Muscle-type MM Creatine Kinase Is Specifically Bound to Sarcoplasmic Reticulum and Can Support Ca2+ Uptake and Regulate Local ATP/ADP Ratios*

Highly purified fractions of sarcoplasmic reticulum (SR) were prepared from chicken pectoralis muscles (Saito, A., Seiler, S., Chu, A., and Fleischer, S. (1984) J. Cell Biol. 99, 875-885) and analyzed for the presence of creatine kinase (CK). Vesicles derived from longitudinal SR contained 0.703 +/- 0.428 IU of CK/mg of (SR) protein. Immunogold localization of muscle-type MM-CK on ultrathin cryosections of muscle, after removal of soluble CK, revealed relatively strong in situ labeling of M-CK remaining bound to the M band as well as to the SR membranes. In addition, purified SR vesicles were also labeled by anti-M-CK antibodies, and the peripheral labeling was similar to that observed with anti-Ca2(+)-ATPase antibodies. Only some particulate CK enzyme was released from isolated SR membranes by EDTA/low salt buffer, and CK was resistant to extraction by 0.6 M KCl. Thus, some of the MM-CK present in muscle displays strong associative behavior to the SR membranes. The SR-bound CK was sufficient to support, in the presence of phosphocreatine plus ADP, a significant portion of the maximal in vitro Ca2+ uptake rate. The ATP regeneration potential of SR-bound CK was similar to the rate of Ca2(+)-stimulated ATP hydrolysis of isolated SR vesicles. Thus, CK bound to SR may be physiologically relevant in vivo for regeneration of ATP used by the Ca2(+)-ATPase, as well as for regulation of local ATP/ADP ratios in the proximity of the Ca2+ pump and of other ATP-requiring reactions in the excitation-contraction coupling pathway.

It is composed of two continuous yet heterogeneous compartments, namely the longitudinal sarcoplasmic reticulum (LSR) surrounding the myofibrils, and the terminal cisternae (TC), junctionally associated with the transverse T tubule membrane by bridging "feet" structures (2).
The major protein component of skeletal muscle SR is the Ca'+/Mg2'-dependent ATPase (Ca'+-ATPase, EC 3.6.1.38). This transmembrane enzyme, which plays a crucial role in Ca*+ homeostasis, is responsible for the ATP-dependent Ca*+ uptake into the lumen of the SR, enabling muscle to relax (3). The Ca2+ pump, requiring a considerable amount of ATP for muscle relaxation, is uniformly distributed within the SR membranes except for the junctional face membrane of the TC, which is essentially free of the enzyme (2, 4).
In muscle it has been shown that small amounts of total CK are bound in an isoenzyme-specific fashion to subcellular structures such as the inner mitochondrial membrane (5, 6), the M band of myofibrils (7-9), and the plasma membrane (10, 11). These observations have lent support to the "CP shuttle" (6,(12)(13)(14) or "phosphocreatine circuit" models (48), which suggest that communication and transfer of energy between ATP-generating (mitochondria and cytosolic glycolysis) and ATP-utilizing sites (myofibrillar actin-activated Mg*+-ATPase, (Na'/K')-ATPase) are facilitated via the CP/ CK system.
Since in muscle a significant amount of energy is spent for Ca2+ sequestration by the SR Ca2+-ATPase, and the local ATP/ADP ratio may be a critical regulatory parameter for the Ca2+ pump, one might predict from the recent phosphocreatine circuit model (48) that CK forms a functionally coupled compartment with the Ca'+-ATPase at the SR, as was suggested earlier (15, 16; see below), and thus would be an in situ ATP regenerator in supporting Ca2+ uptake and an in situ regulator of local ATP/ADP ratios. The present work describes experimental evidence for the presence of creatine kinase (ATP: creatine N-phosphotransferase, EC 2.7.3.2) bound to highly purified SR fractions and for a direct functional coupling of some of the SR-associated CK with the Ca2+ pump. The presence of CK attached to SR has been suggested by Baskin and Deamer in 1970 (15), and some experimental evidence for such an association was provided by Levitsky et al. in 1978 (16). However, since the bulk of CK in muscle is soluble, it was important to exclude adventitious binding of the enzyme to SR membranes and to characterize the specificity of CK binding to the SR with highly purified SR vesicles exposed to various ionic condi-  (30,31).

Evidence for MM-CK Specifically
Bound to SR Membranes-Analysis of the protein composition of purified chicken skeletal muscle SR fractions (Rl-R4) obtained by the procedure described by Saito et al. (17), which also worked very well for chicken muscle, revealed the presence of a major polypeptide band with an apparent M, of 97,000, corresponding to the Ca2+-ATPase subunit (that is the main constituent of the SR membranes) in all four SR fractions (Fig. 1, lanes a-d). However, the relative amount of Ca*'-ATPase was highest in the R2 fraction (lane b), corresponding to longitudinal SR tubules. The R4 or TC fraction (lane d), consisting of two different types of SR membranes, namely the junctional face membrane and the Ca*+ pump membrane, which, in the rabbit, represent 1520% and 80% of the total membranes present in R4 fractions, respectively (17), showed a higher content of calsequestrin, with an apparent M, of 55,000 (25), than Rl, R2, and R3 fractions. Calsequestrin, known to be specifically localized in the lumen of TCs (32), stained blue after staining with Stains-All ( Fig. 1, CS, black asterisk).
Other proteins, also stained blue by this particular dye as already described earlier (33), were the 130-kDa and the 120-kDa =rg,-cs 42- High M, proteins, presumably feet proteins of junctional face membranes, are seen mainly in fraction R4 (Fig. 1, lane d). Besides these proteins, even in highly purified SR preparations, a number of other more or less prominent proteins, some of them still unidentified, can be seen on heavily overloaded sodium dodecyl sulfate gels, as also pointed out by other investigators (17,38). Immunoblotting with anti-M-CK antibodies of the SR fractions, obtained by sucrose gradient centrifugation and seen as four distinctly layered bands (Rl-R4), demonstrated the presence of significant amounts of M-CK in all four of these SR fractions (Fig. 1, lanes a/-d'). None of the four SR fractions contained intact myof'ibrils or significant amounts of mitochondria, as shown by electron microscopical analysis of embedded SR vesicles (see Fig. 4). In addition, only small amounts of myosin contamination were detected by immunoblotting in all four SR membrane fractions (Fig. 20), and the same was true for mitochondria (see below). In order to verify that the CK bound to these SR fractions was not simply due to contamination or adventitious binding from bulk soluble CK or due to small amounts of myosinbound CK (7, 14), aliquots of all four SR fractions were extracted with a 50-loo-fold excess of either high salt (0.6 M KCl) or low salt solution (10 mM Tris, 1 mM EDTA, pH 8.0). Extraction by 0.6 M KC1 released various proteins from the SR fractions, which are recovered in the supernatants as seen in instead, most of the M-CK remained bound to the SR vesicles and appeared on the immunoblot in the salt-extracted SR pellet fractions (Fig. 2C, lanes a', c', e', and g'). Conversely, the EDTA/low salt treatment ( Fig. 2A), known to open SR 97- Heavily loaded Coomassie Blue-stained gel of SR fractions after extraction with 50 volumes of either 10 mM Tris, 2 mM EDTA, pH 8.0 (panel A), or 0.6 M KC1 in buffer A (panel B) and the corresponding supernatants thereof to test for released M-CK. On both panels, Rl, R2, R3, and R4 fractions were loaded in lanes a, c, e, and g, respectively, and the corresponding supernatants in lanes b, cl, f, and h. Lane M contains a myosin standard. Low molecular mass protein standards are indicated in kDa on the fur left on each panel. The lower panels show the region of interest of the blot corresponding to the upper gel, after electrophoretic transfer of protein bands to nitrocellulose followed by staining with specific rabbit anti-chicken M-CK antibodies (diluted 1:lOOO) (panel C) and anti-chicken skeletal myosin antibodies (diluted 1:500) (panel D), both followed by second peroxidase-labeled goat anti-rabbit IgG (diluted 1:2500) and staining for peroxidase activity. The extracted pellets were resuspended in 1 ml of buffer A, and the corresponding supernatants were lyophilized in order to get the same volume (1 ml) as the resuspended pellets. KC1 supernatants were dialyzed against buffer A before lyophilization. The same volume of each fraction was applied/lane. EDTA/low salt buffer opens the vesicles and releases their content; note the calsequestrin band (55 kDa) enriched in the supernatant of R4 (lane h). M-CK is still bound to the SR membranes of all SR fractions after both high and low salt treatment (panel C). No CK is released by KC1 extraction (panel C, lanes b', d', f', and h'), but by contrast, some CK, however not exceeding 30% of the total, as determined by activity measurements (Table I) vesicles and to release their contents (34), seems to set free some CK into the supernatants.
However, the activity directly measured by the enzyme assay in such supernatants (see Table I) amounts only to X-30% of the initial CK activity. Thus, the MM-CK bound to the various SR membranes at the end of the rather lengthy purification procedure, involving many washing and centrifugation steps (17), is resistant to extensive extraction by high and low salt buffers. Therefore,   (26). The values reported are the mean & S.D. for nine determinations from several different SR vesicle preparations.
'SK fractions were extracted either for 1 h with 0.6 M KC1 in buffer A or for 30 min with 10 mM Tris, 1 mM EDTA, pH 8.0 (see legend of Fig. 2). CK activity (NJ/ml) measured both of the extracted pellets, and the corresponding supernatants are expressed as percentages of the initial CK activity (N/ml) of the various SR fractions. The values reported are the mean of three determinations, each from the same preparation.
under both of these extreme conditions, CK still displayed strong associative behavior to SR membranes, indicative for a specific interaction of CK with SR vesicles. Under these conditions, CK bound to myofibrils at a molar ratio of 1 CK molecule/20 myosins (9, 12, 14) is completely solubilized by both high or low salt buffers (9,14). For this reason, the CK activity determined in our SR preparations is unlikely due to myofibrillar or myosin-bound CK, for the myosin contamination present in all four SR fractions was very low (see small and faint bands in Fig. 20 after staining with excess of antimyosin antibody), as indicated by the fact that during low and high salt extraction only small amounts of myosin were either precipitated (Fig. 20, lanes a-h) or released into the supernatants (Fig. 20, lanes a'-!~'). This also indicates that only one of the faint upper bands and not the prominent band itself seen around 200 kDa in Fig. 2, A and B, is myosin (this is seen best by comparing lanes g' and h' of Fig. 2A with the same lanes in Fig. 20).
The CK isoenzymes in the SR fractions were identified by cellulose-polyacetate electrophoresis under native conditions followed by staining for CK activity. All four SR fractions contained MM-CK as the most prominent isoform, as judged by the strong CK activity staining (Fig. 3, MM-CK), whereas the heavier SR fractions (especially R3 and R4) showed in addition a faint band corresponding to mitochondrial CK (Fig. 3, lanes c, d, and e, Mi-CK). However, direct measurements of the specific succinate-cytochrome c reductase activity (Table II) showed that mitochondrial contamination was less than 1% in all cases (for calculations, see legend of Table  II). The low contamination by mitochondria of the SR preparations was also confirmed by electron microscopy (see below). An additional faintly stained band moving toward the anode was most probably due to some MB-CK isoenzyme seen in R2-R4 (Fig. 3, lanes b-d). Omission of CP from the overlay gel resulted in a very weak band only at the MM-CK position within the Rl fraction (not shown), indicating the presence of trace amounts of adenylate kinase.
Amount of CK Activity Bound to the SR-All four SR fractions contained considerable specific CK activities ranging from 0.7 to 1.4 IU/mg of protein (Table I)  Mitochondrial contamination of SR fractions was evaluated by measurmg succinate-cytochrome c reductase activity as described by Fleischer and Kervina (27). Since native beef heart mitochondria display a specific activity of 0.8-0.9 pmol of cytochrome c reductase/ min/mg of protein, the values shown in this This confirmed that CK is rather specifically bound to SR vesicles withstanding high salt extraction.
The release of a certain amount of CK by EDTA/low salt treatment may indicate either that some CK may have gotten trapped inside the SR vesicles during purification or that the interaction of CK bound to the outside of SR vesicles is rather sensitive to alkaline pH. The incomplete recovery of the initial CK activity after 0.6 M KC1 extraction may be explained by inhibition and/or denaturation of some of the SR-bound CK since the lack of any immunoreactive material on the blot (Fig. 2C, lanes b', d', f', and h') suggests that not even inactive enzyme protein is present in the KC1 supernatants. Therefore, we assume that an additional lo-15% of inactivated CK, not detectable by the enzyme assay, would also be contained in the KC1 pellets, as also suggested by the stronger reaction on the immunoblot of the CK bands of the KC1 pellets compared with those of the EDTA/low salt pellets (Fig. 2C). Characterization of SR Vesicles by Electron Microscopy- Fig. 4 (R2) shown at a magnification of ~140,000 with a bar corresponding to 100 nm. Note the rather homogeneous appearance of the R2 SR vesicle population and the absence of myofibrillar and mitochondrial contaminants.
taining mostly longitudinal SR as judged from the rather homogeneous population of vesicles devoid of electron-opaque content and from the absence of myofibrils with no or only very few mitochondria (Fig. 4). Mitochondria were routinely seen only in R4. The size of the LSR vesicles ranged from 70 to 300 nm, which is in agreement with published data (17). Scanning electron microscopy of the R2 fraction showed the rather smooth surface of the vesicles (not shown).
In Situ Localization of CK at the SR in Skeletal Muscle-Since skeletal and cardiac muscle contains large amounts of CK (14), and only fractions of the total enzyme activity are tightly associated with subcellular structures, it was necessary to remove the soluble part of the enzyme for in situ localization of CK bound to SR. Therefore, muscle tissue was either permeabilized and washed with relaxing solution prior to chemical fixation, ultracryotomy, and immunolabeling, or ultrathin cryosections of mildly fixed muscle were incubated extensively in relaxing solution prior to postfixation and immunostaining.
Immunogold labeling of these cryosections with specific anti-M-CK antibodies revealed bound CK at the M line (Fig. 5) of the sarcomer, a well established location of M-CK (7,8,14), and additionally at those places near the Z band and A-I band junctions, where T tubules and SR are prominent (Fig. 5A, arrowheads). The latter is seen more clearly at higher magnification (Fig. 5B, arrowheads). By contrast, no significant labeling was observed with preimmune antibody (Fig. 5C). Even though in negatively stained ultrathin cryosections SR membranes are difficult to visualize directly (Ref. 4), the immunolabeling shown in Fig. 5, A and B, is topologically consistent with the location of SR where membranous structures near the A-I junctions and along the myofibrils were often seen to be stained by gold antibodies (Fig. 5, A and B; see arrowheads).
The in situ immunolocalization of CK at the SR agrees well with the immunogold labeling of isolated SR vesicles with anti-M-CK antibodies (Fig. 6) and argues against adventitious binding of CK to SR vesicles during their isolation. Similar results have been obtained earlier by in situ histochemical staining of muscle for CK enzyme activity (16). However, the diffusibility of the various reaction products generated by the coupled enzymestaining technique for CK activity makes the interpretation of these results very difficult. A direct in situ immunolocalization of the M-CK protein with anti-M-CK antibodies, as shown here, seems to be far superior.
Localization of M-CK and Ca'+-ATPase on Purified SR Vesicles by Immunoelectron Microscopy-LSR vesicles were adsorbed to glow-discharged carbon film, incubated with pri-mary antibodies either against Ca'+-ATPase or M-CK, and then incubated again with secondary gold-conjugated antibodies. After negative staining, LSR vesicles showed specific labeling by both anti Ca'+-ATPase and anti-M-CK antibodies (Fig. 6, A and B). On the other hand, incubation with identical concentrations of preimmune IgG showed only low background labeling (Fig. 6C). The labeling with anti-Ca"-ATPase antibodies appeared very strong and often clustered at the surface of the vesicles (Fig. 6B). Considering the fact that M-CK, relative to the Ca"-ATPase, is only a minor protein of the SR (Fig. l), the extent of labeling with anti-M-CK was rather high (Fig. 6A). The distribution of the gold clusters was similar in both cases, indicating that some epitopes of both enzymes are indeed exposed on the outside of the SR vesicles and thus are accessible to immunogold staining. Extent of CK-supported A TP-dependent Ca" Uptake by SR Vesicles-In order to investigate the functional role of SRbound CK, the Ca"+-loading rate by SR vesicles was measured by the in vitro assay described under "Experimental Procedures," using mainly the homogeneous well characterized R2 SR fraction where Ca"-ATPase was highly enriched. In a first set of experiments, the Ca2+ loading of LSR vesicles in the presence of excess ATP (1 mM) was determined ( Fig. 7B and Table III) and taken as a control value corresponding to 100%. Under these conditions, a maximal Ca*+-loading rate was observed since increasing the ATP concentration above 1 mM did not enhance Ca'+-pumping activity. The reaction was always started by the addition of 25 nmol of Ca*+. When DNFB, a specific inhibitor of CK if used at low concentration (35), was added at 25 pM concentration at the end of the Ca2+ uptake, ATP-supported Ca2+ loading was resumed upon a second addition of Ca*+ (Fig. 7B).
In a second set of experiments, ATP was substituted by 1 mM ADP plus 10 mM CP, both substrates of the CK reaction, and then the CK-supported Ca'+-loading rate was measured again (Fig. 7A and Table III). The relative Ca'+-loading rate, supported by RB-bound CK and CP plus ADP, was approximately 24 + 5% compared with that measured in the presence of ATP (control).
When DNFB was added to the vesicles ( Fig. 7 and Table III), the Ca2+ uptake slowly ceased as a consequence of the inhibition of CK and did not recover after a second addition of Ca2+ (Fig. 7A) due to irreversible inactivation of CK by DNFB. However, upon addition of ATP to the same preparation, Ca2+ loading was resumed, albeit at a somewhat lower absolute rate than the control (Fig. 7A, end of tracing), indicating that the Ca2'-ATPase itself was only slightly affected by DNFB. In fact, the slight inhibition exerted by DNFB (approximately 17%) on the Ca2+-ATPase in the presence of an excess of ATP was actually due to the ethanol used as a solvent for the DNFB (Table III). Taken together, these findings suggest that endogenous SR-bound CK of highly purified SR vesicles is indeed capable of supporting Ca*+ uptake to a considerable extent by local in situ regeneration of ATP. When the assay was performed in the presence of CP and ADP plus an excess of exogenously added CK, the Ca2+loading rate measured was similar to the control value. In a further set of controls, Ca2+ uptake was measured in the presence of ADP alone to evaluate a contribution to ATP regeneration by the adenylate kinase or myokinase reaction. However, myokinase-supported Ca2' loading was negligible (Table III). In order to exclude that the mitochondrial ATP production significantly affected the above results, Ca2+ uptake measurements were made in the presence of either succinate plus ATP or succinate plus ADP as potential energy sources for oxidative phosphorylation.
In the first case, as expected, the Ca*' loading was similar to the control value; whereas in the latter case, no Ca2+ uptake was measured (Table III). The presence of rotenone, a known inhibitor of mitochondrial electron transport chain (37), did not decrease or shorten the ATP-dependent Ca*+ loading by the LSR vesicles (Table III), indicating that under our conditions mitochondria or mitochondrial CK were not involved in supporting Ca2+ uptake. This confirmed that the very low contamination of the SR fractions, especially of R2, by mitochondria (as shown in Table II) had a negligible effect on the Ca*+ uptake measurements. The assay was also performed in the presence of CP alone (data not shown), but no Ca*+ pumping was observed.
Furthermore, rates of Ca*' loading and CK-supported Ca2+ loading were also determined using the R3 fraction, representing a mixture of LSR plus TC. The CK-supported Ca'+loading rate was found to be 42% of the control value (Table  III). As expected, the absolute Ca*+-loading rate of R3 was slower due to the fact that this fraction contained less Ca*+-ATPase. However, since it contained a higher CK content compared with the R2 fraction (Table I), the percentage of CK-supported Ca*+ pumping went up to over 40% (Table III). ATP Regeneration Potential by SR-bound CK-The amounts of CK bound to the different SR vesicle fractions (Table 1) and the observed in vitro Ca'+ uptake rate of the same fractions (Table III)  Immunogold labeling of longitudinal R2 SR vesicles adsorbed to glow-discharged carbon films after incubation with rabbit anti-chicken M-CK (panel A ), rabbit anti-Ca"-ATPase antibodies (panel B), or preimmune IgG (panel C). As second antibody, goat antirabbit IgG conjugated to 10 nm colloidal gold particles was used. Note specific labeling of R2 vesicles by both anti-M-CK and anti-Ca"'-ATPase antibodies and low background staining of the control in panel C. Magnification, ~100,000 with a bar corresponding to 60 nm. FIG. 7. Ca*+ accumulation rate into SR vesicles directly supnorted bv SR-bound CK. The Ca'+-loadinc rate into R2 vesicles was measured under various conditions (see also Table III). Representative tracings are shown. The assay was performed with 50 rg of SR proteins. Tracing a, Ca*+ loading in the presence of 10 mM CP plus 1 mM ADP initiated by the addition of 25 /rmol of Ca'+. The Ca'+-loading rate supported by SR-bound CK was 0.576 pmol of Ca'+ . min-' . mg-' of SR protein. After the addition of 25 j.tM DNFB, Ca2+ loading was abolished due to blockade of CK but recovered upon addition of 1 mM ATP as in the control and was stimulated again by adding Cal+ (Ca'+-loading rate 0.547 pmol min-' mg-' of SR protein). Tracing b, Ca'+ loading in the presence of 1 mM ATP as a control was initiated by the addition of 25 pmol of Ca2+. The Ca'+loading rate was 1.369 gmol of Ca*'. min-' . mg-' of SR protein. The addition of 25 pM DNFB followed by readdition of Ca2+ only slightly inhibited Ca'+ loading. tudinal SR vesicles (Table I), which display a maximal ATPase rate of approximately 1,713 f 0.138/2 = 0.8 ATP.min-'+ mg-' of R2 (Table III) if a loading efficiency of 2Ca*+/ATP is taken as the correct value (49). In the heavier SR fractions (R3 and R4) where the ratio of SR-associated CK to Ca*+-ATPase is higher, the ATP regeneration potential by the bound CK exceeds the ATP required for Ca*+ loading, indicating that CK may, in addition, be involved in processes other than simply regenerating the ATP required for the Ca*+ pump. Thus, even though the direct support of Ca"' loading in SR vesicles by CK is not sufficient in our in vitro assay, the ATP regeneration potential of SR-bound CK should be sufficient to support fully the in uiuo Ca2+ loading (see "Discussion").

DISCUSSION
In a previous study, Volpe et al. (38), using photoactivatable phospholipid analogues, noticed the presence of a 41-kDa integral membrane protein in highly purified SR preparations, which, at that time, was not identified.
We were therefore curious to see whether this protein, showing an apparent M, very similar to that of the M-CK monomer, was membranebound CK. We have demonstrated by biochemical experiments and by immunoelectron microscopy using anti-M-CK antibodies that CK is indeed bound specifically to all four SR fractions obtained from chicken skeletal muscle and that significant amounts of CK were still bound to the SR mem- branes after both low and high salt treatments, indicating a rather tight and specific binding of CK to the SR membranes. After EDTA treatment, only X-30% of the initial CK activity was released into the supernatant.
This relatively small fraction could be derived either from soluble CK that may have gotten trapped inside SR vesicles during their purification or from CK bound to the outside of the SR vesicles, which was dissociated by alkaline pH, low salt, or EDTA. However, the fact that SR-bound CK completely resisted extensive extraction by 0.6 M KC1 suggests a strong associative behavior of this enzyme to SR membranes and argues against a significant contamination of the SR preparation by myofibrillar or myosin-bound CK, which is known to be released into the supernatant by high and low salt buffers (12,14). The presence of CK bound to SR membranes had been suggested earlier by Baskin and Deamer (15) and Levitsky (16); however, these authors neither characterized the SR vesicles to exclude adventitious binding of CK to these vesicles, nor did they specify the CK isoform involved.
In previous studies concerning myofibrils, the M line-bound CK was shown by a similar coupled in vitro assay to suffice completely for the regeneration of ATP hydrolyzed by the actin-activated Mg'+-ATPase during in vitro contraction of myofibrils (26). In the present experiments with SR, the Ca*+ uptake rate was slower if solely supported by endogenous SRbound CK, as compared with the control rate measured in the presence of excess ATP. However, the contribution of the in situ ATP regeneration by the SR-bound CK in the presence of excess CP plus ADP was significant.
In vitro, SR-bound CK supported 24-40% of the maximal Ca2+ uptake rate, depending on which of the SR vesicle fractions were taken. This could mean that some of the CK originally bound to the SR may have been lost during fractionation and extensive washing of the SR vesicles or that in this in vitro assay, working at very low concentrations of vesicles, some of the ATP regenerated in situ on the surface of the SR vesicles may continuously diffuse away and thus may be lost for immediate utilization by the Ca2+ pump. The latter explanation also seems reasonable since a comparison of the ATP regeneration potential of CK on the R2 SR vesicles (0.703 pmol of ATP regenerated/min/mg of R2 protein; see Table I) with the ATP hydrolysis rate of the Ca"-ATPase at maximal speed (1.713 Fmol of Ca*+/min/mg of R2 protein; see Table III) shows that the two numbers are indeed very close, if an ideal coupling ratio of 2Ca2'/ATP is assumed (49). Thus, under ideal conditions and even more so in uiuo where diffusion of nucleotides is severely limited, the SR-bound CK may suffice for supporting fully the Ca*+ uptake.
The fact that the ratio of bound CK to Ca"-ATPase activity is significantly higher in R3 and R4 (compare CK activity values in Table I versus Ca*+-loading rate divided by 2 in Table III) shows that in this region of the SR the ATP regeneration potential of bound CK exceeds the ATP consumption by the Ca2+ pump in vitro. In addition, the fact that the amount of CK activity is not proportional to the Ca*+-ATPase content of the various SR subfractions (Rl-R4) indicates that CK, especially at the junctional face membrane (R4), may have additional physiological functions within the excitation contraction pathway other than simply supporting Ca*+ sequestration (see below). Thus, it seems that even if we assume that the overall ATP concentration in a cell remains constant, the CK present at the SR may be physiologically important for local regeneration of ATP directly in the vicinity of the Ca*+ pump, where, under extremely heavy work load, metabolic stress, or ischemia, the ATP/ADP ratio is bound to be quite different from that in the rest of the myoplasm. The same explanation may hold true for the fact that lowering the ADP concentration to less than 0.5 mM in the CK-and CP-supported Ca*+-loading experiments, using the coupled in vitro assay, started to reduce the Ca'+-loading rate (not shown). A concentration of 0.5 mM ADP is well above the Km of M-CK for ADP (36) and is also higher than the cellular overall ADP concentration found in uiuo, except under heavy work load. However, under the latter conditions, the ADP concentration may increase locally very quickly to such a level, and thus a more efficient functional in situ coupling of CK with the Ca*+ pump may be guaranteed. Champeil et al. (39) have suggested that the SR Ca*+-ATPase itself is regulated by local ATP levels, e.g. that dephosphorylation is regulated by [ATP] and that the catalytic site is the locus of the "regulatory" ATP-binding site. Thus, the membrane-bound CK may not only be important for replenishment of ATP, but also for fine-tuning of local ATP levels and more important for regulation of local ATP/ ADP ratios. The maintenance of high local ATP/ADP ratios increases the thermodynamic efficiency of ATP hydrolysis, which depends critically on the In (ADP).(PJ/(ATP) (see Ref. 50). As a third possibility, CK may direct the ATP needed