Calcium-dependent and -independent Binding of Soybean Calmodulin Isoforms to the Calmodulin Binding Domain of Tobacco MAPK Phosphatase-1*

The recent finding of an interaction between calmodulin (CaM) and the tobacco mitogen-activated protein kinase phosphatase-1 (NtMKP1) establishes an important connection between Ca2+ signaling and the MAPK cascade, two of the most important signaling pathways in plant cells. Here we have used different biophysical techniques, including fluorescence and NMR spectroscopy as well as microcalorimetry, to characterize the binding of soybean CaM isoforms, SCaM-1 and -4, to synthetic peptides derived from the CaM binding domain of NtMKP1. We find that the actual CaM binding region is shorter than what had previously been suggested. Moreover, the peptide binds to the SCaM C-terminal domain even in the absence of free Ca2+ with the single Trp residue of the NtMKP1 peptides buried in a solvent-inaccessible hydrophobic region. In the presence of Ca2+, the peptides bind first to the C-terminal lobe of the SCaMs with a nanomolar affinity, and at higher peptide concentrations, a second peptide binds to the N-terminal domain with lower affinity. Thermodynamic analysis demonstrates that the formation of the peptide-bound complex with the Ca2+-loaded SCaMs is driven by favorable binding enthalpy due to a combination of hydrophobic and electrostatic interactions. Experiments with CaM proteolytic fragments showed that the two domains bind the peptide in an independent manner. To our knowledge, this is the first report providing direct evidence for sequential binding of two identical peptides of a target protein to CaM. Discussion of the potential biological role of this interaction motif is also provided.

cellular fluid is in the high micromolar to millimolar range. In response to a variety of stimuli, the cytosolic Ca 2ϩ concentration rapidly increases and then quickly returns to the basal level. During this spike, Ca 2ϩ binds to a variety of Ca 2ϩ -binding proteins, some of which are simply involved in Ca 2ϩ storage, whereas others regulate target enzymes. One of the most important regulatory Ca 2ϩ -binding proteins in the cytosol is CaM 4 (3,4). CaM is a small (148 amino acids), ␣-helical protein containing four helix-loop-helix Ca 2ϩ -binding motifs commonly known as EF-hands, which are equally distributed between the two distinct N-and C-terminal globular domains. These two domains are connected by a linker that is flexible in solution, as revealed by NMR studies (5,6). At the resting Ca 2ϩ level, CaM is in a Ca 2ϩ -free state (apo-CaM) with the hydrophobic region of both domains almost completely buried (this is usually referred as the "closed conformation"). Upon binding Ca 2ϩ , the ␣-helices within both domains change their relative orientation, exposing Met-rich hydrophobic patches (the "open conformation"), which CaM uses to bind to its target enzymes (7). In the most common binding mode, both the N-and Cdomains of CaM are involved, forming a collapsed complex by wrapping around the target peptide (8). However, several different binding modes have been recently reported, revealing that CaM can also bind to targets in unusual compact or extended conformations (9 -12). Furthermore, an increasing number of interactions between apo-CaM and target proteins and peptides have been reported (13,14).
CaM binding domains (CaMBDs) are generally small, contiguous amino acid sequences of ϳ25 amino acids in length, containing a large number of basic and hydrophobic residues and having the propensity to form an amphiphilic ␣-helix (12). An important feature is the presence of two hydrophobic "anchor" residues, such as Trp or Phe, or a bulky aliphatic residue, such as Leu, usually spaced either 9 or 13 residues apart (11,15,16). The two anchor residues "dock" the helical peptide in the hydrophobic clefts of the two globular domains determining their relative orientation and the extent of uncoiling of the central linker (17,18). To simplify the study of the CaM interactions, chemically synthesized peptides comprising the CaMBD are normally used instead of the large intact enzymes (19,20). It has been shown that this system is an excellent mimetic for the interaction of CaM with the whole enzyme (21). These CaM-binding peptides are typically 20 -30 amino acids long, which makes them long enough to bind simultaneously to both domains (15). Truncated peptides with partial CaM-binding domains and missing one of the two anchor residues can sometimes form 2:1 complexes with CaM, with one peptide bound to each of the two lobes of the protein (22). It has also been shown that CaM can bind to different substrates even in the absence of Ca 2ϩ . In this case, the most common CaMBD is the IQ motif (IQXXXRGXXXR), although many exceptions have been found (13,14).
Although mammalian CaM (mCaM) is highly conserved, plant cells contain multiple CaM isoforms with varying degrees of sequence homology to the single mCaM. Several studies have demonstrated that the different isoforms have dissimilar stimulatory effects on some mammalian target enzymes and different expression patterns in various plant tissue types, suggesting that they play unique roles in the many different Ca 2ϩ signaling pathways of plants (23,24).
Another important signal transduction pathway in animals, yeasts, and plants is the mitogen-activated protein kinase cascade (MAPK cascade). MAPKs are implicated in developmental, hormonal, biotic, and abiotic stress signaling (25). The activity of MAPKs is strictly regulated via phosphorylation of the conserved TXY motif, which is accomplished by a corresponding MAPK kinase. After activation, the dephosphorylation and inactivation of MAPKs is performed by MAPK phosphatases. In recent work, Ohashi and co-workers (26) identified a putative MAPK phosphatase in Nicotiana tabacum (NtMKP1) as a novel plant-specific CaM-binding protein. This result provides the first direct evidence for a molecular interaction between Ca 2ϩ -CaM and a component of the MAPK signaling cascade in plants or any other species. The CaM-binding domain was identified as a 52-amino acid sequence between the conserved gelsolin and Serrich domain in the middle of the NtMKP1 protein ( Fig. 1), and Trp 440 and Leu 443 were identified as residues that are indispensable for the interaction with CaM (26).
In the present work, we have studied the interactions of the most conserved and the most divergent CaM isoforms from soybean, SCaM-1 and SCaM-4, respectively, as well as mCaM, with peptides derived from the CaM-binding domain of NtMKP1. The peptides were chosen based on their propensity to form amphiphatic helices, as suggested by helical wheel prediction, and on the position of the important Trp 440 and Leu 443 residues. Interestingly, none of the peptide sequences correspond either to the typical Ca 2ϩ -CaMBD, where the hydrophobic anchors are separated by 9 or 13 residues (15,16), or to the apo-CaMBD IQ motif. Fig. 1 illustrates the five peptides that were tested, two long peptides (23 amino acids) with Trp 440 close to the N and the C terminus, respectively (NtMKP1-A and NtMKP1-B), and three short peptides, one encompassing the overlapping sequence of the two longer peptides (NtMKP1-sD) and the other two corresponding to the unique sequences of the longer peptides (NtMKP1-sC and NtMKP1-sE). The information obtained by various biophysical studies suggests that the CaMBD is bound to the C-domain of CaM even in a Ca 2ϩ -free environment. However, when either SCaM is saturated with Ca 2ϩ , each lobe of the protein is able to bind a single peptide, leading to a 1:2 overall stoichiometry. Interestingly, we found that the N-and C-domains of CaM bind to the NtMKP1 peptides with very different affinity and in an independent manner. To our knowledge, this is the first reported evidence for sequential binding of two identical target peptides to CaM.

Protein Purification and Peptide Synthesis-Mammalian
CaM and unlabeled and [ 13 C ⑀ ]Met-labeled SCaMs were expressed and purified as described previously (27,28). The mCaM proteolytic fragments mTR1C (residues 1-77) and mTR2C (residues 78 -148) were produced according to the procedure described previously by Itakura and Iio (29). The SCaM-1 proteolytic fragments S1N (residues 1-75) and S1C (residues 76 -148 and 78 -148) and the isolated C-domain of SCaM-4 (S4C; residues 77-149 and 79 -149) were produced according to a procedure described by Murase et al. 5 Protein concentrations were determined by using the Bio-Rad protein assay kit or by using the following molar extinction coefficient: The NtMKP1-A, NtMKP1-B, NtMKP1-sC, NtMKP1-sD, and NtMKP1-sE peptides, all acetylated at the N terminus and amidated at the C terminus, were synthesized commercially by Anaspec, Inc. (San Jose, CA). Peptide purity was confirmed to be more than 95% by high performance liquid chromatography and matrix-assisted laser desorption/ionization mass spectrometry. The concentration of NtMKP1-A, -B, and -sD was determined using its predicted molar extinction coefficient ⑀ 280 ϭ 5690 cm Ϫ1 ⅐M Ϫ1 , whereas NtMKP1-sC and -sE had to be weighed and dissolved in an adequate amount of solvent.
Fluorescence Spectroscopy-Fluorescence spectra were acquired at room temperature on a Varian Cary Eclipse spectrofluorimeter, using an excitation bandwidth of 5 nm and an emission bandwidth of 10 nm. The lone tryptophan residue of 5 T. Murase, T. Iio, and H. J. Vogel, manuscript in preparation. NtMKP1 peptides was excited at 295 nm, and the fluorescence emission was measured from 300 to 450 nm. In this way, the tryptophan could be selectively excited with minimal interference from the tyrosine residues of SCaM-1 and -4 (30). All samples were at pH 7.3 and contained 20 mM Tris-HCl, 100 mM KCl, 1 mM dithiothreitol, and 2 mM CaCl 2 or 5 mM EDTA. The concentration of the NtMKP1 peptides was 10 M. For the quenching experiments with acrylamide, the apo-SCaMs concentration was 15 M, and the Ca 2ϩ -SCaMs concentration was 6 M to ensure the complete saturation of the peptide. Samples were titrated with 4 l of a 4 M acrylamide stock solution up to a final concentration of 0.1 M acrylamide. The fluorescence intensity was monitored at 355 nm for free NtMKP1 peptides and at 325 nm for the complexes and was corrected for dilution effects. The Stern-Volmer constants (K sv ) were calculated according to the equation, where I 0 represents the fluorescence intensity in the absence of acrylamide, I is the dilution-corrected intensity observed at each titration point, and [C] is the molar acrylamide concentration. The error has been estimated as the S.D. over three measurements for the NtMKP1-A⅐SCaM-1 complex and has been assumed to be the same for all of the K sv values. Binding constants were determined in 20 mM HEPES, 100 mM KCl, 1 mM dithiothreitol, pH 7.3. The peptide concentration was 0.7 M in the presence of 2 mM CaCl 2 and 4 M in the presence of 5 mM EDTA, and SCaMs aliquots were added from concentrated stock solutions. Each titration point was obtained by averaging the intensity at the maximum wavelength ( max ) of each complex over 20 scans and fitting the binding curves with Caligator software (31) using the two-site model for Ca 2ϩ titrations and one-site model for EDTA titrations. The resulting binding constants are the average of three independent measurements. Peptide concentration and association constant influences the accuracy of the measurements. The errors calculated as S.D. values over the three measurements are Ͻ0.8 logarithmic units for the first site of Ca 2ϩ -SCaMs, Ͻ1 logarithmic unit for the second site of Ca 2ϩ -SCaMs, and Ͻ0.1 logarithmic units for apo-SCaMs.
Isothermal Titration Calorimetry (ITC)-All isothermal titration calorimetry experiments were performed on a Microcal VP-ITC microcalorimeter. The procedure was similar to that described previously by Yamniuk and Vogel (27). Briefly, both samples were dissolved in a buffer (20 mM HEPES, pH ϳ 7.5) containing KCl ranging from 0 to 200 mM and 2 mM CaCl 2 for the experiment in the presence of free Ca 2ϩ or 5 mM EDTA for those without free Ca 2ϩ . The syringe was loaded with 400 -600 M NtMKP1 peptides, and for each titration point, 4 -6 l were injected into the cell containing ϳ20 M CaM solutions at temperatures ranging from 10 to 30°C. For the SCaMs, protein dimerization through disulfide bonding was prevented by incubating each protein for 24 h in the buffer containing 10 mM dithiothreitol and then desalting into ITC buffer without dithiothreitol using an Econo-Pac 10DG column immediately prior to ITC analysis (27). The heat of dilution/ mixing was measured in separate control experiments and subtracted in each case, and all of the data were analyzed using Microcal Origin software. For each titration, the stoichiometry (N), association constant (K a ), and enthalpy change (⌬H) were obtained directly from the data, whereas the Gibbs free energy (⌬G), entropy (⌬S), and heat capacity (⌬Cp) changes were calculated from Equations 2-4, respectively.
NMR-All NMR spectra were acquired at 298 K on a Bruker Avance 500 NMR spectrometer equipped with a triple resonance inverse Cryoprobe with a single axis z-gradient. Samples contained 400 -500 Met-mCaM alone or in the presence of increasing amounts of NtMKP1 peptides in 100 mM KCl, 99.9% D 2 O, pD ϭ 7.3 Ϯ 0.1 (not corrected for isotope effects), and 5 mM CaCl 2 or 5 mM EDTA. For the two-dimensional 1 H 13 C heteronuclear single quantum coherence (HSQC) spectra, quadrature detection in the F 1 dimension was obtained by using echo/anti-echo time-proportional phase incrementation. The sweep width in the 1 H and 13 C dimensions was 6009.6 and 500 Hz, respectively, and the carrier frequencies were 500.13235 and 125.7596 MHz, respectively. The size of each spectrum was a 2048 ϫ 64 real data matrix, with four scans for each experiment.

Fluorescence Spectroscopy Reveals a Different Stoichiometry in the Presence and
Absence of Free Ca 2ϩ -Because CaM does not have a tryptophan residue, the fluorescent properties of Trp 440 from the NtMKP1 peptides can be utilized to directly follow the peptide binding to SCaM-1 or -4. It is well known that the Trp indole ring emission is very sensitive to the different characteristics of its surroundings. When the Trp side chain in the free target peptide is surrounded by water molecules, the fluorescence spectra are characterized by an emission centered near 355 nm. However, if the Trp residue becomes buried in a hydrophobic region when the peptide binds to CaM, the fluorescence emission undergoes a blue shift and typically an increase in the fluorescence intensity (32,33).
Upon adding either SCaM-1 or -4 to the tryptophan-containing peptides (NtMKP1-A, -B, -sD) in the presence of Ca 2ϩ , an increase in fluorescence quantum yield as well as a ϳ30 nm blue shift was observed (Fig. 2, A, C, and E), confirming that Trp 440 of each peptide is involved in binding to both Ca 2ϩ -SCaMs. At the saturation point, all three peptides complexed with SCaM-1 had a max value at a slightly shorter wavelength as well as lower intensity compared with the complexes with SCaM-4. These differences indicate that the Trp aromatic ring has a unique environment in the SCaM-1 or -4 binding pocket, as was observed in previous work with SCaMs and a peptide derived from another CaM-binding plant protein (27). The best fitting of the titration curves (insets in Fig. 2, A, C, and E) has been obtained with a two-site model, providing evidence that the binding is characterized by two different processes with distinct affinity, the first interaction having a very high association constant (K a1 ϭ 10 7 -10 10 ) and the second being signifi-cantly lower (K a1 ϳ 10 5 ) (Table 1). Therefore, the stoichiometry for each interaction in the presence of free Ca 2ϩ is 1 SCaM:2 peptides. Although for the second interaction, there were not significant differences in the affinity between SCaM-1 and -4, for the first interaction, SCaM-1 had higher affinity for all of the three peptides. Interestingly, like the longer peptides, NtMKP1-sD peptide (Fig. 1), which is shorter than the usual CaMBD (12), bound to the SCaMs with nanomolar affinity. Nevertheless, it is noteworthy that, for the first interaction, NtMKP1-B had an affinity to SCaMs higher than NtMKP1-A and -sD. This result implies that its extra residues (Gly 450 -Lys 460 ) can establish additional interactions to SCaMs.
In contrast to the Ca 2ϩ -dependent interactions, steady-state fluorescence emission spectra recorded in the absence of free Ca 2ϩ show that NtMKP1-B, but not NtMKP1-A or -sD, is capable of binding strongly to apo-SCaM-1 or -4 ( Fig. 2, B, D, and F). This result implies that the additional residues are important for the apo-SCaMs binding as they are in the presence of free Ca 2ϩ . Furthermore, the best fitting of the titration curve was obtained with a single binding model, indicating that without Ca 2ϩ , the stoichiometry for the complex is 1:1 (inset in Fig. 2D). The affinity of NtMKP1-B for both SCaM-1 and SCaM-4 is ϳ10 5 M Ϫ1 (Table 1), whereas NtMKP1-A and -sD bind to SCaMs too weakly under these conditions to measure the K a values.
Fluorescence quenching experiments of the apo-and Ca 2ϩ -SCaM-peptide complexes clearly indicate that the Trp residue becomes shielded from the quencher when bound to SCaMs, as shown by the low K sv values (Fig. 3). Furthermore, near-UV CD spectroscopy, which is very sensitive to the local environment of the aromatic side chains, shows that the Trp mobility is restricted when the peptide is bound to apo-and Ca 2ϩ -SCaMs (data not shown). It is noteworthy that the K sv values of NtMKP1-B in complex with apo-SCaMs are significantly lower than the value of the peptide alone and similar to those of the Ca 2ϩ complexes. This result suggests that the Trp residue is buried when bound to SCaMs even in the absence of free Ca 2ϩ .   The N-and C-domains of CaM Have Different Affinity for NtMKP1 Peptides-The thermodynamic parameters for binding of the NtMKP1 peptides to SCaM-1 or -4 were determined using ITC at different salt concentrations and/or at different temperatures (Table 2). Fig. 4A shows that the binding curves of NtMKP1-B with the SCaMs are clearly characterized by a twostep process, and therefore a two-site binding model was used for fitting the data. The thermodynamics of NtMKP1-sD binding to the Ca 2ϩ -SCaMs are similar to those seen with the two longer peptides (Table 2), whereas the peptides NtMKP1-sC and -sE (Fig. 1) did not have a significant affinity for both SCaMs (data not shown). These results validate the hypothesis suggested by the fluorescence data that NtMKP1-sD is the primary CaM binding region of NtMKP1 protein. Also in agreement with the fluorescence results, the stoichiometry is consistent with the formation of a 1:2 complex, and the two binding processes have very distinct K a values. Because K a values larger than 10 8 are outside the sensitivity range of the ITC technique, it is impossible to quantitatively analyze the affinity (⌬G or T⌬S) of the first binding process (34). However, the ⌬H of the high affinity interaction can be accurately determined (34). Although the ⌬H was highly dependent on the temperature, it was always negative, indicating that the binding of both peptides to SCaMs is an exothermic event in the temperature range tested. Interestingly, ⌬H values for the first binding of SCaM-1

TABLE 2 Thermodynamics of Ca 2؉ -SCaM-peptide interactions measured by ITC
All experiments at different temperature were performed in the presence of 100 mM KCl. The experiments at different salt concentration were performed at 30°C.

Isoforms
Conditions are generally more negative with respect to SCaM-4. This as well as the other small differences in the behavior of the two proteins that have been found (e.g. the smaller ⌬G values for the second binding of NtMKP1-A to SCaM-1 compared with SCaM-4) are probably related to minor dissimilarities in the surface of each protein.
The T⌬S values, which have been cal-culated for the interaction with K a Ͻ 10 8 , are also dependent on the temperature being negative at higher temperatures and becoming positive in some cases at 10°C. The negative entropy values are probably due to the increased order of the peptide that assumes a helical conformation when bound to SCaMs as evidenced by far-UV CD experiments (data not shown), whereas positive contributions, which are predominant at lower temperature, indicate that water molecules are released from the complex surface (35). Despite the fact that entropy and enthalpy values are dependent on temperature, the Gibbs free energy values (for K a Ͻ 10 8 ) remained almost constant over the temperature range tested and are similar for SCaM-1 and -4 ( Table 2). This behavior, which is referred to as enthalpy-entropy compensation, has been observed in numerous other protein-protein and proteinpeptide interactions, including Ca 2ϩ -CaMs binding to several other target peptides (27, 36 -39).
To further study the contribution of electrostatic and hydrophobic interactions to the binding, titrations of NtMKP1-A or -B with SCaM-1 or -4 at different salt concentrations were performed ( Table  2). The K a1 and K a2 values showed a strong salt dependence, with higher K a values at lower KCl concentrations. This behavior indicates that electrostatic interactions play an important role in the complex formation in addition to hydrophobic interactions. This result is not surprising, because it is well known that both hydrophobic and electrostatic interactions are required for the binding of CaM to amphipathic target sequences (15).
For the interactions studied at different temperatures, the change in heat capacity (⌬Cp) was calculated from the slope of a plot of ⌬H versus temperature (Table 2). This value is related to the change in nonpolar surface that is exposed to the solvent during the binding process (40). The ⌬Cp values for the first binding process with NtMKP1-B are more negative (⌬Cp ϳ Ϫ1.9 kJ⅐mol Ϫ1 ⅐K Ϫ1 ) than those of NtMKP1-A (⌬Cp ϳ Ϫ1.1 kJ⅐mol Ϫ1 ⅐K Ϫ1 ), which indicates that a larger hydrophobic sur- face is buried upon binding the NtMKP1-B peptide compared with NtMKP1-A. However, all of the ⌬Cp values for the first binding event for both peptides are closer to the values reported for the binding of peptides to one domain of mCaM (⌬Cp ϭ Ϫ1.6 kJ⅐mol Ϫ1 ⅐K Ϫ1 ) than to both domains (⌬Cp ϭ Ϫ3.2 kJ⅐mol Ϫ1 ⅐K Ϫ1 ) (36). This suggests that the first binding step with each peptide involves predominantly one domain of either SCaM. The ⌬Cp for the second binding event of either peptide with Ca 2ϩ -SCaM-4 (⌬Cp ϭ Ϫ0.9 kJ⅐mol Ϫ1 ⅐K Ϫ1 ) were also characteristic of binding to a single lobe of CaM (Table 2). In contrast, we found a nonlinear temperature dependence for the second binding event in the Ca 2ϩ -SCaM-1⅐NtMKP1-A complex. This could indicate that there are temperature-dependent differences in this second binding event, although we note that the ⌬H values for this interaction are also less accurate due to the lower affinity of the complex.
To confirm that the N-and C-domains bind the peptides independently, we sought to study peptide binding to the isolated N-and C-domains of SCaM-1 and -4. Unfortunately, proteolytic generation of the isolated SCaM was unsuccessful for the N-domain of SCaM-4 (S4N) and produced only low amounts of pure S1N, S1C, and S4C. 5 Therefore, we tested only the primary binding site peptide (NtMKP1-sD) with the SCaM fragments. We also verified the binding behavior using mCaM and its proteolytically isolated mTR1C and mTR2C, since the fragments can be obtained in higher yields (29). Intact Ca 2ϩ -mCaM binds NtMKP1-A, -B, and -sD in a manner similar to Ca 2ϩ -SCaMs although with affinities 1 order of magnitude weaker (Tables 2 and 3). Furthermore, the thermodynamic values obtained with the isolated C-and N-domains are consistent with higher and lower affinity binding events that have been detected with the intact proteins (Tables 2 and 3). This result clearly shows that the peptides bind first to the C-domain and then to the N-domain with different affinity and that the two domains can act in an independent manner.
As first shown by fluorescence spectroscopy, ITC titrations demonstrated that NtMKP1-B (Fig. 5A), but not NtMKP1-A and -sD (data not shown), can interact strongly with apo-SCaM-1 or -4 as well as with apo-mCaM (Fig. 5B) in the presence of 100 mM KCl. Like the Ca 2ϩ -CaMs, peptide binding to the apo-SCaMs was highly dependent on the salt concentration, showing a stronger affinity without KCl (Table 4). Furthermore, NtMKP1-A and -sD peptides, which were unable to bind SCaMs in the presence of 100 mM KCl, bound to both apo-SCaM-1 and -4 with a K a of about 10 5 in the absence of salt. This result suggests that ionic interactions are prevalent in each complex, in addition to the hydrophobic contacts. The affinity of the NtMKP1-sD peptide for both SCaMs and to mCaM is significantly lower than that of NtMKP1-B (Table 4), confirming that its sequence is not the complete binding region for apo-SCaMs. NtMKP1-sC and -sE did not have a significant affinity for apo-SCaMs (data not shown). As with Ca 2ϩ -mCaM, we performed titrations with the proteolytic fragments of apo-mCaM (Fig. 5, C and D) as well as with S1C, S1N, and S4C (Table 4). These experiments showed that NtMKP1-B binds almost exclusively to the C-domain of apoproteins, whereas there is essentially no binding to the N-domain. This result is consistent with numerous earlier studies showing that CaMBD peptides usually exhibit a Ca 2ϩ -independent interaction with the C-terminal but not the N-terminal domain of CaM (41)(42)(43)(44)(45).
Enthalpy-entropy compensation was observed with NtMKP1-B binding to the apo-SCaMs, since T⌬S and ⌬H but not ⌬G are affected by the temperature. Because ⌬H is linearly dependent on temperature, ⌬Cp values were calculated ( Table  4). The value for apo-SCaM-1 (⌬Cp ϭ Ϫ1.4 kJ⅐mol Ϫ1 ⅐K Ϫ1 ) is significantly different from that for apo-SCaM-4 (⌬Cp ϭ Ϫ2.6 kJ⅐mol Ϫ1 ⅐K Ϫ1 ), indicating that, despite their similar affinity, the interaction surfaces are possibly quite different, with the apo-SCaM-4⅐NtMKP1-B complex having a more extensive nonpolar binding interface. However, we note that apo-SCaM-4 might weakly self-associate (46), which could influence the observed ⌬Cp values for this complex. a Due to the lack of a reliable optically active residue, there is a large degree of uncertainty in the mTR1C and S1N concentration that affects the binding stoichiometry of these two proteins.
The N-and C-domains of CaM Act Independently-The hydrophobic patches of CaM are rich in methionine residues, and it is well known that they play an essential role in the promiscuous binding behavior of the protein (11,47). Therefore, NMR spectroscopy studies that monitor the chemical shift of the terminal methyl group of the Met side chains are useful to study the complexes of CaM with different target peptides (27,42,48,49). Since SCaM-1, SCaM-4, and mCaM contain con-served Met residues, the comparison of their spectra allows investigating similarities and differences between the CaM complexes with NtMKP1 peptides. Therefore, we studied the complexes of SCaM-1, SCaM-4, and mCaM with the NtMKP1-A, -B, or -sD peptides by heteronuclear 1 H, 13 C-HSQC NMR spectroscopy using [ 13 C ⑀ ]Met-labeled proteins. Because there were no distinct differences between the different CaM isoforms, and since the assignments of all methionine methyls for apo-mCaM and Ca 2ϩ -mCaM are known (50), we have only presented the mCaM spectra. Furthermore, since NtMKP1-sD is the primary binding site for Ca 2ϩ -CaM, and NtMKP1-B has the highest affinity for the C-domain in the presence of free Ca 2ϩ , we have presented only the data obtained with these peptides (Fig. 6). NtMKP1-B is the primary binding site for apo-CaMs, and therefore its spectra are also presented (Fig. 7).
In the presence of ϳ1.0 molar eq of NtMKP1-sD, all of the C-domain Met peaks of Ca 2ϩ -mCaM had almost disappeared (Fig. 6B), whereas a new set of peaks emerged (a slow exchange process), indicating that this domain is saturated by the peptide, whereas the N-domain Met residues were not affected. Upon the addition of a second molar equivalent of peptide (Fig. 6C) all of the N-domain peaks then shifted to their peptide bound conformation (a fast exchange process). This confirms that the peptide binds to the protein with a 2:1 molar stoichiometry and that both CaM domains independently bind to one peptide, with the C-terminal domain having the higher affinity.
Interestingly, the NtMKP1-B peptide had a different behavior from NtMKP1-sD. Even at a 1:1 ratio with mCaM (Fig. 6E), the N-domain Met peaks were shifted from the unbound position, implying the existence of weak interactions of the extra residues (Gly 450 -Lys 460 ) with the N-domain along with the strong interactions with the C-domain. This result is in agreement with the fluorescence data showing that NtMKP1-B has a higher affinity than the other two peptides for CaMs. Furthermore, it explains the larger ⌬Cp values found for the first binding process of NtMKP1-B com- pared with the NtMKP1-A peptide. At an ϳ2.5 peptide/protein ratio (Fig. 6F) all N-domain Met residues had disappeared except Met 76 , which was only slightly shifted. Since the number of peaks present in the spectrum was lower than the number of Met residues in the labeled protein, some of the regions of the complex must undergo intermediate chemical exchange.
NMR spectra recorded in the absence of free Ca 2ϩ showed that in the presence of 1 molar eq of NtMKP1-B, all peaks belonging to Met residues in the C-domain of apo-mCaM had completely shifted to the peptide-bound conformation, consistent with the 1:1 stoichiometry, whereas those of the N-domain are not affected (Fig. 7). The presence of a set of weaker peaks, as shown in the inset, indicates that the C-domain Met residues in the complex are in slow-to-intermediate chemical exchange between two conformations, which explains why these peaks have lower intensity than those belonging to the peptidefree N-domain. Higher concentration of the peptide resulted in the formation of precipitate.

DISCUSSION
The discovery of an interaction between the Ca 2ϩ signaling protein CaM and a plant MAPK phospha-  tase establishes an important connection between two of the major signal transduction pathways in plant cells. SCaM-1 and -4 represent a good system for investigating the binding of NtMKP1 peptides to CaM, because they represent the extremes of the sequence homology within the well studied soybean plant CaM family (24,46,51). Like other CaM target peptides (27,46), we found that in vitro, NtMKP1 peptides bind with similar affinity to both SCaM-1 and -4 and also to mCaM in the presence of free Ca 2ϩ . In contrast to the results of Ohashi and coworkers (26), where no interaction with apo-CaM was observed in electrophoretic mobility shift assays, we observed an interaction between NtMKP1-B and -sD peptides and the three CaM isoforms even in the presence of EDTA. This discrepancy is probably due to the higher sensitivity of the techniques described herein, which can detect weaker interactions. However, we cannot exclude the possibility that the extra residues in the peptides studied by Ohashi can inhibit the binding to apo-CaM.
Our results show that NtMKP1 peptides interact predominantly with the C-domain of the apo-CaMs and form 1:1 complexes, which is similar to other apo-CaM-binding peptides (42,43,46), whereas the N-domain is not involved in the binding. This interaction involves both hydrophobic and electrostatic components and includes burial of Trp 440 of NtMKP1. Although hydrophobic interactions are a well known component of apo-CaM target complexes (52), it is quite unusual to observe burial of a Trp residue, since the hydrophobic patches of the protein should not be solvent-accessible in the closed conformation of the domain. Although our current data cannot reveal the details of how Trp 440 of NtMKP1 is anchored to CaM, the interaction could resemble the CaM complex with peptides derived from small conductance Ca 2ϩ -activated potassium channels, where the Trp is anchored in a hydrophobic cavity of the apo-CaM C-domain (45,53). Interestingly, the different fluorescence features of apo-and holocomplex spectra reflect the differences of the binding pockets where Trp is buried, as revealed by NMR structural studies. The increased apo-CaM affinity for NtMKP1-B with respect to NtMKP1-sD also suggests that additional interactions might occur with the downstream Ile 451 , Phe 455 , Ile 456 residues. As a result, NtMKP1-B provides a more complete apo-CaM binding region of NtMKP1. The reduced binding constant in the presence of salt is probably due to the disruption of the interactions between the peptide's positive amino acid side chains and the negative side chains of apo-CaM. In fact, these types of electrostatic interactions have been shown to be essential for apo-CaM target recognition (52)(53)(54). Furthermore, Koch and co-workers (55) have shown that increasing the number of acidic residues in the target sequence can disrupt binding to apo-CaM. Therefore, the three additional Glu residues (Glu 428 , Glu 430 , Glu 437 ) at the N-terminal region of NtMKP1-A are the likely cause for its weaker binding to the apo-CaMs.
In the presence of free Ca 2ϩ , all of the three NtMKP1 peptides bind to CaMs with the same 2:1 stoichiometry and very similar thermodynamic behavior. These results suggest that the primary CaM binding region is restricted to the 12-amino acid sequence of NtMKP1-sD (Asn 438 -Ser 449 ). The larger affinity of NtMKP1-B for the C-domain of Ca 2ϩ -SCaMs is probably due to additional interactions between the peptide's extra residues with the N-domain of the Ca 2ϩ -SCaM, as shown by NMR. Small differences in the CaMBDs for apo-and holo-CaM have also previously been suggested in a model of mCaM bound to an IQ motif peptide (54) and in the complex of mCaM with the ryanodine receptor (56). Like the apo-CaM complexes, NtMKP1-peptide binding to the Ca 2ϩ -CaMs involves both hydrophobic and electrostatic interactions and includes the burial of Trp 440 . Trp burial as well as electrostatic interactions are common features for Ca 2ϩ -CaM-peptide complexes (57). The relative position of the two key residues (Trp 440 and Leu 443 ) in the NtMKP1-sD sequence does not correspond to a typical CaMBD. A similar 1-5 pattern is found in the CaMBD of plant glutamate decarboxylase (58). The C-terminal peptide of glutamate decarboxylase also binds to Ca 2ϩ -CaM with a 2:1 stoichiometry, with both peptides binding simultaneously (58,59). The NMR structure of the complex showed that each glutamate decarboxylase peptide binds exclusively to one CaM domain. The two peptides form a dimer that forces the CaM domains in a previously unseen relative orientation with a number of interdomain contacts larger than usual (58). Despite the similarities between NtMKP1 and glutamate decarboxylase peptides, the key difference for NtMKP1-sD is that the binding to CaM is sequential, and the two CaM domains are independent. Interestingly, as with NtMKP1-A, the glutamate decarboxylase peptide has a high number of acidic residues. These residues are involved in interactions between the dimer interface. However, since the acidic residues of NtMKP1-A are located out of the actual CaM-binding region of NtMKP1 and do not affect the binding constant values, we suggest that they are unlikely to play a similar role in the interaction with Ca 2ϩ -CaMs.
A novel characteristic of the NtMKP1-CaM interaction is the very different affinity of the peptides for the two lobes of Ca 2ϩ - CaM in vitro. We can hypothesize that in vivo, Ca 2ϩ -CaM induces dimerization of NtMKP1 analogous to the CaM-induced dimerization of the oxygenase domain of human endothelial nitric-oxide synthase (60). However, the relatively low affinity of NtMKP1 for the N-domain suggests that such dimerization would only take place when the local phosphatase concentration is quite high. It seems more likely that NtMKP1 is preassociated with the C-domain of CaM at resting and elevated Ca 2ϩ concentrations, leaving the N-domain free to bind a second target protein or even a distal region of NtMKP1. Other examples of Ca 2ϩ -binding proteins having two domains with different roles include troponin C binding to troponin I (61,62) or centrin binding to the yeast centrosomal protein Kar1p (63). The target protein-binding mechanism observed in centrin is particularly noteworthy with respect to the present study. The C-domain of centrin is bound to a peptide derived from the yeast centrosomal protein Kar1p (K 19 ) either at low or high free Ca 2ϩ concentration. The N-domain does not have any appreciable affinity for K 19 peptide. It has recently been shown that the isolated Ca 2ϩ -loaded N-domain is able to bind to a different peptide (64), confirming that the N-domain of centrin serves as a Ca 2ϩ sensor. Such a mechanism of action could explain the finding that CaM is not able to activate NtMKP1 (65), since its function could be to transmit the Ca 2ϩ signal to a presently as yet unknown third protein. This would allow CaM to act as a Ca 2ϩ -dependent "adaptor" protein, similar to the adaptor proteins that organize large protein complexes in transmembrane protein kinase signaling (66,67). MAPKs are also known as structural adaptors in the formation of transcription complexes (68). Further structural and biological studies are required to more deeply understand the biological importance of these unique CaM-NtMKP1 complexes.