Inactivation and self-association of Ca2+/calmodulin-dependent protein kinase II during autophosphorylation.

The time-dependent loss in enzyme activity associated with the autophosphorylation of Ca/calmodulin-dependent protein kinase II (CaM-kinase) was altered by both pH and ATP concentration. These parameters also influenced the extent to which soluble CaM-kinase undergoes self-association to form large aggregates of sedimentable enzyme. Specifically, autophosphorylation of CaM-kinase in 0.01 mM ATP at pH 6.5 resulted in the formation of sedimentable enzyme and a 70% loss of enzyme activity. Under similar conditions at pH 7.5, the enzyme lost only 30% of its activity, and no sedimentable enzyme was detected. In contrast to 0.01 mM ATP, autophosphorylation of CaM-kinase at pH 6.5 in 1 mM ATP did not result in a loss of activity or the production of sedimentable enzyme, even though the stoichiometry of autophosphorylation was comparable. Electron microscopy studies of CaM-kinase autophosphorylated at pH 6.5 in 0.01 mM ATP revealed particles 100-300 nm in diameter that clustered into branched complexes. Inactivation and self-association of CaM-kinase were influenced by the conditions of autophosphorylation in vitro, suggesting that both the catalytic and physical properties of the enzyme may be sensitive to fluctuations in ATP concentration and pH in vivo.

The time-dependent loss in enzyme activity associated with the autophosphorylation of Ca 2؉ /calmodulin-dependent protein kinase II (CaM-kinase) was altered by both pH and ATP concentration. These parameters also influenced the extent to which soluble CaM-kinase undergoes self-association to form large aggregates of sedimentable enzyme. Specifically, autophosphorylation of CaM-kinase in 0.01 mM ATP at pH 6.5 resulted in the formation of sedimentable enzyme and a 70% loss of enzyme activity. Under similar conditions at pH 7.5, the enzyme lost only 30% of its activity, and no sedimentable enzyme was detected. In contrast to 0.01 mM ATP, autophosphorylation of CaM-kinase at pH 6.5 in 1 mM ATP did not result in a loss of activity or the production of sedimentable enzyme, even though the stoichiometry of autophosphorylation was comparable. Electron microscopy studies of CaM-kinase autophosphorylated at pH 6.5 in 0.01 mM ATP revealed particles 100 -300 nm in diameter that clustered into branched complexes. Inactivation and self-association of CaM-kinase were influenced by the conditions of autophosphorylation in vitro, suggesting that both the catalytic and physical properties of the enzyme may be sensitive to fluctuations in ATP concentration and pH in vivo.
We have demonstrated a loss of CaM-kinase activity in both in vivo and in situ models of ischemia (25,26). Coincident with inactivation in these models was a redistribution of soluble CaM-kinase to particulate fractions. One possible explanation for the loss of activity and redistribution of soluble CaM-kinase is that ischemic-induced alterations in the cellular environment alter the properties of the enzyme, potentially through an autophosphorylation-mediated process. Biochemical alterations observed during ischemia include ionic imbalances with marked reductions in intracellular pH and ATP concentrations (see "Discussion"). Although several investigators have observed a time-dependent decrease in the activity of autophosphorylated CaM-kinase (i.e. inactivation) in vitro (18 -24), the physiological significance and role of autophosphorylation in this loss of activity is unknown. Lou et al. (22) reported that ATP concentration influences autophosphorylation-associated losses in the activity of CaM-kinase, with low ATP maximizing and high ATP preventing enzyme inactivation. The generation of autonomous activity and the sites of autophosphorylation also were influenced by ATP concentration (22). Other parameters reported to influence the loss of enzyme activity and sites of autophosphorylation include reaction temperature (27) and the type and concentration of divalent ions (24,27). These findings indicate that the conditions of autophosphorylation are important determinants for alterations in the activity of CaM-kinase. The redistribution and loss of soluble CaM-kinase activity associated with ischemic conditions led us to further examine the influence of conditions of autophosphorylation, specifically pH and ATP concentration on the activity and physical characteristics of CaM-kinase in vitro. In this study, we observed that pH and ATP concentration influenced the inactivation of CaM-kinase associated with autophosphorylation. These parameters also influenced the self-association of soluble CaM-kinase into sedimentable enzyme complexes, suggesting that cells undergoing fluctuations in pH and ATP concentration could produce autophosphorylation-associated changes in both the physical and catalytic properties of the enzyme in vivo.

EXPERIMENTAL PROCEDURES
Materials-ATP was purchased from Pharmacia Biotech Inc., [␥-32 P]ATP (3,000 Ci/mmol) from Amersham Corp., and AMP-PCP from Boehringer Mannheim. Tabbed-copper grids (400 mesh) and 90-nm polystyrene sizing beads were obtained from Ted Pella. Ultrafiltration units (regenerated cellulose, M.W.C.O. 100,000) were purchased from Millipore. Low molecular weight standards and Tween-20 were obtained from Bio-Rad. CaM was purified from bovine brain as described by Gopalakrishna and Anderson (28). Soluble CaM-kinase was purified from rat forebrain essentially as described by Kelly and Schenolikar with the omission of gel filtration (29). The enzyme was quantified using either Bradford or BCA (Pierce) methods, with bovine serum albumin as the standard. Multiple forebrain preparations of CaMkinase generated similar results throughout this study. Specific activity of enzyme preparations under the assay conditions described below were between 12 and 30 mol/min/mg. The peptide substrate syntide (PLRRTLSVAA) was synthesized on an automated peptide synthesizer (Applied Biosystems) and high performance liquid chromatography purified.
Autophosphorylation and Processing of CaM-kinase-Standard autophosphorylation conditions are defined as 50 mM MES/Hepes, pH 6.5, 0.4 mM DTT, 0.5 mM CaCl 2 , 10 mM MgCl 2 , 1 M CaM, 10 M ATP, 3 Ci [␥-32 P]ATP, 10% (v/v) glycerol and 0.1% (v/v) Tween-20. The reactions (100 l, final volume) were initiated with the addition of enzyme (1 g). The enzyme was autophosphorylated (or simply incubated depending upon the reaction conditions) for 0.5 min on ice before an initial time 0 point was collected. The purpose of the preincubation step was to allow the equilibration of the reaction components and to generate a similar phosphorylation state on ice before allowing the enzyme to undergo further autophosphorylation at 30°C. In the absence of this step, similar alterations in sedimentation and Ca 2ϩ /CaM-dependent inactivation were observed. Following preautophosphorylation, the timer was started, and the reaction mixture was placed in a 30°C shaking water bath for collection of samples at subsequent time points. Aliquots (10 l, 100 ng of enzyme) were collected over time and diluted 6-fold in 50 l of ice-cold 50 mM Hepes, pH 7.2 with 10 mM EDTA, terminating further autophosphorylation of the enzyme. Samples were then subjected to centrifugation for 30 min at 15,000 ϫ g at 4°C. The supernatant and pellet were separated and resuspended in SDS-loading buffer (10% glycerol, 15 mM DTT, 2.3% SDS, and 62.5 mM Tris, pH 6.8) for SDSpolyacrylamide electrophoresis (PAGE).
Electrophoresis was performed as described elsewhere (30), using a 10% resolving gel. Following SDS-PAGE, the enzyme was visualized using either Coomassie Blue staining or Western blotting. Coomassie staining of proteins was accomplished by incubation of gels in fixative (20% methanol, 7% acetic acid) containing 0.002% of Coomassie Brilliant Blue R-250 for 30 min and destained overnight in fixative. For Western blotting, the enzyme was transferred to nitrocellulose using a Genie transfer apparatus (Idea Scientific). Transfer was performed at 4°C in 20% methanol, 190 mM glycine, and 26 mM Tris, pH 8.0, for 1-2 h at a constant 15-20 V. After transfer, the nitrocellulose was blocked overnight in 5% dry milk in phosphate-buffered saline. Monoclonal antibodies to the 50-(2D5) (31) and 60-(Cb␤-1) (32) kDa subunits of CaM-kinase were diluted 1:2000 in TBST (10 mM Tris, pH 8.0, 150 mM NaCl, and 0.05% Nonidet P-40), and the membrane was incubated at room temperature for 1 h on a platform shaker. The membrane was washed in TBST, followed by incubation in anti-mouse-IgG AP-conjugated secondary antibody (Promega) diluted 1:5000 in TBST for 1 h. The membrane again was washed in TBST and the immunoreactive proteins were visualized using the alkaline phosphatase substrates, nitro blue tetrazolium and 5-bromo-1-chloro-3-indolyl phosphate as described by Promega. Autoradiographs were produced by exposing Kodak XAR-5 film at Ϫ80°C for 1-2 h in the presence of an intensifying screen. The antibody to the 60-kDa subunit (Cb␤-1) was kindly provided by Dr. Howard Schulman (Stanford University).
Enzyme Activity-To measure changes in kinase activity during autophosphorylation, a second stage assay was utilized. An aliquot of autophosphorylated enzyme (a 2.5-l aliquot, 25 ng of the enzyme) was added to a second stage reaction mix (final volume, 50 l) consisting of 50 mM MES/Hepes, pH 7.2, 0.4 mM DTT, 0.5 mM CaCl 2 , 5 mM MgCl 2 , 1 M CaM, 50 or 100 M ATP, 3 Ci of [␥-32 P]ATP, 50 M syntide, 10% glycerol, and 0.1% Tween-20. During autophosphorylation conditions where ATP concentration was varied, the final ATP concentration of the second stage mix was adjusted to 100 M. The second stage mix was preincubated for 1 min at 30°C before addition of the enzyme. Following addition of the enzyme, the reaction was incubated for 0.5 min before 20 l of the reaction mix was spotted onto a Whatman P-81 filter paper. The filters were processed as described previously (33), and the bound radioactivity was quantified by Cerenkov counting. Although the reaction mixture supported a linear incorporation of 32 PO 4 into 50 M syntide for at least 3 min, a 0.5-min reaction time was chosen to minimize further autophosphorylation-dependent changes in the properties of the enzyme during the second stage reaction. Enzyme activity throughout the study is expressed as percent initial activity and reflects normalization to the time 0 measurement. Measurement of Ca 2ϩ /CaMindependent activity was performed under similar reaction conditions as described above, with the following modifications in the second stage reactions: 1) Ca 2ϩ /CaM was omitted, and 5 mM EGTA was included, and 2) a 15-s incubation at 30°C was performed before addition of the substrates (Mg 2ϩ /ATP and syntide) to the reaction mixture.
Stoichiometry of Autophosphorylation-Autophosphorylation was performed essentially as described for standard reaction conditions, with modifications as indicated in the figure legends. Aliquots of the reaction mixture (10 -20 l) were collected at the indicated times and diluted with SDS-PAGE sample buffer. SDS-PAGE, Coomassie staining, and destaining were performed as described above. An aliquot of the reaction mixture was taken before addition of the enzyme and spotted directly onto the filter paper. The counts/min from each reaction mix (filter paper) and the radioactivity incorporated into the subunits of CaM-kinase (dried gel) were determined simultaneously using a Beta-Scope 603 (Betagen, counting efficiency of 19%). Densiometric analysis of Coomassie Blue stained 50-and 60-kDa subunits after SDS-PAGE was used to determine a ratio of 3:1 (50:60 kDa) giving an average molecular mass per mol of CaM-kinase of 52.5 kDa. Stoichiometry is reported as moles of phosphate incorporated per mol of CaM-kinase.
Transmission Electron Microscopy (TEM)-Autophosphorylation was performed under standard conditions with the indicated changes in ATP concentration. Reactions were terminated by the addition of EDTA (final concentration of 50 mM) to each reaction mix and held on ice for 10 min. The samples were deposited onto carbon-coated Formvar grids in a Millipore 100,000 MW sample concentrator using a refrigerated microcentrifuge at 4,000 ϫ g for 3 min. The grids were incubated in 2% uranyl acetate in 25% methanol for 1-2 min and then washed in distilled water and air dried. Additionally, 90-nm polystyrene sizing beads were applied to the grid by the drop method for a size reference. Transmission electron microscopy was performed using conventional irradiation procedures on a JOEL-100CX operating at 80 kV.

RESULTS
Autophosphorylation of CaM-kinase was associated with a time-dependent decrease in Ca 2ϩ /CaM-dependent enzyme activity at pH 7.5 (Fig. 1A). Activity measurements from enzyme autophosphorylated at pH 7.0 and 6.5 indicated that a greater percentage of enzyme activity was lost over time with a decrease in reaction pH (Fig. 1A). Autophosphorylation of CaMkinase at pH 7.5 resulted in a 15% decrease in enzyme activity at 2 min, whereas pH 7.0 and pH 6.5 were both associated with a 40% loss of enzyme activity at 2 min. After 5 min, the extent of enzyme inactivation at pH 7.5 (30%) was less than both pH 7.0 (50%) and pH 6.5 (70%). Activity measurements also were performed in the presence of EGTA to monitor changes in Ca 2ϩ /CaM-independent activity (Fig. 1B). The time course of Ca 2ϩ /CaM-independent inactivation at pH 6.5 (65 and 40% at 2 and 5 min, respectively) and pH 7.0 (80 and 58% at 2 and 5 min, respectively) were similar to decreases in Ca 2ϩ /CaM-dependent activity over time. However, at pH 7.5 an increase in independent activity was observed at 2 min (120%) and a modest decrease in activity observed at 5 min (90%). These data suggested that maximal Ca 2ϩ /CaM-independent activity had not been obtained during the 0.5-min preautophosphorylation at pH 7.5. This conclusion was supported by measuring the percent Ca 2ϩ /CaM-independent activity generated at time 0 for pH 6.5, 7.0, and 7.5, which were 55 Ϯ 2.1, 54.2 Ϯ 5.4, and 42 Ϯ 0.2, respectively (n ϭ 3, ϮS.D.) following the standard 0.5-min preautophosphorylation protocol. Extending the preautophosphorylation to 1, 2, or 3 min on ice increased the percent Ca 2ϩ /CaM-independent activity generated at pH 7.5 at time 0 to 52.1 Ϯ 0.09, 55.3 Ϯ 0.06, and 54.6 Ϯ 0.04, respectively (n ϭ 3, ϮS.D.). After extending the preautophosphorylation on ice to 2 min, we observed no differences in the time course of autonomous inactivation for pH 6.5 and 7.0 conditions (data not shown). However, at pH 7.5, the percent inactivation observed for Ca 2ϩ /CaM-independent activity (90 and 60% at 2 and 5 min, respectively) was similar to the Ca 2ϩ /CaM-dependent loss of activity over time (see solid bar in Fig. 1B). These data indicated that the loss of CaM-kinase activity observed during autophosphorylation was influenced by pH and that the extent of inactivation observed is similar for both Ca 2ϩ /CaMdependent and independent activities.
At 0, 2, and 5 min of autophosphorylation, aliquots of the reaction mixture were subjected to centrifugation for 30 min at 15,000 ϫ g to determine the influence of pH on the sedimentability of autophosphorylated CaM-kinase. The Western blot in Fig. 2 depicts the distribution of CaM-kinase in the supernatant and pellet following autophosphorylation at pH 6.5, 7.0, and 7.5. As expected, the 50-and 60-kDa subunits were detected in the supernatant at time 0 in all pH conditions. Interestingly, after 2 min of autophosphorylation at pH 6.5, the enzyme was completely in the pellet. After 2 min of autophosphorylation at pH 7.0, the enzyme was detected in both the supernatant and pellet fractions, while the enzyme autophosphorylated at pH 7.5 remained in the supernatant. Although enzyme activity continued to decrease (see Fig. 1), no further changes in the distribution of CaM-kinase between supernatant and pellet were observed at 5 min. The electrophoretic mobility of the 50-kDa subunit was altered during autophosphorylation and decreased to a 54-kDa form in a time-dependent manner in all three pH conditions (see Fig. 2). Anomalies in the electrophoretic mobility of CaM-kinase subunits have been correlated with a high stoichiometry of autophosphorylation (3,18) and may result from differential SDS binding to the autophosphorylated subunits (22). Although the immunostaining of the 60-kDa subunit at the 2-and 5-min time points often was below the sensitivity of the immunoassay, autophosphorylation-associated changes in sedimentability and electrophoretic mobility were also observed for the ␤ subunit. The loss of immunostaining is not due to a loss of the 60-kDa subunit (see autoradiogram in Fig. 3) or specific to any reaction pH and may represent changes in the affinity of the monoclonal antibody to the autophosphorylated subunit and/or a decrease in the threshold for immunostaining. It should be noted that an increase in the preautophosphorylation period from the standard 0.5 min to 2 min did not alter the formation of sedimentable enzyme in the different pH conditions, nor did omitting the preautophosphorylation step completely (data not shown).
A more comprehensive time course was constructed to observe changes in sedimentability and activity of CaM-kinase during autophosphorylation at pH 6.5. The Western blot in Fig.  3 demonstrates that, under standard reaction conditions, CaMkinase was detected in the pellet as early as 1 min. Similar to Fig. 2, changes in the electrophoretic mobilities of the subunits of CaM-kinase occurred with autophosphorylation. A 54-kDa form of the 50-kDa subunit was observed as early as 1 min, appearing temporally with the detection of sedimentable enzyme. By 5 min, the 50-kDa form was predominantly shifted to the 54-kDa form. The mobility of the 60-kDa subunit was also altered during autophosphorylation and although not visible by immunostaining in Fig. 3, the formation of a 62-kDa form was evident after 5 min of autophosphorylation on the autoradiogram (Fig. 3). The autoradiogram of the Western blot in Fig. 3 also demonstrates alterations in the distribution of CaM-kinase between supernatant and pellet and altered mobilities of the 50-and 60-kDa subunits of the enzyme. The signal from the 32 P-labeled phosphoproteins clearly illustrates both the presence of the 60-kDa subunit throughout the period of autophosphorylation and the generation of the 62-kDa form after 5 min (Fig. 3).
The time course of CaM-kinase inactivation during autophosphorylation as described in Fig. 3, is shown in Fig. 4 (left ordinate axis). At 1.5 min, the enzyme was completely sedimentable (see Fig. 3), yet retained over 60% of its initial activity. After 7 min, the enzyme retained less than 25% of the initial activity. Quantitative analysis of 32 PO 4 incorporation into CaM-kinase during autophosphorylation indicated an increase in the incorporation of 32 PO 4 over time, with approximately 2.3 mol of phosphate incorporated per mol of enzyme after 7 min of autophosphorylation (Fig. 4, right ordinate axis). The time course of enzyme inactivation parallels an increase in the stoichiometry of autophosphorylation, suggesting that autophosphorylation contributes to the time-dependent loss in enzyme activity.
The autophosphorylation dependence of the formation of sedimentable CaM-kinase at pH 6.5 was investigated by sub- FIG. 2. Western blot analysis demonstrating that pH influences the formation of sedimentable CaM-kinase during autophosphorylation. Enzyme autophosphorylated at pH 6.5, 7.0, or 7.5 as described in Fig. 1 was subjected to sedimentation analysis. Aliquots of the reaction mix were collected at the indicated times and diluted 6-fold in 10 mM EDTA buffered with 10 mM Pipes, pH 7.2. Samples were centrifuged at 15,000 ϫ g for 30 min and supernatant (S) and pellet (P) fractions isolated for each time point. A complete description of enzyme processing for centrifugation and Western blotting is provided under "Experimental Procedures." Western blotting was performed using monoclonal antibodies to the 50-and 60-kDa subunits of CaM-kinase. The relative position of 50-and 60-kDa subunits of CaM-kinase are indicated by arrows.
stituting a nonhydrolyzable analogue for ATP (AMP-PCP) in the reaction mixture (Fig. 5A). In the absence of autophosphorylation, sedimentable enzyme was not formed. Similar results were obtained when the reaction mix contained 0.01 mM ADP substituted for ATP (data not shown). A decrease in the activity of enzyme incubated in AMP-PCP was observed over time in parallel second-stage reactions (Fig. 5B). The enzyme retained approximately 60% of its initial activity after 5 min in the presence of AMP-PCP. We observed that enzyme incubated in similar reaction conditions, yet in the absence of Ca 2ϩ /CaM retained 90% of its activity after 5 min (data not shown). This loss of substrate phosphorylation by CaM-kinase in the presence of Ca 2ϩ /CaM is similar to previous reports and may reflect a "thermal instability" of activated enzyme in the absence of ATP (21,23).
The role of autophosphorylation in the inactivation and formation of sedimentable enzyme was further investigated by altering the conditions of autophosphorylation. Autophosphorylation on ice previously was demonstrated to produce the autonomous form of CaM-kinase without enzyme inactivation (21,22). Lou et al. (22) suggested that this condition not only slows the kinase reaction, but also modifies the sites autophosphorylated compared to similar conditions at 30°C.  To explore the effects of temperature during autophosphorylation, limited autophosphorylation was accomplished at 30°C by substituting the divalent Mn 2ϩ for Mg 2ϩ (27). The Western blot in Fig. 5A demonstrates that autophosphorylation in the presence of 10 mM Mn 2ϩ does not result in the formation of sedimentable enzyme. In addition, no change was observed in the relative mobility of the 50-kDa subunit. The 60-kDa subunit clearly shifted to a 62-kDa form after 2 min and a 64-kDa form after 5 min of autophosphorylation. The stoichiometry of autophosphorylation was approximately 0.3 Ϯ 0.05, 0.8 Ϯ 0.13, and 1.2 Ϯ 0.18 mol of phosphate per mol of CaMkinase at 0, 2, and 5 min, respectively (n ϭ 5, ϮS.D.). Restricted or limited autophosphorylation in Mn 2ϩ , was supported by a maximal stoichiometry of Ϸ1, whereas in 10 mM Mg 2ϩ at 30°C, the maximal stoichiometry was Ϸ2 mol of phosphate per mol of CaM-kinase (see Fig. 4). Similar to autophosphorylation at 4°C, CaM-kinase autophosphorylated in Mn 2ϩ retained 90% of its initial activity after 5 min (Fig. 5B). During these experiments, we observed that the specific activity of CaM-kinase measured in the second-stage reaction was influenced by the concentration of Mn 2ϩ carried over from the autophosphorylation reaction (0.4 mM Mn 2ϩ final second-stage concentration). Although we did not extensively pursue the influence of Mn 2ϩ on the catalytic activity of CaM-kinase, the specific activity observed in the second-stage reaction was significantly improved if the concentration of Mn 2ϩ was decreased from 10 to 1 mM during the initial autophosphorylation reaction, lowering the final concentration in the second stage reaction to 0.04 mM. Autophosphorylation in 1 mM Mn 2ϩ produced similar results to the data reported above for 10 mM Mn 2ϩ , including a similar stoichiometry and no appreciable changes in the activity or sedimentability of the enzyme over time (data not shown). As seen in Fig. 5B, the inactivation associated with autophosphorylation in standard conditions (10 mM Mg 2ϩ , 10 M ATP, pH 6.5, at 30°C) was greater than inactivation in the absence of autophosphorylation, suggesting that additional processes may lead to further enzyme inactivation. Both inactivation and the formation of sedimentable enzyme are influenced by autophosphorylation. Restricted autophosphorylation does not produce sedimentable enzyme nor significant inactivation and may even prevent or protect the enzyme from "thermal inactivation." Lou et al. (22) previously demonstrated that autophosphorylation of CaM-kinase in high (0.5 mM) versus low (0.005 mM) ATP produced a similar stoichiometry of autophosphorylation associated with different sites of autophosphorylation. The functional consequence of autophosphorylation was characteristic of the ATP concentration (i.e. inactivation with little autonomous activity was associated with low ATP, and autonomous activity with little inactivation was associated with high ATP). Fig. 6 demonstrates the effects of ATP concentration (0.01, 0.1, and 1 mM) on the sedimentability of CaM-kinase autophosphorylated at pH 6.5. After 2 min of autophosphorylation in 0.01 mM ATP, the enzyme was completely in the pellet. However, CaM-kinase autophosphorylated at 0.1 mM ATP remained partially in the supernatant after 2 min, and enzyme autophosphorylated for 2 min in 1 mM ATP remained completely in the supernatant. Fig. 6 also illustrates that ATP concentration influenced the formation of the 54-kDa form of the 50-kDa subunit. In 1 mM ATP, the formation of the 54-kDa form was not observed. However, the formation of the 54-kDa form was enhanced with decreased ATP concentration. Autophosphorylation in 1 mM ATP produced changes in the mobility of the 60-kDa subunit without altering the mobility of the 50-kDa subunit. This is reminiscent of mobility changes observed when the enzyme is autophosphorylated in 0.01 mM ATP in the presence of Mn 2ϩ (see Fig. 5A). After 2 min of autophosphorylation in 0.01, 0.1, or 1 mM ATP, the percent activity remaining was 53 Ϯ 7.2, 83 Ϯ 6.4, and 103 Ϯ 13.8, respectively (n ϭ 3, ϮS.D.). The percent Ca 2ϩ /calmodulin-independent-activity remaining after 2 min of autophosphorylation in 0.01 and 1 mM ATP paralleled changes in the Ca 2ϩ / calmodulin-dependent activity; 65 Ϯ 6.4 and 122 Ϯ 2.5, respectively (n ϭ 3, ϮS.D.). The generation of independent activity in the 0.01 and 1.0 mM ATP conditions was similar, 53.3 Ϯ 0.6 and 53.3 Ϯ 3.6 (n ϭ 3, ϮS.D.), respectively. Autophosphorylation for 2 min in 0.01 mM and 1 mM ATP produced a stoichiometry of 1.2 Ϯ 0.23 and 1.1 Ϯ 0.32 mol of 32 PO 4 per mol of CaM-kinase, respectively (n ϭ 3, ϮS.D.). These data indicate that both inactivation and the formation of sedimentable enzyme are not a consequence of the total 32 PO 4 incorporated into CaM-kinase during autophosphorylation. In addition, these data demonstrate that inactivation of the enzyme and the formation of sedimentable CaM-kinase at pH 6.5 are prevented at high concentrations of ATP, while low ATP concentrations appear to promote this process.
The formation of sedimentable CaM-kinase suggested dramatic alterations in the physical characteristics of the soluble enzyme. Enzyme autophosphorylated in conditions that either promote (0.01 mM ATP at pH 6.5) or prevent (1 mM ATP at pH 6.5) the formation of sedimentable enzyme was visualized using TEM. CaM-kinase autophosphorylated in 0.01 mM ATP for 5 min produced distinct uranyl acetate-staining structures that were uniformly distributed over the grid (Fig. 7A, ϫ 5000). At a higher magnification (Fig. 7B, ϫ 33,000), particles ranging from 100 nm to 300 nm in diameter were evident and appeared to interconnect and associate to form branched structures, termed enzyme complexes. Under these conditions (0.01 mM ATP at pH 6.5) enzyme complexes were not observed after pre-autophosphorylation on ice (time 0), yet were evident after 2 min of autophosphorylation at 30°C (data not shown). In contrast, after 5 min of autophosphorylation in 1 mM ATP at pH 6.5, the grid was void of uranyl acetate-staining structures, and increased magnification revealed only the sizing beads (data not shown). The TEM conditions utilized in these experiments did not generate sufficient contrast to expose the 20 -30-nm structure of individual CaM-kinase holoenzymes (34). In addition, we directly analyzed the sedimented form of CaM-kinase following centrifugation (15,000 ϫ g for 30 min) by TEM and as expected, the enzyme complexes observed following centrifugation and resuspension were more tightly packed than enzyme complexes deposited directly to the grid (data not shown).
Throughout this study we have used immunostaining to identify the soluble and sedimentable enzyme. Fig. 7C demonstrates the Coomassie Blue staining pattern of enzyme autophosphorylated in standard conditions for 0 or 5 min. The majority of the non-CaM-kinase-staining protein remains in the supernatant after 5 min of autophosphorylation. The 50and 60-kDa subunits of CaM-kinase were the major constituents of the sedimentable protein following autophosphorylation. These data indicate that the formation of enzyme complexes is due primarily to interactions between soluble holenzymes, presumably through self-association; however, a minor constituent of the sedimentable protein was detected at approximately 75 kDa, and it remains possible that proteins that co-purify with CaM-kinase could influence the formation and/or associate with sedimentable enzyme. We feel that this is an unlikely possibility because the formation of sedimentable enzyme during autophosphorylation also was observed using a recombinant preparation of the alpha subunit of CaM-kinase that is greater than 95% pure by Coomassie staining of SDS-PAGE analyzed protein. 2

DISCUSSION
In this study we observed that the conditions of autophosphorylation influence both the formation of sedimentable CaMkinase and a time-dependent loss of substrate phosphorylation. Autophosphorylation at 4°C or in Mn 2ϩ at 30°C did not result in a significant loss of enzyme activity (less than 10%) nor in the formation of sedimentable enzyme, indicating that limited autophosphorylation either prevents or is not sufficient for these changes. Lou and Schulman demonstrated that these conditions restrict the sites of autophosphorylation and maintain the autonomous form of the enzyme without its inactivation (27). Furthermore, the substitution of Mn 2ϩ for Mg 2ϩ was reported to increase the affinity of the kinase for ATP (27). The K m for Mg 2ϩ /ATP is reported to be within the range of 1.6 -20 M for both substrate phosphorylation and autophosphorylation (4,15,18). The low ATP concentration used in this study, 10 M, is within the range of the calculated K m for ATP and, therefore, may not be saturating. The potential consequence or role of an unoccupied ATP binding site on the inactivation and formation of sedimentable CaM-kinase is currently under investigation.
We observed that ATP concentration influenced both enzyme inactivation and the formation of sedimentable enzyme. The inactivation associated with autophosphorylation in low ATP (10 M) was further influenced by pH, with maximal inactivation and formation of sedimentable enzyme occurring at pH 6.5, whereas autophosphorylation in high ATP (1 mM) prevented these changes. The stoichiometry of autophosphorylation was comparable between low and high ATP conditions, indicating that formation of sedimentable enzyme and/or inactivation were not due to the extent of 32 PO 4 incorporation. Lou et al. (22) also reported a similar ATP sensitivity to the inactivation of CaM-kinase associated with autophosphorylation and similar stoichiometries of autophosphorylation in high (500 m) and low (5 m) ATP. These authors also reported that tryptic mapping of CaM-kinase autophosphorylated in low ATP produced unique phosphopeptides compared to high ATP (22). In addition, Hanson et al. (9) reported that Thr-253 on CaMkinase was autophosphorylated preferentially at low ATP. Site specific autophosphorylation-associated alterations in the properties of CaM-kinase are complex and well documented (e.g. Thr-286 phosphorylation and autonomous activity); however, it is currently unknown whether the differential autophosphorylation associated with low ATP induces the inactivation and/or formation of sedimentable enzyme or is merely a consequence of these processes.
Inactivation and the formation of sedimentable enzyme during autophosphorylation in low ATP were influenced by pH. Autophosphorylation at pH 6.5 in low ATP produced robust inactivation and formation of sedimentable enzyme. Autophosphorylation at pH 7.5 was the only condition observed that produced enzyme inactivation, yet was not accompanied by the formation of sedimentable enzyme, indicating that inactivation FIG. 7. Electron micrographs of sedimentable CaM-kinase and SDS-PAGE analysis of sedimented enzyme. A and B, CaM-kinase autophosphorylated in standard reaction conditions was processed for TEM as described under "Experimental Procedures." A, a field of uranyl acetate-staining structures at ϫ 5000 following autophosphorylation in 0.01 mMATP for 5 min at pH 6.5. B, the same field as in A but at a higher magnification (ϫ 33,000). Arrows indicate polystyrene sizing beads 90 nm in diameter. Grid fields at high and low magnification are representative of the entire grid. C, analysis of sedimentable enzyme by SDS-PAGE and Coomassie Blue staining was performed on enzyme autophosphorylated in standard reaction conditions, with the exception of a decrease in reaction volume (50 l) and an increase in the amount of enzyme (2.0 g). Complete reactions were terminated at 0 and 5 min with EDTA (100 mM final). Centrifugation and processing of autophosphorylated enzyme was performed as described in Fig. 2 through SDS-PAGE. The gel was processed as described under "Experimental Procedures." Arrows indicate the 50-and 60-kDa subunits of CaM-kinase. Supernatant (S) and pellet fractions (P) fractions are indicated. Positions of molecular mass standards are indicated on the right. and the formation of sedimentable enzyme may occur by distinct processes. Enzyme inactivation could proceed at elevated pH, yet potential protein-protein associations necessary for the formation of sedimentable enzyme may be unstable or may not form. A potential limitation of this interpretation is that smaller complexes may still form during autophosphorylation at pH 7.5 but are simply not detected by the sedimentation criteria used in this study. Interestingly, we also observed that increased pH slowed the generation of maximal independentactivity in low ATP. Maximal independent activity at higher pH values (pH 7.5) required increasing the time period of preautophosphorylation on ice.
Lai et al. (21) attributed the loss of enzyme activity associated with CaM-kinase autophosphorylation to increased thermal lability. However, consistent with previous data (22,27), we observed that both autophosphorylation in low ATP with Mn 2ϩ and in high ATP with Mg 2ϩ prevented the inactivation of CaM-kinase associated with autophosphorylation at 30°C, indicating that autophosphorylated enzyme is differentially susceptible to thermal inactivation. The time course of inactivation following autophosphorylation in low ATP at pH 6.5 was correlated temporally with the incorporation of 32 PO 4 into the enzyme (see Fig. 4). Interpreting the role of autophosphorylation in the inactivation of CaM-kinase is complicated by a loss of enzyme activity in the presence of Ca 2ϩ /CaM and in the absence of autophosphorylation (see Fig. 5A). This process is temperature-sensitive and also is attributed to thermal lability (21,23). Ishida et al. (35) recently reported that this mode of inactivation may be produced via Ca 2ϩ /CaM interaction with a low affinity CaM-binding domain. The relationship of this autophosphorylation-independent inactivation to the inactivation observed during autophosphorylation is unclear. Determining the role of autophosphorylation in enzyme inactivation is further complicated by potential alterations in the structure of CaM-kinase associated with the formation of sedimentable enzyme. Stearic constraints due to enzyme aggregation may also contribute to the loss of substrate phosphorylation observed.
Interestingly, the enzyme complexes formed during autophosphorylation of purified CaM-kinase in our study (Fig. 7B) appeared similar in ultrastructure to preparations of isolated cytoskeletal complexes enriched in CaM-kinase previously described (36). The formation of the enzyme complexes in our study is apparently due to the self-association of soluble CaMkinase holoenzymes. The results of our study were consistent with multiple CaM-kinase preparations, with enzyme purities ranging from 70 -90% of the total protein, and although potential interactions and associations with proteins that co-purify with CaM-kinase cannot be excluded completely, they do not significantly co-sediment with sedimentable enzyme (Fig. 7C). Whether CaM-kinase self-association has any role in the formation of cytoskeletal specializations such as postsynaptic densities or the cytoskeletal complexes described by Sahyoun et al. (36) is unknown.
The enzyme complexes shown in the micrographs (Fig. 7, A  and B) are representative of fields throughout the entire grid and were observed only in autophosphorylation conditions that produced sedimentable enzyme. However, what percentage of the sedimentable enzyme has adopted this macromolecular structure is difficult to ascertain. As suggested by the reviewer, we attempted to quantify these structures by modeling the volume for both holoenzymes and the enzyme complexes and determining the theoretical number of holoenzymes required to construct the enzyme complexes observed in Fig. 7. Based on the assumptions and analysis provided under "Appendix," we estimated that 7% of the holoenzymes in the reaction were recovered as enzyme complexes. Of the many assumptions required to complete the analysis, the 7% estimation was highly dependent upon the assumed volume of the holoenzyme. This includes the values chosen for determining the volume of holoenzyme as well as potential alterations in the holoenzyme's volume associated with formation of enzyme complexes (i.e. packing and compression due to aggregation). For example, if the packing of the holoenzyme causes a decrease in their apparent diameter from 30 to 20 nm, the enzyme recovered increased to 15%, whereas an additional decrease in the diameter to 10 nm produces a recovery of over 60%. One interpretation from our calculations is that the formation of the enzyme complexes could be associated with significant packing and condensation of the holoenzymes, and structural alterations of the enzyme over time could represent an important contribution to the time-dependent losses in the enzyme's activity.
Although the role of autophosphorylation in the formation of sedimentable enzyme is unknown, this phenomena was not observed in the absence of autophosphorylation in the reaction conditions of this study. However, in the absence of 10% glycerol, we observed that 20 -30% of the enzyme sedimented when incubated at 30°C in the presence of Ca 2ϩ /CaM at pH 6.5 for 5 or 10 min (data not shown). The formation of sedimentable enzyme in the absence of autophosphorylation was not as rapid nor as complete as the formation of sedimentable enzyme observed with autophosphorylation, and was not detected in the standard assay conditions containing 10% glycerol. Glycerol and ethylene glycol are used routinely to stabilize purified CaM-kinase and one possibility suggested from these data is that these agents decrease enzyme aggregation.
CaM-kinase is isolated from both cytosolic and particulate fractions (36 -39). How CaM-kinase interacts or associates within these subcellular compartments, and whether the enzyme's distribution is regulated is unknown. The changes in the sedimentability of purified CaM-kinase observed during conditional autophosphorylation in this study may reflect conformational changes in the enzyme that enable or promote protein interactions with itself and/or other proteins. Subcellular translocation of particulate to soluble CaM-kinase during autophosphorylation has been reported in the Aplysia ganglia (6), and the redistribution of the cytosolic enzyme into the particulate fraction has been reported during neuronal activation (40,41) and following epileptic (42) and ischemic conditions (25,26,43). Cerebral ischemia is one well documented example in which dramatic and rapid fluctuations occur to both pH and ATP. Silver and Erecinska (44) demonstrated that, within minutes, an ischemic insult altered the intracellular pH in hippocampal neurons from approximately pH 7.4 to pH 6.2. Folbergrova et al. (45) reported that the physiological ATP concentration is approximately 2.8 mmol/kg of cortical tissue, and that during ischemic insult, ATP stores are virtually depleted within minutes. It is currently unknown whether autophosphorylation plays a role in the inactivation and redistribution of soluble CaM-kinase into the particulate fraction during ischemia.
Our in vitro data predict that both a decrease in pH and ATP concentration would result in alterations in the physical and catalytic properties of CaM-kinase analogous to the changes observed in vivo and in situ during an ischemic insult (25,26,43). Our in vitro data further predict that activation of cellular CaM-kinase in conditions of neutral pH and high ATP would result in the production of a predominantly soluble enzyme with little corresponding inactivation. In contrast, the activation of CaM-kinase in conditions of neutral pH and low ATP ultimately would result in the production of a predominantly inactive soluble enzyme. The data presented in this report suggest that different intracellular environments will modify the functional output of CaM-kinase through autophosphorylation-associated changes in the activity and/or association of the enzyme with itself and potentially other proteins. The reversibility of autophosphorylation-associated inactivation and/or self-association is unknown. Bidirectional regulation is a hallmark of physiological regulatory mechanisms and the potential to yield active, soluble enzyme from sedimentable enzyme possibly through dephosphorylation and/or other conditions warrants future investigation. Acknowledgments-We thank Dr. James Stoops, Dr. Jack Waymire, Dr. Pramod Dash, and Anthony Moore for helpful discussions and critical review of this manuscript. We are indebted to Tara Vicknair for assistance in preparation of figures. We also thank Dr. Howard Schulman for generously providing monoclonal antibodies to CaM-kinase. APPENDIX In an attempt to estimate the number of holoenzymes recovered as enzyme complexes (see Fig. 7, A and B), the micrographs were digitized and the electron dense particles quantified using Image software (version 1.44, Wayne Rasband at NIMH). The steps in analysis are given in Table I. Several assumptions were required to complete the calculations.
First Assumption: Holoenzyme Is a Cylinder-The diameter of a holoenzyme has been estimated to be approximately 30 nm (34). The "dumbbell" shaped models proposed by Kanaseki et al. (34) appear cylindrical; however, no published information is available concerning the holoenzyme's height. We utilized a uniform height of 5 nm, a size slightly smaller value than the predicted size of the "lollipops" (6 nm) described by Kanaseki et al. (34).
Second Assumption: Enzyme Complexes Are Spherical-The average diameter of an enzyme complex is roughly 200 nm. However, two-dimensional projections make it impossible to determine whether the volume occupied is spherical, elliptical or discoidal. A spherical volume is the easiest geometry to model in the absence of any three-dimensional information.
Third Assumption: Holoenzymes Packed into Complexes Are Modeled as Nonoverlapping 30-nm Diameter Spheres-The packing density of the holoenzymes within the enzyme complexes is unknown (i.e. whether the holoenzyme's volume is altered by aggregation). We hypothesize that the space occupied by aggregating holoenzymes condenses; however, we have no information to model these changes. Therefore, we have taken the simplest possibility and assumed that the holoenzymes within the enzyme complex retain their initial starting volume.
Fourth Assumption: Total Number of Enzyme Complexes Is Uniform over the Entire Grid-We have assumed a uniform protein distribution over the entire grid because we have no way to determine whether the copper mesh and frame of the grid influences the distribution of the deposited protein.
Fifth Assumption: 27% of the Starting Protein Is Deposited on the Grid-Although we were unable to quantify the protein deposited on the carbon-coated grid, the difference in the protein recovered in the microconcentrator in the presence and absence of a grid was quantifiable. Approximately 73% Ϯ 12% (n ϭ 4, with 2 microconcentrators per n) of the protein was recovered in the microconcentrator in the presence of the grid, indicating that only 27% of the protein applied to the microconcentrator was potentially deposited on the grid. Whether all of this protein (27% of the total or 0.27 g) is actually deposited on the grid or whether staining and washing further alters this value is unknown. This corrected value was used as the amount of starting enzyme for the analysis.