RC3/Neurogranin and Ca2+/calmodulin-dependent protein kinase II produce opposing effects on the affinity of calmodulin for calcium.

The interaction of calmodulin with its target proteins is known to affect the kinetics and affinity of Ca(2+) binding to calmodulin. Based on thermodynamic principles, proteins that bind to Ca(2+)-calmodulin should increase the affinity of calmodulin for Ca(2+), while proteins that bind to apo-calmodulin should decrease its affinity for Ca(2+). We quantified the effects on Ca(2+)-calmodulin interaction of two neuronal calmodulin targets: RC3, which binds both Ca(2+)- and apo-calmodulin, and alphaCaM kinase II, which binds selectively to Ca(2+)-calmodulin. RC3 was found to decrease the affinity of calmodulin for Ca(2+), whereas CaM kinase II increases the calmodulin affinity for Ca(2+). Specifically, RC3 increases the rate of Ca(2+) dissociation from the C-terminal sites of calmodulin up to 60-fold while having little effect on the rate of Ca(2+) association. Conversely, CaM kinase II decreases the rates of dissociation of Ca(2+) from both lobes of calmodulin and autophosphorylation of CaM kinase II at Thr(286) induces a further decrease in the rates of Ca(2+) dissociation. RC3 dampens the effects of CaM kinase II on Ca(2+) dissociation by increasing the rate of dissociation from the C-terminal lobe of calmodulin when in the presence of CaM kinase II. This effect is not seen with phosphorylated CaM kinase II. The results are interpreted according to a kinetic scheme in which there are competing pathways for dissociation of the Ca(2+)-calmodulin target complex. This work indicates that the Ca(2+) binding properties of calmodulin are highly regulated and reveals a role for RC3 in accelerating the dissociation of Ca(2+)-calmodulin target complexes at the end of a Ca(2+) signal.

␣CaM kinase II is the major neuronal isoform of this multifunctional CaM-dependent enzyme (1,23). CaM binding in the regulatory domain of CaM kinase II activates the enzyme by relieving autoinhibition. CaM kinase II is then able to autophosphorylate at Thr 286 , which leads to both Ca 2ϩ /CaM independent activity and CaM-trapping, a phenomenon in which the affinity of the enzyme for CaM is increased by more than 1,000-fold (24). Our previous work identified that binding of Ca 2ϩ /CaM to unphosphorylated CaM kinase II increases the Ca 2ϩ binding affinity of CaM by about 40-fold (14). Because phosphorylation increases the CaM binding affinity of CaM kinase II, we predict a further increase in the Ca 2ϩ binding affinity of CaM when the kinase is in the phospho-Thr 286 state.
We previously showed that the interaction of PEP-19 with CaM accelerates the exchange of Ca 2ϩ from the C-terminal lobe of CaM (14). We now show that interaction of RC3 with CaM accelerates the dissociation of Ca 2ϩ from the C-terminal lobe, but has no effect on association rates, leading to an up to 60-fold decrease in the affinity of Ca 2ϩ for the C-terminal lobe of CaM. The different effects of RC3 and PEP-19 on the interactions of Ca 2ϩ with CaM suggest that the IQ domain proteins have diverse modulatory effects on CaM function. Interaction of CaM with CaM kinase II causes a decrease in the rates of Ca 2ϩ dissociation from both lobes. When present together, RC3 and CaM kinase II have non-linear effects on Ca 2ϩ dissociation; RC3 dampens the effects of CaM kinase II by causing Ca 2ϩ to dissociate more quickly from the CaM/CaM kinase complex, as we saw with PEP-19 (14). The data suggest that RC3 has the potential to decrease the lifetime of the Ca 2ϩ /CaM bound state of CaM kinase II presumably by preventing reassociation of CaM kinase with CaM. However, RC3 does not increase the rates of dissociation of Ca 2ϩ from the CaM/CaM kinase II complex after autophosphorylation of the kinase, a state in which slower Ca 2ϩ dissociation is seen, likely because the dissociation rate of CaM from phospho-CaM kinase is so slow that this dissociation pathway (kinase dissociates first, then Ca 2ϩ ) is not favored. The refractory nature of the Thr 286 phosphokinase to RC3 adds an additional layer of complexity to the intricate modulation of CaM function in neurons. The opposite effects of RC3 and CaM kinase II on the interaction of CaM with Ca 2ϩ indicate that bidirectional tuning of the CaM affinity for Ca 2ϩ plays an important role in regulating the CaM response to Ca 2ϩ signals.

EXPERIMENTAL PROCEDURES
CaM and CaM Kinase II Expression and Purification-A cDNA for sea urchin CaM was codon optimized for expression and mutated to produce an identical protein sequence to vertebrate CaM. The cDNA in pET-23d was expressed in Escherichia coli and purified as described (25). A rat ␣CaM kinase II cDNA was expressed in Sf21 cells as previously described (26) and purified on a phosphocellulose cation exchange column as described by Bradshaw et al. (27).
RC3 Expression and Purification-A cDNA encoding the full-length RC3 protein from rat was originally obtained from Dr. Dan Gerendasy and the cDNA sequence was confirmed in our laboratory. RC3 was expressed from the pET23 vector in the BL21 DE3 pLys S strain of E. coli following the expression protocol detailed by Gerendasy et al. (28). To purify RC3, the following steps were performed at 4°C. Cell pellets were lysed in a cold hypotonic lysis buffer (10 mM HEPES, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride), with sonication. The lysate was cleared by centrifugation at 25,000 ϫ g, and the resulting supernatant was brought to 30% ammonium sulfate. The precipitate was removed by centrifugation at 10,000 ϫ g for 30 min, and then the supernatant was brought to 65% ammonium sulfate, causing precipitation of RC3. After centrifugation at 10,000 ϫ g for 30 min, the pellet was resuspended in 15 ml of 50 mM Tris, pH 6.8, 50 mM NaCl, 1 mM DTT, and perchloric acid was added to 2.5% final concentration. The mixture was centrifuged at 10,000 ϫ g for 15 min, and then the supernatant from this spin containing RC3 was brought to 90% ammonium sulfate. After a final spin at 10,000 ϫ g for 30 min, the pellet was resuspended in 10 mM HEPES, 200 mM KCl, 1 mM DTT, and dialyzed overnight into the same buffer. The RC3 was then subjected to high performance liquid chromatography using a Vydac C4 column and was eluted with a 0 -60% gradient of water ϩ 0.13% trifluoroacetic acid to acetonitrile ϩ 0.13% trifluoroacetic acid, over 30 min. Peak fractions were detected by UV absorption and the peak containing RC3 (that eluted at 38 min) was identified by SDS-PAGE. Peak fractions were pooled and lyophilized, resuspended in water, and relyophilized. For each set of experiments, a 1 mM stock solution of RC3 was made in 20 mM MOPS, pH 7.5, 100 mM KCl, and 1 mM DTT. Before each experiment, the RC3 was treated with 50 mM DTT at 55°C for 15 min to minimize the disulfide-bonded intermediate.
Decalcification of Buffers and Proteins-For Ca 2ϩ titration experiments and Ca 2ϩ association experiments, all solutions and proteins were decalcified before use by passing them over a small column of Calcium Sponge resin (Molecular Probes, Eugene, OR). All pipette tips, cuvettes, and other labware were rinsed with 0.1 M HCl and MilliQ water to remove Ca 2ϩ . The absence of Ca 2ϩ was confirmed by comparing Tyr fluorescence of diluted CaM before and after the sequential addition of 0.1 mM EGTA and 1 mM Ca 2ϩ . Residual Ca 2ϩ in decalcified buffers was detected using Indo-1: 1 M Indo-1 in a decalcified solution of 20 mM MOPS, pH 7.0, 100 mM KCl was excited at 355 nm and monitored emission was varied from 350 to 600. In the absence of Ca 2ϩ , a single peak was seen at ϳ475 nm. Binding of Ca 2ϩ to Indo-1 (K D ϭ 0.23 M) caused a decrease in this peak and an increase at ϳ400 nm. The above procedures remove essentially all Ca 2ϩ from CaM (Ͻ0.1 mol Ca 2ϩ /mol CaM), and reduced the background Ca 2ϩ in buffers to Ͻ10 Ϫ7 M.
Stopped Flow Fluorometry-All stopped flow experiments were performed on an Applied Photophysics Ltd. (Leatherhead, UK) model SV.17 MV sequential stopped flow spectrofluorometer with a dead time of 1.7 ms. For all experiments, data from five to six injections were averaged and then fit with either single, double, or triple exponential functions. All solutions were made in a buffer containing 20 mM MOPS, pH 7.5, 100 mM KCl and, where applicable, 1 mM MgCl 2 . The concentrations of reagents given below are those in the syringes, before mixing. For Ca 2ϩ dissociation using quin-2, the excitation wavelength was 334.5 nm, and the detection of fluorescence emission was controlled by a 435 nm cut-on filter (Oriel 51282). 2 M CaM and 20 or 100 M CaCl 2 (along with the CaM target and nucleotide, as indicated in the figure legends) were mixed with 150 M quin-2 (Molecular Probes). For Tyr fluorescence, the excitation wavelength was 276 nm, and emission was monitored with a bandpass emission filter with peak transmittance at 355 nm (Oriel 51662). To measure Ca 2ϩ dissociation, 10 M CaM and 100 M CaCl 2 with 50 M RC3 where indicated was mixed with 5 mM EGTA. To measure rates of Ca 2ϩ association with CaM, 2 M CaM with or without 10 M RC3 was mixed with 20 -100 M CaCl 2 .
Steady State Ca 2ϩ Binding-Titration of CaM and 5,5Ј-Br 2 BAPTA with CaCl 2 was according to the method of Linse et al. (29) as used by us previously (14). Briefly, CaCl 2 was titrated into a cuvette containing 30 M decalcified CaM, 30 M 5,5Ј-Br 2 BAPTA, with or without 60 M RC3, all in 20 mM MOPS, pH 7.5 100 mM KCl. After each addition of CaCl 2 , the absorbance of the solution was measured at a wavelength of 263 nm using a Cary/Varian 100 spectrophotometer. Note that RC3 does not contain Trp or Tyr residues and its presence contributes little to the absorbance at 263 nm.
Free Energy Calculations-The free energy coupling of Ca 2ϩ binding to CaM associated with CaM kinase II was calculated as described in Keller et al. (30). The free energies for the interaction of Ca 2ϩ with CaM (⌬G (Ca) ), the interaction of Ca 2ϩ with the CaM/CaM kinase II complex (⌬G (Ca/CK) ), and the interaction of Ca 2ϩ 4 -CaM with CaM kinase II (⌬G (CK/Ca 4) ) were calculated with the formula ⌬G ϭ ϪRTlnK D , where R is the gas constant (0.586135 kcal/mol) and T is the temperature in K. The free energy for the interaction of apo-CaM with CaM kinase II (⌬G (CK) ) was calculated using Equation 1, (Eq. 1) and the K D for this interaction was calculated using the above relationship between ⌬G and K D . The magnitude of free energy coupling (⌬G CaCK ) was calculated according to Equation 2.
For all these calculations, the free energy of Ca 2ϩ binding used was the average free energy of the four sites, using K D values of 13 M and 2 M for the N-terminal and C-terminal lobes of uncomplexed CaM, respectively (17), and assuming that the rate of association of Ca 2ϩ with CaM was not altered by the binding of CaM kinase II.

RESULTS
Because RC3 can bind to the Ca 2ϩ -free form of CaM, we reasoned that it might stabilize this form of CaM and decrease its affinity for Ca 2ϩ , either by increasing its rate of dissociation and/or decreasing its rate of association. To address these possibilities, two different stopped flow methods were used to assess the rate of Ca 2ϩ dissociation from CaM: quin-2 fluorescence and Tyr fluorescence. Quin-2 is a fluorescent Ca 2ϩ chelator, which exhibits an increase in fluorescence upon binding Ca 2ϩ . Free Ca 2ϩ associates with quin-2 so quickly that the reaction is complete within the 1.7 ms dead time of the fluorometer leaving a fluorescence increase that results from the slower release of Ca 2ϩ from CaM and thus the rate of dissociation of Ca 2ϩ from CaM can be determined. As shown in Fig. 1 and summarized in Table I, the rate of Ca 2ϩ dissociation from CaM is dramatically increased in the presence of RC3. Without RC3, the dissociation of Ca 2ϩ from CaM at 3°C fits to 2 exponentials, with rates of 476 Ϯ 18 s Ϫ1 and 1.9 Ϯ 0.2 s Ϫ1 , which correspond to the rates of Ca 2ϩ dissociation from the N-terminal and C-terminal lobes of CaM respectively (Fig. 1A). The entire amplitude of fluorescence change associated with the slower rate was easily captured in the stopped flow instrument, and we used this amplitude to calibrate the amount of Ca 2ϩ dissociation in the quin-2 experiments: it was considered to be equal to the dissociation of 2 mol of Ca 2ϩ per mole of CaM (11,17,31). Dissociation rates were confirmed by performing stopped-flow experiments while monitoring the intrinsic Tyr fluorescence of CaM, which decreases upon release of Ca 2ϩ . These experiments gave rise to a single rate of 1.69 Ϯ 0.02 s Ϫ1 for the dissociation of Ca 2ϩ . Because the Tyr residues in CaM all reside in the C-terminal lobe, this rate represents the dissociation of Ca 2ϩ from the C-terminal lobe.
In the presence of RC3 at 3°C, the quin-2 fluorescence data ( Fig. 1A) was fit with a triple exponential curve, with rates of Ͼ500 s Ϫ1 , 98.6 Ϯ 9.1 s Ϫ1 , and 9.7 Ϯ 0.7 s Ϫ1 . The Tyr fluorescence data was fit with a double exponential, with rates of 115 Ϯ 1 s Ϫ1 and 12.0 Ϯ 2.7 s Ϫ1 . Because the fast rate measured with quin-2 fluorescence is absent from the Tyr fluorescence data, it can be assumed to correspond to the dissociation of Ca 2ϩ from the N-terminal lobe. The other two rates correspond to Ca 2ϩ release from the C-terminal lobe. By setting the amplitude of the Ca 2ϩ release from the C-terminal lobe of CaM alone (without RC3) to be approximately equal to the release of 2 mol of Ca 2ϩ per mole of CaM (11,17,31), we calculate that in the presence of RC3, 1.64 Ϯ 0.16 mol of Ca 2ϩ dissociate from the C-terminal lobe at the faster rate of about 100, while 0.46 Ϯ 0.06 mol dissociate with the slower rate of about 10 s Ϫ1 .
We also measured Ca 2ϩ dissociation from CaM at room temperature and looked at the effect of RC3 interacting with CaM. Tyr fluorescence stopped-flow experiments using CaM alone gave a C-terminal dissociation rate of 9.54 Ϯ 0.06 s Ϫ1 . In the presence of RC3, the fluorescence data fit with a double exponential as it did at 3°C. The two rates were 557 Ϯ 55 s Ϫ1 and 71.2 Ϯ 2.8 s Ϫ1 . The dissociation of Ca 2ϩ from CaM in the presence of RC3 as measured by Tyr fluorescence is shown in Fig. 1B. Quin-2 experiments performed at room temperature were not able to separate the accelerated release of Ca 2ϩ from the C-terminal lobe in the presence of RC3 from the N-terminal Ca 2ϩ dissociation, and the data fit with double exponential curves, giving rates of Ͼ500 s Ϫ1 and 60.7 Ϯ 5.3 s Ϫ1 . At room temperature, the slower rate represented a greater proportion of the amplitude than at 3°C, when measured with either quin-2 or Tyr fluorescence. The quin-2 data indicate that at room temperature, the slower rate represents the release of ϳ1.4 mol of Ca 2ϩ per mol of CaM.
We next determined whether binding to RC3 causes a change in the overall affinity of CaM for Ca 2ϩ , using a steady state titration method. 5,5Ј-Br 2 BAPTA absorbance decreases when the chelator binds Ca 2ϩ , and this can be used to quantify the amount of free Ca 2ϩ in solution (29). By titrating Ca 2ϩ into a solution containing both CaM and 5,5Ј-Br 2 BAPTA, and measuring the absorbance, we can derive the affinity of Ca 2ϩ for CaM. As shown in Fig. 2A, the presence of RC3 causes a significant left shift of this binding curve. The data could not be fit to a model of competition for Ca 2ϩ between CaM and 5,5Ј-Br 2 BAPTA, which suggests that binding to RC3 decreases the affinity of CaM for Ca 2ϩ so that it is much lower than the affinity of 5,5Ј-Br 2 BAPTA for Ca 2ϩ (K D ϭ 1.6 M).
Because it is possible that the shift in the 5,5Ј-Br 2 BAPTA absorbance curve results from a nonspecific effect of RC3 on 5,5Ј-Br 2 BAPTA absorbance; we decided to quantitate the RC3 effect on the affinity of CaM for Ca 2ϩ by directly determining the association rate constant for Ca 2ϩ binding to CaM. cence was measured in the presence or absence of RC3 for a range of Ca 2ϩ concentrations, and the observed rates of binding (determined by monoexponential fits of averaged data) were plotted against the concentration of Ca 2ϩ . The slope of a linear fit to these data gives the association rate constant, which was 6.0 ϫ 10 6 M Ϫ1 s Ϫ1 for CaM alone, and 6.5 ϫ 10 6 M Ϫ1 s Ϫ1 for CaM plus RC3. Therefore, the rate of association of Ca 2ϩ with the C-terminal lobe of CaM is not altered by CaM interaction with RC3.
Because RC3 does not alter the rate of association of Ca 2ϩ with CaM, the change in affinity of CaM for Ca 2ϩ when bound to RC3 is proportional to the quantified changes in dissociation rates. The fast rate of dissociation of Ca 2ϩ from CaM in the presence of RC3 is more than 60-fold faster than in the absence of RC3; the slower rate is ϳ8-fold faster. Therefore, the affinity of CaM for Ca 2ϩ can be decreased up to 60-fold by CaM interaction with RC3.
We have shown that proteins such as RC3 and PEP-19 (14) that bind to apo-CaM can lead to an increase in the rate of dissociation of Ca 2ϩ from CaM and/or a decrease in the affinity of CaM for Ca 2ϩ . In contrast, Ca 2ϩ -dependent targets of CaM lead to an increase in the affinity of CaM for Ca 2ϩ (15). We demonstrated previously that CaM binding to CaM kinase II significantly increases CaM affinity for Ca 2ϩ (14). However, the additional impact that CaM kinase II autophosphorylation might have on CaM Ca 2ϩ binding affinity was not assessed. In the absence of any interacting protein partner, dissociation of Ca 2ϩ from CaM at room temperature occurs at two rates: one which is too fast to measure practically, and a second rate of 9.1 Ϯ 0.6 s Ϫ1 (Fig. 3). As discussed above, these are the rates of dissociation from the N-terminal and C-terminal lobes, respectively. In the presence of CaM kinase II, the data fit best to a triple exponential curve, as evidenced by the residuals shown in Fig. 3, panels B and C. About 1 mol of Ca 2ϩ dissociates at a rate of 48.9 Ϯ 15.2 s Ϫ1 , roughly 2 mol dissociate with a rate of 7.6 Ϯ 0.9 s Ϫ1 , and less than 1 mol of Ca 2ϩ dissociates with a rate of 0.45 Ϯ 0.06 s Ϫ1 (Fig. 3). These data are summarized in Table II.
When CaM kinase II is autophosphorylated at Thr 286 , it has a much higher affinity for CaM than unphosphorylated CaM kinase II (24). Based on thermodynamic principles, phosphorylated CaM kinase II should therefore stabilize the Ca 2ϩbound form of CaM to an even greater extent than unphosphorylated CaM kinase II. We measured Ca 2ϩ dissociation from CaM bound to CaM kinase II phosphorylated in the presence of ATP␥S, which promotes phosphorylation only at Thr 286 (32). We again measured three rates of dissociation: less than 1 mol of Ca 2ϩ dissociates at a rate of 11.2 Ϯ 2.8 s Ϫ1 , less than 1 mol of Ca 2ϩ dissociates at a rate of 1.24 Ϯ 0.54 s Ϫ1 , and approximately 3 mol of Ca 2ϩ dissociate at 0.38 Ϯ 0.03 s Ϫ1 (Fig. 3A and Table II). Phosphorylation of CaM kinase II in the presence of ATP (as opposed to ATP␥S) can lead to phosphate incorporation at a number of sites in addition to Thr 286 (33,34). Using CaM kinase II phosphorylated in this manner, the Ca 2ϩ dissociation data was less consistent than with the ATP␥S phosphorylated kinase. In about half of the experiments, the rates were similar to those with ATP␥S, while in the other half the rates were considerably slower (data not shown). The cause of this variability is not readily discernible; however, variable autophosphorylation of other sites on the kinase might produce effects on its interaction with CaM that are yet to be characterized.
We next considered the fact that both CaM kinase II and RC3 are known to be present at high concentrations in neurons, and both reside in the same compartments: soma, dendrites, and dendritic spines (35,36). We previously showed that PEP-19 can increase the rates of Ca 2ϩ dissociation from CaM bound to CaM kinase II (14). Therefore, we asked the question: if both RC3 and CaM kinase II are present, how does this tripartite system affect Ca 2ϩ dissociation from CaM? Representative results from quin-2 stopped flow experiments are shown in Fig. 4. Panel A shows that Ca 2ϩ dissociation from CaM in the presence of both RC3 and CaM kinase II occurs at rates intermediate between those in the presence of RC3 or in the presence of CaM kinase II alone. Changing the ratio of RC3 to CaM kinase II did not alter the measured rates of Ca 2ϩ dissociation (data not shown). When we examined the effect of thiophosphorylated CaM kinase II and RC3 on Ca 2ϩ dissociation from CaM, however, different results were obtained, as shown in Fig. 4B. The presence of RC3 did not affect Ca 2ϩ dissociation from CaM when CaM was associated with the thiophosphorylated form of CaM kinase II. The measured dissociation rates were not significantly different from those measured in the presence of thiophosphorylated CaM kinase II alone.

DISCUSSION
In this work, we present data showing that RC3 and CaM kinase II have opposing effects in altering CaM affinity for Ca 2ϩ , and that RC3 can lead to an increase in the observed rates of dissociation of Ca 2ϩ from CaM complexed with CaM kinase II. RC3 causes a significant decrease in CaM affinity for Ca 2ϩ by increasing the rate of dissociation of Ca 2ϩ from the C-terminal lobe of CaM. CaM kinase II causes an increase in the affinity of Ca 2ϩ for CaM by decreasing the rate of dissociation of Ca 2ϩ from both the N-terminal and C-terminal lobes. Quin-2 Ͼ500 -b 9.1 Ϯ 0.6 2 n/a n/a Tyr n/a n/a 9.54 Ϯ 0.06 1 n/a n/a CaM ϩ RC3 22°C Quin b For rates Ͼ400 s Ϫ1 , for example see Fig. 1, panel A, it was not possible to accurately determine the amplitude of the curve, because most of the Ca 2ϩ dissociation was complete within the dead time of the fluorometer.
c Not applicable.
A Kinetic Scheme for Dissociation of Ca 2ϩ and Target from CaM-The dissociation of Ca 2ϩ and Ca 2ϩ -dependent target peptides from CaM induced by chelating Ca 2ϩ has been studied in detail for peptides derived from the CaM binding domain of skeletal muscle myosin light chain kinase (37,38). A major conclusion from this work was that the observed dissociation rates cannot be directly assigned to any kinetic step in the dissociation pathway. This is because several competing dissociation pathways exist in the Ca 2ϩ -CaM-target system, as shown in Fig. 5. This scheme assumes that the two Ca 2ϩ ions in each lobe bind cooperatively, and so the dissociation of two Ca 2ϩ ions from one lobe can be considered as a single step. The first step in the dissociation can be either Ca 2ϩ dissociation from the N-terminal lobe (Fig. 5, step 1), Ca 2ϩ dissociation from the C-terminal lobe (Fig. 5, step 2), or target dissociation from Ca 2ϩ -CaM (Fig. 5, step 3). The target protein can reassociate with (Ca 2ϩ ) 4 -CaM or the (Ca 2ϩ ) 2 -CaM intermediates, altering the overall observed rates of Ca 2ϩ dissociation, depending on the relative rates of Ca 2ϩ dissociation and target dissociation and reassociation. The dissociation of the Ca 2ϩ -CaM-target complex is therefore under kinetic control, in which the order of dissociation steps and the overall speed of the dissociation depend on the relative kinetics of the individual steps in the pathways. This idea of kinetic control of the dissociation process applies to the dissociation of Ca 2ϩ from CaM bound to both Ca 2ϩ -dependent targets like CaM kinase II, and those like RC3, which interact with both Ca 2ϩ -bound and Ca 2ϩ -free CaM.
Dissociation of Ca 2ϩ from CaM Bound to RC3-Previous work from our laboratories has shown that PEP-19, which contains an IQ-type CaM binding domain homologous to that in RC3, also increases the rate of dissociation of Ca 2ϩ from the C-terminal lobe of CaM (14). However, while PEP-19 causes an increase in the rate of Ca 2ϩ association with CaM, RC3 does not. Because of this difference, PEP-19 does not alter the overall affinity of Ca 2ϩ for CaM, while RC3 causes a significant decrease in CaM affinity for Ca 2ϩ . When CaM is bound to RC3 Quin-2 was excited at 334.5 nm and emission was monitored through a 435 nm cut-on filter. The representative traces shown are averages of five runs each. The data were converted to moles of Ca 2ϩ released/mole of CaM by assuming that the amplitude of Ca 2ϩ release from CaM alone was equal to the release of 2 mol of Ca 2ϩ . The data are shown as circles, and the lines represent single (for CaM alone) or triple (with CaM kinase II) exponential fits of the data. B, residuals from a double exponential fit of quin-2 data for the dissociation of Ca 2ϩ from CaM bound to CaM kinase II. The peak in the residuals at ϳ0.05 s followed by a trough at ϳ0.2 s was seen in each data set and indicates that a double exponential fit is inadequate to describe the data. C, residuals from a triple exponential fit of quin-2 data for the dissociation of Ca 2ϩ from CaM bound to CaM kinase II. at 3°C, we measure two rates of dissociation of Ca 2ϩ from the C-terminal lobe of CaM: one of about 100 s Ϫ1 , with an amplitude of 1.6 mol of Ca 2ϩ released per mole of CaM, and a second slower rate of about 10 s Ϫ1 , with an amplitude of 0.4 mol of Ca 2ϩ released per mole of CaM. We could not resolve whether RC3 led to any change in the dissociation of Ca 2ϩ from the N-terminal lobe, due to the extremely fast dissociation rates. Our previous NMR studies of Ca 2ϩ /CaM bound to PEP-19 showed that PEP-19 interacts with residues on the C-terminal lobe of CaM that border the hydrophobic pocket (14). Others showed that a peptide derived from the CaM binding domain of RC3 interacted with similar sites in the C-terminal lobe of apo-CaM, as well as a small region in the N-terminal domain (39). Therefore, RC3 may interact with CaM in a similar manner as PEP-19, with the major interactions occurring with the C-terminal lobe of CaM. This would be consistent with the observed effects on Ca 2ϩ dissociation occurring predominantly in the C-terminal lobe.
Because the dissociation of Ca 2ϩ from the N-terminal lobe of the Ca 2ϩ -CaM-RC3 complex is so fast, it is likely that the first step in the dissociation pathway is the release of the N-terminal Ca 2ϩ ions from this complex (Fig. 5, step 1). The remaining Ca 2ϩ dissociation occurs therefore through dissociation of Ca 2ϩ from the C-terminal lobe of RC3-bound CaM (step 7) or through dissociation of RC3 followed by dissociation of Ca 2ϩ from the C-terminal lobe of uncomplexed CaM (steps 5 and 6). Based on our quin-2 data, we propose that the dissociation rate in step 7 is very fast (ϳ500 s Ϫ1 at room temperature), and this step is responsible for the observed fast rate of Ca 2ϩ dissociation from the C-terminal lobe. The observed slow rate of Ca 2ϩ dissociation is not due simply to dissociation of RC3 followed by dissociation of Ca 2ϩ from the uncomplexed C-terminal lobe, because that rate would be at most 9 s Ϫ1 , which is much slower than we observe. Therefore, the observed slow rate is likely also in part due to dissociation and reassociation of RC3 with CaM before dissociation of Ca 2ϩ .
At higher temperatures, the relative amplitude of the slower observed rate of dissociation from the C-terminal lobe of CaM complexed with RC3 increases, and the amplitude of the faster rate decreases. This indicates that at higher temperatures, the direct dissociation pathway through steps 7 3 8 decreases in prominence, which implies that temperature does not affect all rates equally. The dissociation rate of RC3 from (Ca 2ϩ ) 2 -CaM and/or the dissociation of Ca 2ϩ from uncomplexed CaM increases more than rate of reassociation of RC3 with (Ca 2ϩ ) 2 -

TABLE II
Effect of CaM kinase II on the rates of dissociation of Ca 2ϩ from CaM Ca 2ϩ dissociation from CaM or from CaM bound to CaM kinase II was measured at 22°C using quin-2 stopped-flow fluorometry as described under "Experimental Procedures" and as shown in Fig. 3. The data were fit with single or triple exponential equations using Sigma Plot. Values shown represent mean Ϯ S.D. of the fits to at least four curves. Rate   CaM. The differential effect of temperature on these rates could also be due to the fact that RC3 largely exists as an unstructured molecule (40), i.e. sampling a large variety of conformations, and the kinetics of interaction of RC3 with CaM may be different for the different conformers. Lower temperatures may preferentially stabilize certain conformers, leading to differences in the effects of temperature on the kinetic steps described in Fig. 5.
Ca 2ϩ Dissociation from CaM Bound to CaM Kinase II-Our studies on the effect of the interaction of CaM kinase II with CaM are consistent with the idea that Ca 2ϩ -dependent targets of CaM stabilize the Ca 2ϩ -bound form of CaM (30). Dissociation of Ca 2ϩ from CaM when bound to CaM kinase II was slower than from CaM in the absence of target protein. We measured three dissociation rates: for each mole of CaM, 1 mol of Ca 2ϩ dissociates at a rate of ϳ50 s Ϫ1 , and about 2 mol of Ca 2ϩ dissociate at 8 s Ϫ1 , and less than 1 mol of Ca 2ϩ dissociates at a rate of 0.4 s Ϫ1 . If these data are considered in terms of the dissociation scheme presented in Fig. 5, the rates of dissociation of CaM kinase II from (Ca 2ϩ ) 4 -CaM (step 3) must be compared with the fast rate of dissociation of Ca 2ϩ from the full complex (step 1 or 2) in order to see whether dissociation of the kinase is an important pathway. The first step in the dissociation pathway cannot logically be slower than the fastest observed rate. The kinase dissociation in step 3 occurs at a rate of 1.6 s Ϫ1 , 2 which is much slower than the observed fast dissociation rate of ϳ50 s Ϫ1 , leading to the conclusion that the pathway through step 3 will not be significant in the overall dissociation. Therefore, most of the release occurs through pathways starting with dissociation of Ca 2ϩ from CaM complexed with CaM kinase II (steps 1 or 2). A reasonable assumption is that the dissociation of Ca 2ϩ from the N-terminal lobe is faster than dissociation from the C-terminal lobe, causing step 1 to predominate (38). Dissociation therefore occurs through steps 1, 7, and 8 (i.e. Ca 2ϩ dissociates from each lobe, then CaM kinase II dissociates from Ca 2ϩ -free CaM) or through steps 1, 5, and 6, which includes dissociation of CaM kinase II from the (Ca 2ϩ ) 2 -CaM intermediate, followed by dissociation of Ca 2ϩ from the C-terminal lobe of uncomplexed CaM at a rate of 9 s Ϫ1 . Our measurements indicate that in total about 2 mol of Ca 2ϩ dissociate at ϳ8 s Ϫ1 , so dissociation through steps 5 and 6 must be an important pathway. The observed slow rate of dissociation can be assigned to dissociation through step 7, perhaps convolved with rebinding of CaM kinase II to (Ca 2ϩ ) 2 -CaM in step 5.
Dissociation of Ca 2ϩ from CaM Bound to Thr 286 -phosphorylated CaM Kinase II-For CaM bound to Thr 286 -phospho-CaM kinase II, less than 1 mol of Ca 2ϩ dissociates at 11.2 s Ϫ1 , less than 1 mol dissociates at 1.24 s Ϫ1 , and about 2.5 mol of Ca 2ϩ dissociate at the slower rate of 0.4 s Ϫ1 . Dissociation of phosphorylated CaM kinase II from (Ca 2ϩ ) 4 -CaM is known to be extremely slow, on the order of 10 Ϫ3 to 10 Ϫ5 s Ϫ1 (24,41), therefore dissociation of this complex must occur through pathways starting with steps 1 or 2. Again, we will assume that the N-terminal Ca 2ϩ ions have a faster dissociation rate, however, the entire description that follows could apply with the Cterminal dissociation occurring first. Since the fastest observed rate must be because of the first step in the pathway, one explanation of our data is that there is altered cooperativity in the interaction of Ca 2ϩ to CaM in the presence of phospho-CaM kinase II. The first step is the dissociation of one Ca 2ϩ ion at a rate of about 11 s Ϫ1 (step 1, but with only one Ca 2ϩ ), leading to the formation of a Ca 2ϩ 3 -CaM-CaM kinase II intermediate. Further dissociation occurs by release of the 3 remaining Ca 2ϩ ions at rates of about 1 s Ϫ1 and 0.4 s Ϫ1 (via step 7 and/or steps 5 and 6) followed by phospho-CaM kinase II dissociation from Ca 2ϩ -free CaM.
Another issue that must be considered is that in our experiments it is difficult to assess whether all subunits of CaM kinase II are in the phosphorylated state. If not, the faster rates of dissociation that we see might be because of dissociation of Ca 2ϩ from unphosphorylated subunits. If true, this would imply that all dissociation of Ca 2ϩ from CaM bound to phospho-CaM kinase II occurs at a rate of about 0.4 s Ϫ1 .
Thermodynamic Coupling of Ca 2ϩ and Target Binding to CaM-The effect of binding to target protein on CaM affinity for Ca 2ϩ has been investigated previously with respect to the free energy coupling of Ca 2ϩ binding to CaM. The binding of CaM to target reduces the free energy for the interaction of Ca 2ϩ and CaM. This free energy coupling was first demonstrated by Keller et al. (30) for the binding of CaM to troponin I (TnI). They calculated that the free energy coupling for four Ca 2ϩ ions and TnI binding to CaM was Ϫ5 kCal/mol, which indicated that the affinity of TnI for CaM was increased 4,500fold by the binding of Ca 2ϩ to CaM. Similarly, Olwin et al. (42) calculated the free energy coupling for the binding to CaM of skeletal muscle myosin light chain kinase (skMLCK) and concluded that the difference in free energy for Ca 2ϩ binding to CaM was Ϫ8.44 kCal/mol. They determined from this that the affinity of skMLCK for apo-CaM was about 28 mM. Persechini et al. (43,44) reinvestigated the free energy coupling in CaM/ skMLCK complex and determined that the change in Ca 2ϩ affinity was due mostly to changes in dissociation rate, with little effect on association rates of Ca 2ϩ with CaM. If we assume that there are negligible changes in the rate of association of Ca 2ϩ with CaM in the presence of CaM kinase II, the calculated average free energy coupling is Ϫ7.12 kCal/mol for unphosphorylated CaM kinase II and Ϫ11.22 kCal/mol for phosphorylated CaM kinase II. This leads to the conclusion that the affinity of Ca 2ϩ -free CaM for unphosphorylated CaM kinase II is about 11 mM, while for phosphorylated CaM kinase II it is 0.41 mM. It should be noted that these calculations used the observed dissociation rates, which represent the upper limit for the actual dissociation rates for each step. If the actual rates of dissociation of Ca 2ϩ from CaM complexed to CaM kinase II are slower, the free energy coupling is even greater and the affinity of CaM kinase II for apo-CaM is even weaker.
Effect of RC3 on Dissociation of the Ca 2ϩ -CaM-CaM Kinase II Complex-The interaction of RC3 with CaM while it is bound to CaM kinase II is consistent with our experiments on PEP-19, which similarly increases the rate of dissociation of Ca 2ϩ from CaM complexed with CaM kinase II (14). Depending on the rates of each dissociation step, it may be possible to explain this result with a kinetic argument, by suggesting that RC3 and PEP-19 act as a shunt in the dissociation pathway. The dissociation of the complex of Ca 2ϩ -CaM with CaM kinase II has a large component that occurs through step 5; dissociation of the kinase from (Ca 2ϩ ) 2 -CaM. What then follows is dissociation of Ca 2ϩ from uncomplexed CaM or reassociation of (Ca 2ϩ ) 2 -CaM with CaM kinase II. In the presence of RC3, what happens instead is that (Ca 2ϩ ) 2 -CaM binds to RC3, which induces a fast dissociation of Ca 2ϩ from the C-terminal lobe of CaM. Ca 2ϩ dissociation occurs before CaM dissociates from RC3, so it does not have the chance to reassociate with CaM kinase II. The result is an overall increase in the rates of dissociation of Ca 2ϩ . This interpretation of the results is likely an oversimplification, because the presence of RC3 appears to accelerate the dissociation of Ca 2ϩ from all four binding sites on CaM, but it illustrates the principle that RC3 has can alter the rates of dissociation of the Ca 2ϩ -CaM-CaM kinase II complex through a kinetic competition for CaM.
RC3 has no effect on the observed rates of dissociation of Ca 2ϩ from CaM complexed with phosphorylated CaM kinase II. The dissociation of this complex occurs mainly through dissociation of Ca 2ϩ from CaM complexed to CaM kinase II because the dissociation of the phosphorylated kinase from Ca 2ϩ -CaM is so slow as to be negligible. Therefore, RC3 cannot bind to Ca 2ϩ -CaM and has no effect on the rates of dissociation.
A second possible explanation of this data is that RC3 (or PEP-19) and CaM kinase II can bind to CaM at the same time. The N-terminal lobe of CaM could remain bound to CaM kinase II while the C-terminal lobe dissociates from CaM kinase II and is then available to bind to RC3 or PEP-19. Phosphorylated CaM kinase II makes additional contacts with CaM, which may preclude the binding of RC3 (45,46). This may occur through phospho-CaM kinase II interacting competitively with sites on CaM that are required for interaction with RC3, or by altering the conformation of CaM so that RC3 no longer has access to its binding sites. Alternatively, RC3 may interact with CaM bound to phospho-CaM kinase II, but have no effect on Ca 2ϩ dissociation.
Cellular Effects of Observed Ca 2ϩ Dissociation Rates-The ability of RC3 to accelerate Ca 2ϩ release from CaM bound to CaM kinase II indicates that in the cell, RC3 has a role at the end of a Ca 2ϩ stimulus in increasing the rate at which Ca 2ϩdependent targets release CaM. The impact of RC3 at the end of the Ca 2ϩ stimulus is likely critical for pulsatile Ca 2ϩ stimuli, in which the decrease in binding of CaM to Ca 2ϩ -dependent target sets the startpoint for the response to the next pulse of Ca 2ϩ . An important cellular phenomenon that depends on such pulsatile stimulation is synaptic plasticity, which is induced experimentally by stimulating neurons at frequencies from 1 to 100 Hz. Stimulation at low frequencies has been shown to lead to long-term depression of synaptic strength, while high frequency stimulation leads to long term potentiation of synaptic strength. A role for RC3 at the end of a Ca 2ϩ pulse is supported by the observation that mice lacking RC3 have altered synaptic plasticity (47,48), but the details of the role of RC3 in plasticity remain to be determined. Krucker et al. (48) propose that preferential activation of the Ca 2ϩ -dependent phosphatase calcineurin over activation of CaM kinase II may explain the results they observe. However, it is also clear that the RC3 Ϫ/Ϫ mice have widespread alterations in intracellular signaling (49) and Ca 2ϩ dynamics (50), suggesting that it is not possible to assign the deficits in synaptic plasticity to the effect of RC3 on any one particular Ca 2ϩ -dependent target. Indeed, because CaM is an activator in so many pathways, one would expect that removal of a protein such as RC3, which dynamically regulates the ability of CaM to interact with Ca 2ϩ and target proteins, would lead to broad signaling deficits.
We calculate that RC3 causes an up to 60-fold increase in the K D for the interaction of Ca 2ϩ with the C-terminal lobe of CaM. This increase in K D affects both the activation of CaM by Ca 2ϩ , and, additionally, the levels of free Ca 2ϩ in the neuron. RC3 decreases CaM's affinity for Ca 2ϩ , so it will bind less Ca 2ϩ for a given influx of Ca 2ϩ into the cell. There will therefore be more free Ca 2ϩ . Indeed, Van Dalen et al. (50) found that in neurons from transgenic mice lacking RC3, Ca 2ϩ transients in response to NMDA or the metabotropic glutamate receptor agonist DHPG were significantly smaller than in neurons containing RC3. Our results provide a possible explanation for this finding, by suggesting that in the neurons containing RC3, less of the Ca 2ϩ that enters is able to bind to CaM, and so the free Ca 2ϩ levels are higher; conversely, in neurons without RC3, more of the entering Ca 2ϩ binds to CaM and free Ca 2ϩ levels are lower.
How might the CaM kinase II effect on the dissociation rate of Ca 2ϩ from CaM affect cellular signaling? When CaM is bound to phosphorylated CaM kinase II, the slow rate for Ca 2ϩ dissociation is 0.4 s Ϫ1 . If this corresponds to Ca 2ϩ dissociation from the C-terminal lobe, then the K D for Ca 2ϩ interaction with this lobe of CaM is about 40 nM, assuming that CaM kinase II does not alter the association rates of Ca 2ϩ with CaM. This means that at a resting Ca 2ϩ level of 50 -100 nM, most of the CaM bound to phospho-CaM kinase II will still be in the Ca 2ϩbound form. The presence of RC3 acts to mitigate this effect, however, because once Ca 2ϩ dissociates from CaM, CaM will be bound up by RC3, making it less likely for Ca 2ϩ to reassociate with it.
Further analysis of the data presented in this manuscript is ongoing in our laboratories. We aim to fit these data to the kinetic scheme shown in Fig. 5 and thereby gain an estimate of the kinetics of the individual steps in the dissociation pathways for the Ca 2ϩ -CaM-RC3 and Ca 2ϩ -CaM-CaM kinase II complexes. This computational effort will allow us to better predict the dynamics of the interaction of Ca 2ϩ with CaM in the context of the cell.
The effect of RC3 and CaM kinase II in altering CaM affinity for Ca 2ϩ clearly alters the level of activation of CaM by a given Ca 2ϩ transient. A kinetic explanation of the observed rates is useful for understanding the dissociation pathways important for each protein, and shows that RC3 can have an important role in terminating the activity of Ca 2ϩ -CaM-dependent enzymes when Ca 2ϩ levels fall. These two CaM-binding proteins are examples of two classes of CaM binding interactions: Ca 2ϩdependent, and Ca 2ϩ -independent. There are many other proteins which fall into these two classes, each with slightly different affinity for CaM; each probably alters Ca 2ϩ binding to CaM in a slightly different manner. In this way, the response of CaM to Ca 2ϩ signals is tuned by the binding proteins that are present, so that the appropriate response to a given Ca 2ϩ signal is achieved. This response is also affected by the state of the proteins, since phosphorylation and other modifications of the proteins are known to alter their CaM binding properties, which in turn may alter their effect on Ca 2ϩ dissociation from CaM. This allows the cell to dynamically regulate the response to a Ca 2ϩ signal by regulating the Ca 2ϩ binding properties of CaM. Because Ca 2ϩ signaling is critical to many neuronal functions, it is important for the neuron to be able to carefully modulate its response to Ca 2ϩ signals. Regulation of the Ca 2ϩ binding properties of CaM by interaction with CaM-binding proteins may be a central mechanism for tuning the cellular response to Ca 2ϩ influx.