liver (54) all of which are stimulated not only by ATP but also by polyphosphate compounds, because the adrenocortical protease was not stimulated by tripolyphosphate and a nonhydrolyzable ATP analogue. The adrenocortical ATP-dependent protease resembles rat liver ATP-dependent protease (45) in its intracellular localization, molecular weight, sensitivity to inhibitors, and nucleotide specificity and seems to be a homologous enzyme. Some differences between adrenal and liver enzymes were observed, however. The rat liver enzyme was inhibited by Triton X-100 (45), while the adrenocortical enzyme was not affected by the detergent. This rendered the assay method for this enzyme in adrenal cortex convenient because the detergent not only solubilized the enzyme but also inhibited ATP-degrading activity. Other differences were that adrenocortical ATP-dependent protease was more sensitive to Nethylmaleimide and less sensitive to vanadate inhibition than the rat liver enzyme. Nevertheless, it is still premature to conclude that the adrenal enzyme differs from the liver one in terms of enzyme protein. Rather, we would emphasize here that the ATP-dependent protease activity in adrenal cortex mitochondria is approximately 30 times higher than that in liver mitochondria from rat.
Adrenal enzyme was specific to ATP. Only nucleoside triphosphates could partially substitute ATP, but ADP, AMP, and tripolyphosphate could not. Furthermore, nonhydrolyzable ATP analogue, @,y-methylene ATP failed to stimulate adrenal proteolytic activity. These results suggest hydrolysis of ATP is required for the proteolysis. Another ATP analogue, &methylene ATP, slightly stimulated the degradation of casein. This suggests that the hydrolysis of ATP presumably occurs between @-and y-phosphates.
The physiological significance of a high activity of ATPdependent protease is interesting with respect to the adrenocortical functions. Wheeldon et al. (55), Kalnov et at. (56), and Desautels and Goldberg (57) suggested that mitochondrially synthesized proteins are degraded inside the mitochondria by an ATP-dependent manner in yeast and rat liver, namely protein subunits without associating subunits which are synthesized in cytoplasmic ribosomes seem to be degraded by the ATP-dependent protease.
The ATP-dependent enzyme must play a role on the degradation of mitochondrial proteins. The reported values for half-lives of rat liver (58), heart (59), kidney (60), brain (60), and adrenal (12-14) mitochondria seem to have no apparent correlation with the ATP-dependent protease activity in the respective tissues (Fig. 4). After ACTH withdrawal, the apparent half-life of rat adrenal mitochondria is shortened (12), while the ATP-dependent protease activitylg of the tissue decreases. Lysosomal enzyme activities rather increase in terms of specific activities. This is perhaps parallel to the fact that the specific activities of lysosomal enzymes decreased in hypertrophic rat adrenal (35). From these facts, it is implicated that lysosomal enzymes rather than the mitochondrial one may be responsible for the autophagic degradation of mitochondria as an entity. However, some mitochondrial proteins have shorter half-lives than others (58, 61). In adrenal cortex, P-450,, has a shorter half-life (3.5 days) than some other mitochondrial enzymes (5-6 days) in hypophysectomized rat (8,9). The proteins with short half-lives might be degraded by ATP-dependent protease. Since the activity depends on ATP, the mitochondrial integrity must be maintained in order to generate ATP by oxidative phosphorylation. Therefore, the enzyme must work on the relatively intact mitochondria. The denatured cytochrome, which is generated during the catalytic processes in a suicide manner (62), may be digested by the ATP-dependent enzyme. In addition, proteins which are transferred to mitochondria after their synthesis in ribosomes may be a second type of the substrate. In this sense, so-called labile protein (t++ = about 8 min), which is a ribosomal product promoting the cholesterol side chain cleavage reaction in the mitochondrial inner membrane upon response to ACTH stimulation (3), may be an interesting candidate for the substrate. These possibilities are under investigation in this laboratory. We recently succeeded in purifying the ATP-dependent protease from bovine adrenocortical mitochondria to near homogeneity. The purification and characterization of the enzyme will be published separately.

53.
l l l 0 l