Effects of myosin light chain kinase and peptides on Ca2+ exchange with the N- and C-terminal Ca2+ binding sites of calmodulin.

Myosin light chain kinase and peptides from the calmodulin (CaM) binding domains of myosin light chain kinase (RS-20, M-13), CaM kinase II, and the myristoylated alanine-rich protein kinase C substrate protein slowed Ca2+ dissociation from CaM's N-terminal sites from 405 ± 75/s to 1.8-2.9/s and from CaM's C-terminal sites from 2.4 ± 0.2/s to 0.1-0.4/s at 10°C. Since Ca2+ dissociates 5-29 times faster from the N-terminal in these CaM·peptide complexes and both lobes are required for activation, Ca2+ dissociation from the N-terminal would control target protein inactivation. Ca2+ binds 70 times faster to the N-terminal (1.6 × 108M−1 s−1) than the C-terminal sites (2.3 × 106M−1 s−1). In a 0.6-ms half-width Ca2+ transient, Ca2+ occupied >70% of the N-terminal but only 20% of the C-terminal sites. RS-20 produced a 9-fold and CaM kinase II a 6.3-fold increase in C-terminal Ca2+ affinity, suggesting that some target proteins may be bound to the C-terminal at resting [Ca2+]. When this is the case, Ca2+ exchange with the faster N-terminal sites may regulate CaM's activation and inactivation of these target proteins during a Ca2+ transient.

Methods-All static fluorescence measurements were performed on a Perkin-Elmer LS5 Spectrofluorometer at 10°C. Free [Ca 2ϩ ] was calculated as described by Robertson and Potter (1984). Kinetic measurements were performed by mixing an equal volume (50 l) of each reagent together in an Applied Photophysics Ltd. (Leatherhead, UK) model SF.17 MV stopped flow instrument. This instrument had a dead time of 1.6 ms and flow rate of 17 l/ms. The samples were excited using a 150-watt Xenon arc source at the specified wavelength. Fluorescence emission was monitored through the specified interference filters. The curve fitting program (by P. J. King, Applied Photophysics Ltd.) uses the non-linear Levenberg-Marquardt algorithm.
The changes in Quin fluorescence as it dissociates Ca 2ϩ from CaM were converted to moles of Ca 2ϩ dissociating from CaM by mixing increasing concentrations of Ca 2ϩ (0, 10, 20, 30, 40, 50, and 60 M) with Quin 2. Quin 2 fluorescence increased linearly as a function of increasing [Ca 2ϩ ], allowing us to directly relate a change in Quin 2 fluorescence to the number of moles of Ca 2ϩ dissociating per mole of CaM. Calibration curves were performed at the end of each experiment using the same Quin 2 solutions and experimental conditions as used in the experiments. The amplitude of the change in Quin 2 fluorescence was extrapolated from an exponential fit of the data in Fig. 1A, inset.
Computer Modeling-The computer simulations were performed using KSIM version 1.1 (N. C. Millar, UCLA School of Medicine, Los Angeles). The K d of EGTA for Ca 2ϩ at 10°C and low ionic strength (pH 7.0) was calculated to be 4.23 ϫ 10 Ϫ7 M from the program of Robertson and Potter (1984). The rate of Ca 2ϩ dissociation from EGTA (k off ) was determined by stopped flow experiments where EGTA (10 M) and Ca 2ϩ (20 M) were mixed with Quin 2 (150 M) in the same buffer (20 mM Hepes, pH 7.0), at 10°C. The Quin 2 fluorescence increased at the k off of EGTA, 0.55 Ϯ 0.03/s. Our calculated k on of 1.3 ϫ 10 6 M Ϫ1 s Ϫ1 (from k on ϭ k off /K d ) was in agreement with the value of 9.2 ϫ 10 5 M Ϫ1 s Ϫ1 determined by Smith et al. (1984) at 16°C in 0.1 M ionic strength (pH 6.8). Modeling of the Ca 2ϩ transient and the Ca 2ϩ occupancy of the Nand C-terminal sites of CaM assumed the following initial concentrations: N-and C-terminal Ca 2ϩ binding sites (4 M), Ca 2ϩ (100 M), and EGTA (1 mM), which represent the reaction conditions in the stopped flow experiments of Fig. 5 after mixing. The kinetic parameters used for Ca 2ϩ exchange with the N-and C-terminal sites of CaM were those determined as described under "Results" and as cited in Table I.

Ca 2ϩ Dissociation from the N-and C-terminal Ca 2ϩ Binding
Sites of CaM- Bayley et al. (1984) have previously used Quin 2 fluorescence to measure the rates of Ca 2ϩ dissociation from CaM. Fig. 1A shows Ca 2ϩ dissociation from CaM's N-and C-terminal Ca 2ϩ binding sites as measured by the increase in Quin 2 fluorescence, which occurs upon Ca 2ϩ binding. Ca 2ϩ dissociated from CaM as a biphasic process with 1.2 Ϯ 0.2 mol of Ca 2ϩ dissociating at 405 Ϯ 75/s (Fig. 1A, inset) and 1.6 Ϯ 0.2 mol of Ca 2ϩ dissociating nearly 170 times more slowly at 2.4 Ϯ 0.2/s. Thus, under the conditions used in our studies, we clearly observe Ca 2ϩ dissociation from both the N-and C-terminal Ca 2ϩ binding sites of CaM.
Effect of MLCK and CaM Binding Peptides on Ca 2ϩ Dissociation from CaM-The effects of MLCK and several CaM binding peptides on Ca 2ϩ dissociation from CaM's N-and C-terminal Ca 2ϩ binding sites were determined using Quin 2 fluorescence. Fig. 1B shows the effect of smooth muscle MLCK, its CaM binding peptide , and the MARCKS peptide on Ca 2ϩ dissociation from CaM. In the presence of protein or peptide, the rapid phase of Ca 2ϩ dissociation (405/s) was eliminated. Instead, biphasic Ca 2ϩ dissociation rates of 2.28 Ϯ 0.20/s and 0.39 Ϯ 0.15/s, 1.8 Ϯ 0.2/s and 0.07 Ϯ 0.01/s, and 2.6 Ϯ 0.3/s and 0.3 Ϯ 0.02/s were observed for MLCK, RS-20, and MARCKS peptide, respectively. M-13 peptide, from skeletal muscle MLCK, was similar to RS-20 with biphasic Ca 2ϩ dissociation rates of 2.6 Ϯ 0.6/s and 0.06 Ϯ 0.02/s. C 2 K peptide was similar to the MARCKS peptide with biphasic Ca 2ϩ dissociation rates of 2.9 Ϯ 0.3/s and 0.29 Ϯ 0.03/s. Thus, MLCK and all of these high affinity peptides abolish the rapid rate of Ca 2ϩ dissociation from the N-terminal of CaM and result in ϳ2 mol of Ca 2ϩ dissociating at 1.8 -2.9/s and ϳ2 mol of Ca 2ϩ dissoci-ating at 0.1-0.4/s.
Effect of Peptide on Ca 2ϩ Dissociation from the CaM Cterminal Ca 2ϩ Binding Sites-CaM's two tyrosine residues  are in the C-terminal half of the molecule and undergo a large fluorescence increase when Ca 2ϩ binds to the C-terminal sites of CaM (Dedman et al., 1977;George et al., 1993). Fig. 2 shows the time course of the EGTA-induced decrease in Ca 2ϩ -CaM tyrosine fluorescence. These data agree with our Quin 2 studies and verify that Ca 2ϩ dissociates from the C-terminal sites of CaM at a rate of 2.1 Ϯ 0.1/s at 10°C. The addition of MARCKS peptide, which has no Tyr or Trp residues, reduces the rate of Ca 2ϩ dissociation from these C-terminal sites to 0.32 Ϯ 0.02/s. CaM41/75 is a mutant CaM in which cysteine residues have been introduced at positions 41 and 75 by site-directed mu-FIG. 1. Effect of MLCK and peptides on the rates of Ca 2ϩ dissociation from the CaM N-and C-terminal Ca 2ϩ binding sites as measured by Quin 2 fluorescence. In panel A, the time course of the increase in Quin 2 fluorescence is shown as Ca 2ϩ dissociates from CaM's N-terminal (Fast) and C-terminal (Slow) Ca 2ϩ binding sites. The inset shows the time course of the Quin 2 fluorescence increase that occurs upon Ca 2ϩ dissociation from CaM's low affinity N-terminal Ca 2ϩ binding sites over 0 -20 ms. CaM (8 M) ϩ Ca 2ϩ (60 M) in 20 mM Hepes buffer, pH 7.0, was rapidly mixed with an equal volume of Quin 2 (150 M) in the same buffer at 10°C. The 0 -2-s trace is an average of four traces fit with a double exponential (variance Ͻ 2.7 ϫ 10 Ϫ4 ). The 0 -20-ms trace is an average of six traces fit with a single exponential (variance Ͻ 1.6 ϫ 10 Ϫ4 ). All kinetic traces were triggered at time zero; the first 1.6 ms of premixing is shown, and all traces were fit after mixing was complete. In panel B, the time course of the increase in Quin 2 fluorescence is shown as Ca 2ϩ dissociates from CaM's N-and Cterminal Ca 2ϩ binding sites in the presence of MLCK, RS 20, or MARCKS peptide. CaM(8 M) ϩ Ca 2ϩ (60 M) and 16 M of RS-20 or MARCKS peptide was rapidly mixed with an equal volume of Quin 2 (150 M) in the same buffer (20 mM Hepes, pH 7.0) at 10°C. CaM (2 M) ϩ Ca 2ϩ (60 M) ϩ MLCK (2 M) was rapidly mixed with an equal volume of Quin 2 (150 M) in the same buffer, and the signal was amplified to match the amplitude of the peptide experiments. Each trace is an average of four to six traces, fit with a double exponential (variance Ͻ 1.1 ϫ 10 Ϫ4 ). Control experiments where Ca 2ϩ (60 M) was mixed with an equal volume of Quin 2 (150 M) were flat lines. Quin 2 emission was monitored at 510 nm, with excitation at 330 nm. tagenesis. When these two cysteines are cross-linked by a disulfide bridge, the N-terminal hydrophobic pocket of CaM cannot be opened in a Ca 2ϩ -dependent manner; CaM is thereby inactivated and exhibits a greatly reduced Ca 2ϩ affinity at its N-terminal Ca 2ϩ binding sites (Grabarek et al., 1991). 2 When CaM41/75 (ϩCa 2ϩ ) is mixed with Quin 2, the rapid phase of Ca 2ϩ dissociation from its N-terminal Ca 2ϩ binding sites (which occurs over 0 -20 ms) is not observed. 2 mol of Ca 2ϩ still dissociate from the C-terminal sites of CaM41/75 and cause an increase in Quin 2 fluorescence at a rate of 2.4 Ϯ 0.2/s (Fig. 2). Thus, CaM41/75 allows us to specifically monitor Ca 2ϩ dissociation from the C-terminal Ca 2ϩ binding sites. The addition of MARCKS peptide to CaM41/75 slowed the rate of Ca 2ϩ dissociation from its C-terminal Ca 2ϩ sites from 2.4/s to 0.3/s (Fig.  2). Similar studies with RS-20, M-13, and MLCK indicated that these peptides/protein reduced the rate of Ca 2ϩ dissociation from the C-terminal sites of CaM41/75 from 2.4/s to 0.07 Ϯ 0.01/s, 0.11 Ϯ 0.02/s, and 0.5 Ϯ 0.02/s, respectively. Similar results were obtained with a tryptic fragment of CaM, CaM 78 -148, which contains only the C-terminal half of the molecule (residues 78 -148). As shown with CaM41/75, Ca 2ϩ dissociated from CaM 78 -148 at a rate of 2.3/s, and addition of the MARCKS peptide slowed Ca 2ϩ dissociation to 0.25/s (data not shown). Thus, these studies with the C-terminal fragment of CaM and with the N-terminal mutant CaM (CaM41/75) allowed us to determine the effect of these peptides on Ca 2ϩ dissociation from the C-terminal of CaM, selectively.
Knowing the effect of MLCK and these peptides on the rates of Ca 2ϩ dissociation from the C-terminal sites of CaM, it follows (from the data in Fig. 1B) that MLCK, RS-20, M-13, MARCKS peptide, and C 2 K peptide slow Ca 2ϩ dissociation from the Nterminal Ca 2ϩ binding sites of CaM from 405/s to 2.3 Ϯ 0.2/s, 1.8 Ϯ 0.2/s, 2.6 Ϯ 0.6/s, 2.6 Ϯ 0.3/s, and 2.9 Ϯ 0.3/s, respectively. The effects of MLCK and each of these peptides on the rates of Ca 2ϩ dissociation from the N-and C-terminal Ca 2ϩ binding sites of CaM could, therefore, be unambiguously assigned as listed in Table I.

Effects of Temperature and KCl on Ca 2ϩ Dissociation from the N-and C-terminal Sites of CaM in the Presence of RS-20-
Since many physiological studies are conducted at higher temperatures and ionic strength, we have examined the effects of RS-20 on Ca 2ϩ dissociation from CaM's N-and C-terminal Ca 2ϩ binding sites at various temperatures in the presence of KCl (90 mM). The addition of KCl to CaM ϩ RS-20, at 10°C, did not alter the rates of Ca 2ϩ dissociation from either the N-or C-terminal sites from those reported in Table I. At 20, 30, and 37°C in the presence of KCl and RS-20, Ca 2ϩ dissociated from the N-and C-terminal sites 2.5, 5.8, and 10 times faster, respectively, than at 10°C. This suggests a Q 10 of ϳ2.3-2.5 for these processes.
Effects of CaM Binding Peptides on Ca 2ϩ Affinity at the CaM C-terminal Ca 2ϩ Sites-Ca 2ϩ affinity at the C-terminal sites was determined by the Ca 2ϩ -dependent increase in tyrosine fluorescence. Fig. 3 shows Ca 2ϩ titrations of CaM in the presence or absence of C 2 K or MARCKS peptide and of CaM41/75 in the presence or absence of RS-20. Ca 2ϩ half-maximally binds to the C-terminal sites of CaM and CaM41/75 at pCa 5.96. In the presence of C 2 K and MARCKS peptide, half-maximal Ca 2ϩ binding occurs at pCa 6.8 and 6.4, respectively. RS-20 shifted the half-maximal binding of Ca 2ϩ to the C-terminal sites of CaM41/75 from pCa 5.96 to 7.0 as monitored by the increase in its Trp fluorescence. These results suggest that CaM and CaM41/75 have the same C-terminal Ca 2ϩ affinities and that RS-20, C 2 K, and the MARCKS peptide produce 9.1, 6.3, and 2.6-fold increases, respectively, in C-terminal Ca 2ϩ affinity. Thus, RS-20, which produces the largest (ϳ9-fold) increase in C-terminal Ca 2ϩ affinity, produces the largest (40-fold) decrease in C-terminal Ca 2ϩ off-rate.
Determination of Ca 2ϩ Affinity at the CaM N-and C-terminal Ca 2ϩ Binding Sites-The hydrophobic polarity probe TNS undergoes a large fluorescence increase when it binds to the Ca 2ϩ -dependent hydrophobic pockets in both the N-and Cterminal halves of CaM (Suko et al., 1985;Johnson et al., 1986).
Since Ca 2ϩ binding to both the N-and C-terminal sites exposes TNS binding sites, the Ca 2ϩ affinity at each class of site can be determined by the Ca 2ϩ dependence of the increase in TNS fluorescence. Ca 2ϩ titrations of TNS in the presence of CaM41/75 or CaM85/112 (where the exposure of the N-or C-terminal hydrophobic pocket has been blocked by a disulfide bond, respectively) were half-maximal at pCa 5.9 and 5.6, respectively (data not shown). Thus, the Ca 2ϩ titrations of CaM41/75 with TNS and of CaM or CaM41/75 tyrosine fluorescence all indicate a C-terminal Ca 2ϩ affinity of ϳ1.3 ϫ 10 Ϫ6 M at 10°C. Ca 2ϩ titrations of CaM85/112 with TNS indicate a Ca 2ϩ affinity of ϳ2.5 ϫ 10 Ϫ6 M at the N-terminal sites. Knowing the rates of Ca 2ϩ dissociation from the N-terminal (405/s) and the C-terminal sites (2.4/s), the Ca 2ϩ association rate (k on ) was calculated from k on ϭ k off /K d . These calculations suggest that Ca 2ϩ binds to the lower affinity N-terminal Ca 2ϩ binding sites at ϳ1.6 ϫ 10 8 M Ϫ1 s Ϫ1 and to the higher affinity Cterminal Ca 2ϩ binding sites at ϳ2.3 ϫ 10 6 M Ϫ1 s Ϫ1 . These calculations suggest that Ca 2ϩ binds to the N-terminal sites ϳ70 times faster than it binds to the C-terminal sites.
Transient Occupancy of the N-terminal Ca 2ϩ Binding Sites of CaM-We have utilized the slow Ca 2ϩ on-rate of EGTA (Smith et al., 1984) to produce a rapid Ca 2ϩ transient and determined if Ca 2ϩ can bind to the N-or C-terminal sites of CaM and induce TNS binding during a brief Ca 2ϩ transient. A computer simulation of the Ca 2ϩ transient that is produced when 100 M Ca 2ϩ is instantaneously mixed with 1 mM EGTA is shown in 2 Z. Grabarek, personal communication.   (Fig. 4, MF2 trace). Ca 2ϩ bound to Mg-Fura-2 during the dead time and then dissociated at ϳ1300/s, as EGTA chelated Ca 2ϩ . Thus, the simulation and Mg-Fura-2 fluorescence precisely define the time course of this Ca 2ϩ transient. Fig. 4 also simulates the time course of occupancy of the Nand C-terminal Ca 2ϩ binding sites of CaM (using the on-and off-rates shown in Table I) during this Ca 2ϩ transient. This simulation suggests that the faster N-terminal sites would be 96% occupied at 0.4 ms and that they would lose this Ca 2ϩ to EGTA at ϳ240/s. The slower C-terminal Ca 2ϩ binding sites would only be 18% occupied, and they would then lose this Ca 2ϩ to EGTA at 2.4/s. Ca 2ϩ binding to both the N-and C-terminal lobes of CaM exposes hydrophobic sites that bind TNS and cause a large increase in TNS fluorescence. Fig. 5 (trace A) shows the time course of the decrease in TNS fluorescence when the Ca 2ϩ ⅐CaM⅐TNS complex is mixed with EGTA. Similar to Suko et al. (1985), we observed a biphasic process in which 54% of the TNS fluorescence decrease occurs at 405 Ϯ 20/s (Fig. 5, trace A (N-EGTA)), and 46% occurs at 2.1 Ϯ 0.1/s (Fig. 5, trace A (C-EGTA)). These rates correspond to the rates of Ca 2ϩ dissociation from the N-and C-terminal sites of CaM, respectively (as measured by Quin 2), and suggest that both hydrophobic pockets close and displace TNS as quickly as Ca 2ϩ dissociates. Thus, these changes in TNS fluorescence provide an accurate way of following Ca 2ϩ dissociation from both the N-and Cterminal Ca 2ϩ binding sites of CaM.
If the N-and/or C-terminal Ca 2ϩ binding sites of CaM are transiently occupied during a rapid Ca 2ϩ transient (as in Fig.  4), then TNS fluorescence would increase with Ca 2ϩ and TNS binding and then decrease at the rates of Ca 2ϩ removal from the N-and C-terminal sites. Fig. 5 (trace B) shows the change in TNS fluorescence when Ca 2ϩ (200 M) is rapidly mixed with CaM (4 M) and TNS (200 M) in the presence of 2 mM EGTA. Ca 2ϩ binds and produces a rapid increase in TNS fluorescence, and then TNS fluorescence decreases as a biphasic process. Most (82%) of this decrease occurs as Ca 2ϩ dissociates from the N-terminal sites (Fig. 5, trace B (N-Transient)) at 321 Ϯ 33/s, and the remaining 18% occurs as Ca 2ϩ dissociates from the C-terminal sites at 2.1 Ϯ 0.1/s (Fig. 5, trace B (C-Transient)). When the amplitudes for the decreases in TNS fluorescence  , which is produced when 2 mM EGTA is rapidly mixed with 200 M Ca 2ϩ , was achieved by using the KSIM program (N. C. Millar) by setting the initial [Ca 2ϩ ] to 100 M at time ϭ 0 and using a measured Ca 2ϩ off-rate from EGTA of 0.55/s and a Ca 2ϩ on-rate to EGTA of 1.3 ϫ 10 6 M Ϫ1 s Ϫ1 at 10°C. The time course of this Ca 2ϩ transient was verified by stopped flow, using the rapid Ca 2ϩ indicator Mg-Fura-2. The trace labeled MF2 (q) shows the change in Mg-Fura-2 fluorescence (excitation, 380 nm; emission, 510 nm; trace inverted for comparison) when 1 M Mg-Fura-2 in the presence of 2 mM EGTA is rapidly mixed with 400 M Ca 2ϩ . This trace is an average of six traces fit with a single exponential with a variance of Ͻ1.7 ϫ 10 4 . The simulation of Ca 2ϩ exchange with the N-(f) and C-terminal (Ⅺ) Ca 2ϩ binding sites of CaM assumes that Ca 2ϩ binds to both halves of CaM as a simple bimolecular process with the Ca 2ϩ on-and off-rates given in Table I for CaM's N-and C-terminal sites. that occur in the transient occupancy experiments (Fig. 5, trace  B) are compared to the amplitudes of the decreases in the EGTA experiments (Fig. 5, trace A), we find that 70% of the N-terminal and 20% of the C-terminal Ca 2ϩ binding sites are occupied during the transient occupancy experiments. These data are consistent with the simulation of Fig. 4, where 96% of the N-terminal sites (with a Ca 2ϩ on-rate of 1.2 ϫ 10 8 M Ϫ1 s Ϫ1 ) and 18% of the C-terminal sites (with a Ca 2ϩ on-rate of 1.2 ϫ 10 6 M Ϫ1 s Ϫ1 ) would be occupied during this Ca 2ϩ transient. Increasing the [Ca 2ϩ ] produced greater occupancy of both the N-and C-terminal hydrophobic pockets by TNS. Thus, Ca 2ϩ binding is the rate-limiting step and not the rate of opening of the hydrophobic pocket or TNS binding. These data confirm the Ca 2ϩ on-rates we calculated in Table I and the simulation using these on-rates (Fig. 4), which shows that the N-terminal Ca 2ϩ binding sites are almost fully occupied during a rapid Ca 2ϩ transient, while the C-terminal sites, with a 70-fold slower Ca 2ϩ on-rate, are not.
The observation that the C-terminal sites were only 20% occupied by Ca 2ϩ during this rapid Ca 2ϩ transient was verified using CaM-tyrosine fluorescence. When CaM (4 M) and Ca 2ϩ (200 M) were mixed with 2 mM EGTA, the C-terminal tyrosine fluorescence decreased at a rate of ϳ2.1/s as Ca 2ϩ dissociated from the C-terminal sites (Fig. 6, CaM ϩ Ca versus EGTA trace). When a rapid Ca 2ϩ transient was produced by mixing Ca 2ϩ (200 M) with CaM (4 M) in the presence of 2 mM EGTA, the tyrosine fluorescence rose to only 23% of the intensity seen with Ca 2ϩ saturated CaM and then decayed back at 2.1/s (Fig.  6, CaM ϩ EGTA versus Ca trace). When the [Ca 2ϩ ] was increased to 400 M, the tyrosine fluorescence increased to 42% of the intensity seen with Ca 2ϩ saturated CaM. This suggests that Ca 2ϩ binding to the C-terminal is the rate-limiting step and not a slower structural change at the tyrosine residues. Thus, both the TNS and the tyrosine data indicate that the N-terminal but not the C-terminal Ca 2ϩ binding sites are maximally occupied under the conditions of this rapid Ca 2ϩ transient. DISCUSSION Our studies show that Ca 2ϩ dissociates from the N-terminal Ca 2ϩ binding sites of CaM at 405/s and from the higher affinity C-terminal sites at 2.4/s at 10°C. These results agree with the work of Bayley et al. (1984) who have shown that Quin 2 chelates Ca 2ϩ from the N-and C-terminal sites of CaM at 293 Ϯ 93/s and 2.1 Ϯ 0.4/s, respectively, at 11°C. Martin et al. (1992) have also shown that the N-terminal Ca 2ϩ binding sites lose Ca 2ϩ rapidly (389 Ϯ 64/s) while the higher affinity Cterminal Ca 2ϩ binding sites lose Ca 2ϩ more slowly (11 Ϯ 2.4/s) at 18°C and low ionic strength.
Peptide and MLCK binding to CaM reduced the rate of Ca 2ϩ dissociation from the N-terminal sites from 405/s to 1.8 -2.9/s (a 140 -225-fold decrease) and from the C-terminal Ca 2ϩ binding sites from 2.4/s to 0.1-0.4/s (a 6 -24-fold decrease). Martin et al. (1985) have shown that the CaM antagonist trifluoroperazine slows Ca 2ϩ dissociation from the CaM Nterminal sites from 310/s to 15/s and from the CaM C-terminal sites from 3.6/s to 0.55/s at 13.4°C. Further, calmidazolium slowed Ca 2ϩ dissociation to 4/s and 0.2/s for the N-and Cterminal sites, respectively (data not shown). Thus, these CaM antagonist drugs, which bind to both the N-and C-terminal hydrophobic pockets, produce similar decreases in Ca 2ϩ offrate as peptide.
The high resolution structure of CaM complexed with RS-20 (Meador et al., 1992), C 2 K peptide (Meador et al., 1993), and M-13 (Ikura et al., 1992) suggest that peptide binding dramatically alters CaM structure. The central helix of CaM unwinds, allowing the hydrophobic pockets on the N-and C-terminal lobes to engulf these peptides. The amphipathic nature of these peptides allows them to shield the N-and C-terminal hydrophobic pockets of CaM from solvent (O'Neil and Degrado, 1990) and to stabilize the Ca 2ϩ bound state of CaM. This results in peptide-induced increases in Ca 2ϩ affinity, which we see expressed as large decreases in the rate of Ca 2ϩ dissociation from both halves of CaM.
While all of the peptides produced similar dramatic decreases in the rate of Ca 2ϩ dissociation from the N-terminal of CaM, both RS-20 and M-13 slowed Ca 2ϩ dissociation from the C-terminal ϳ3-fold more than either C 2 K or MARCKS. Comparison of the high resolution CaM-peptide structures indicates that there are ϳ40% less contacts between CaM-C 2 K compared to CaM-RS-20 (Meador et al., 1993). Furthermore, both RS-20 and M-13 have an aromatic Trp residue in their N-terminal domain, which makes extensive contacts with the C-terminal of CaM (Ikura et al., 1992;Meador et al., 1992). This could explain why RS-20 and M-13 reduce Ca 2ϩ dissociation from the C-terminal to ϳ0.1/s while C 2 K and MARCKS peptide reduce Ca 2ϩ dissociation to only ϳ0.3/s. While every residue of RS-20, M-13, and C 2 K form contacts with CaM, the Trp residue in the N-terminal of RS-20 and M-13 may further stabilize the Ca 2ϩ bound state, increase Ca 2ϩ affinity, and decrease the rate of Ca 2ϩ dissociation from the C-terminal.
Our results show that the C-terminal of CaM has a 2.6-fold greater affinity for Ca 2ϩ than the N-terminal of CaM. This agrees with the results of Minowa and Yagi (1984) who have shown a 3.4-fold higher Ca 2ϩ affinity at the C-terminal of the molecule. Since the N-terminal of CaM has a 170-fold faster Ca 2ϩ off-rate than the C-terminal but only a 2-3-fold lower Ca 2ϩ affinity, it follows that the N-terminal of CaM must have a much faster Ca 2ϩ on-rate than the C-terminal. Utilizing the K d and Ca 2ϩ off-rates for the N-and C-terminal Ca 2ϩ binding sites of CaM, we calculated their Ca 2ϩ on-rates to be 1.6 ϫ 10 8 M Ϫ1 s Ϫ1 and 2.3 ϫ 10 6 M Ϫ1 s Ϫ1 , respectively. Estimates of Ca 2ϩ on-rate from K off /K d could be complicated by the cooperativity that exists between the two Ca 2ϩ binding sites in each lobe of CaM. Therefore, it was necessary to verify the faster Ca 2ϩ on-rate to the N-terminal sites by an additional method. Currently, there is no way to directly measure the Ca 2ϩ on-rate to the N-terminal Ca 2ϩ binding sites in a stopped flow apparatus. Our computer simulations (Fig. 4) were conducted assuming the Ca 2ϩ off-rates determined by our stopped flow studies and the calculated Ca 2ϩ on-rates cited above and in Table I. They suggest that during a rapid Ca 2ϩ transient (half-width 0.6 ms), Ca 2ϩ would occupy 96% of the N-terminal sites and 18% of the C-terminal sites. We have produced this rapid Ca 2ϩ transient in the stopped flow and verified (using CaM and TNS fluorescence) that the N-terminal sites are Ͼ70% saturated by this Ca 2ϩ transient, while the slower Cterminal sites are only 20% occupied. Thus, our calculated Ca 2ϩ on-rates for the N-and C-terminal sites of CaM are verified by modeling (Fig. 4) and by the Ca 2ϩ transient occupancy experiments using CaM-TNS (Fig. 5).
Ca 2ϩ not only binds quickly to the N-terminal Ca 2ϩ sites, but it also rapidly exposes the N-terminal hydrophobic pocket to accommodate TNS and presumably protein binding. Thus, only the N-terminal Ca 2ϩ binding sites on CaM can respond to a very rapid rise and fall in Ca 2ϩ by opening and closing their hydrophobic binding pocket. The Ca 2ϩ transient (0.6 ms halfwidth) produced in our stopped flow experiments is much faster than physiological Ca 2ϩ transients, which exhibit half-widths from milliseconds to minutes. Clearly, during these longer Ca 2ϩ transients, both the N-and C-terminal Ca 2ϩ binding sites would be occupied. The rapid "artificial" Ca 2ϩ transient produced in our stopped flow experiments was designed and used to verify the faster Ca 2ϩ on-rate to the N-terminal sites of CaM relative to the C-terminal sites.
The N-terminal Ca 2ϩ binding sites of CaM have a ϳ70-fold faster Ca 2ϩ on-rate and a ϳ170-fold faster Ca 2ϩ off-rate than the higher affinity C-terminal Ca 2ϩ binding sites. Therefore, the N-terminal sites of CaM resemble the N-terminal sites of TnC, which have a fast Ca 2ϩ on-rate (ϳ2 ϫ 10 8 M Ϫ1 s Ϫ1 ) and off-rate (400/s) (Johnson et al., 1979 compared to the higher affinity C-terminal Ca 2ϩ -Mg 2ϩ sites. Both the N-and C-terminal lobes of CaM must bind Ca 2ϩ and expose their hydrophobic pockets to activate target proteins (Persechini and Kretsinger, 1988). Our Ca 2ϩ titrations (Fig. 3) indicate that RS-20 and C 2 K increase Ca 2ϩ affinity at CaM's C-terminal Ca 2ϩ binding sites so dramatically that these sites may be 50 -80% occupied even at the resting levels of Ca 2ϩ found in smooth muscle cells (ϳpCa 6.8, Cornwell and Lincoln (1989)). Therefore, the higher affinity C-terminal lobe of CaM may be bound to some target proteins at resting Ca 2ϩ levels. While this would not result in enzyme activation, the subsequent rapid binding of Ca 2ϩ to the N-terminal sites of CaM would allow it to bind also, resulting in target protein activation. A similar mechanism exists in skeletal muscle troponin C, where the higher affinity C-terminal Ca 2ϩ -Mg 2ϩ sites stabilize the troponin complex at resting levels of Ca 2ϩ , and the rapid exchange of Ca 2ϩ with the faster N-terminal sites regulates contraction and relaxation .
Ca 2ϩ dissociates from CaM after cellular free [Ca 2ϩ ] falls in a Ca 2ϩ transient. This results in the disruption of most CaMtarget protein complexes and target protein inactivation. Since both halves of CaM are required for activation, the removal of Ca 2ϩ from either the N-or C-terminal Ca 2ϩ binding site would result in enzyme inactivation. Our data (Table I) indicate that in the presence of MLCK and peptides, Ca 2ϩ dissociates from the N-terminal sites of CaM 5-29 times faster (1.8 -2.9/s) than it dissociates from the C-terminal half of the molecule (0.1-0.4/ s). This suggests that as the [Ca 2ϩ ] falls, the N-terminal half of CaM would dissociate first, resulting in enzyme inactivation at a rate slower than 2-3/s (at 10°C and low ionic strength). At 20°C, in the presence of RS-20 and KCl, Ca 2ϩ dissociates from the N-terminal sites of CaM at 4.5/s and from the C-terminal sites of CaM at 0.18/s. If Ca 2ϩ dissociation from the N-terminal of CaM controls the disruption and inactivation of the CaM⅐MLCK complex, then these events must occur slower than 4.5/s at 20°C. Consistent with this, we have previously shown that EGTA disrupts the CaM-skeletal muscle MLCK complex at 2/s  and the CaM-smooth muscle MLCK complex at 3.2/s (Kasturi et al., 1993) at 20°C. Stull et al. (1986) have also demonstrated that EGTA can inactivate CaMactivated MLCK at a rate of ϳ1/s. Thus, Ca 2ϩ dissociation from the N-terminal sites of CaM in the CaM⅐RS-20 and CaM⅐MLCK complexes is fast enough to control disruption of these complexes and the inactivation of MLCK. Ca 2ϩ dissociation from the C-terminal sites at 0.18/s is too slow to be responsible for the disruption of this complex and inactivation of MLCK. Thus, Ca 2ϩ dissociation from the N-terminal sites of CaM, in the CaM⅐MLCK complex, appears to control the rate of enzyme inactivation as the [Ca 2ϩ ] falls. Ca 2ϩ dissociates from the N-terminal sites of CaM hundreds of times more slowly when a target protein is bound (1.8 -2.9/s). This allows a rapid Ca 2ϩ transient to result in the binding and presumable activation of a CaM target protein for over 300 -400 ms. Thus, the fast rate of Ca 2ϩ binding to the N-terminal of CaM allows it to rapidly "sense" the Ca 2ϩ transient and bind target proteins. The dramatic increases in Ca 2ϩ affinity and the reductions in the rate of Ca 2ϩ dissociation observed in the CaM-target protein complex ensures that CaM-activated enzymes can remain active long after a rapid Ca 2ϩ transient has subsided. gift of MARCKS peptide, and Dr. Zenon Grabarek (Boston Biomedical Research Institute) for helpful comments on the manuscript and for the gift of CaM41/75 and CaM85/112.