The Solution Structures of Two Soybean Calmodulin Isoforms Provide a Structural Basis for Their Selective Target Activation Properties*

The intracellular calcium ion is one of the most important secondary messengers in eukaryotic cells. Ca2+ signals are translated into physiological responses by EF-hand calcium-binding proteins such as calmodulin (CaM). Multiple CaM isoforms occur in plant cells, whereas only a single CaM protein is found in animals. Soybean CaM isoform 1 (sCaM1) shares 90% amino acid sequence identity with animal CaM (aCaM), whereas sCaM4 is only 78% identical. These two sCaM isoforms have distinct target-enzyme activation properties and physiological functions. sCaM4 is highly expressed during the self-defense reaction of the plant and activates the enzyme nitric-oxide synthase (NOS), whereas sCaM1 is incapable of activating NOS. The mechanism of selective target activation by plant CaM isoforms is poorly understood. We have determined high resolution NMR solution structures of Ca2+-sCaM1 and -sCaM4. These were compared with previously determined Ca2+-aCaM structures. For the N-lobe of the protein, the solution structures of Ca2+-sCaM1, -sCaM4, and -aCaM all closely resemble each other. However, despite the high sequence identity with aCaM, the C-lobe of Ca2+-sCaM1 has a more open conformation and consequently a larger hydrophobic target-protein binding pocket than Ca2+-aCaM or -sCaM4, the presence of which was further confirmed through biophysical measurements. The single Val-144 → Met substitution in the C-lobe of Ca2+-sCaM1, which restores its ability to activate NOS, alters the structure of the C-lobe to a more closed conformation resembling Ca2+-aCaM and -sCaM4. The relationships between the structural differences in the two Ca2+-sCaM isoforms and their selective target activation properties are discussed.

five distinct CaM genes (sCaM1-5) encoding four distinct isoforms of CaM (18). In the latter organism, sCaM1, -2, and -3 are highly homologous to aCaM with Ͼ90% sequence identity, whereas sCaM4 and -5 are divergent with only ϳ78% sequence identity. These sCaM isoforms show unique target activation profiles that have been categorized into three different groups (19,20). The target enzymes belonging to group 1 include for example, myosin light chain kinases (MLCKs), and they are activated in a similar manner by both sCaM1 and -4. In contrast, group 2 and 3 target enzymes are exclusively activated by sCaM1 or sCaM4, respectively. It seems therefore that each CaM isoform is utilized to control enzymes involved in specific physiological responses in plants. For example, expression of the two most divergent isoforms, sCaM4 and -5, is markedly induced by treatment with a fungal elicitor or following a pathogen attack, and they both can activate the group 3 enzyme nitric-oxide synthase (NOS) (21). Generation of nitric oxide is thought to be one of the early events in plant defense reactions (22,23). Interestingly, even though all sCaM isoforms retain the ability to bind strongly to enzymes belonging to the other group(s), they cannot properly activate them, which raises a question about how they distinguish their own target enzymes for activation rather than acting as a competitive inhibitor (19,20). Site-directed mutagenesis studies of aCaM and sCaM isoforms have identified some of the residues that are responsible for the target selectivity. For example, the substitution of Lys-30 and Gly-40 in sCaM1 by Glu-30 and Asp-40 in sCaM4 are the main reason that sCaM4 fails to activate NAD kinase and smooth muscle MLCK (24,25) (Fig. 1). These two residues are located on a hydrophilic surface that is opposite to the exposed hydrophobic patch; as such these interactions are not seen in studies with typical CaMBD target peptides. On the other hand, the Met-144 3 Val substitution within the C-lobe hydrophobic patch seems to be the main reason that sCaM1 fails to activate NOS (26). Recently the target-binding properties of these sCaM isoforms have been tested using spectroscopic and calorimetric studies (27,28). However, the detailed three-dimensional structures of the sCaM isoforms and the structural features that contribute to their different enzyme activation properties have not been analyzed to date. Here we describe the solution structures of Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4, the soybean CaM isoforms, that are the most closely related and the most divergent from aCaM, respectively. The solution structures of these CaM isoforms have unexpectedly revealed that the C-lobe of Ca 2ϩ -sCaM1 has a more open conformation relative to those of Ca 2ϩ -sCaM4 and Ca 2ϩ -aCaM. Additional biophysical experiments further confirm the presence of an enlarged exposed hydrophobic pocket in Ca 2ϩ -sCaM1. Interestingly, the single Val-144 3 Met substitution was found to reduce the size of this hydrophobic pocket of the C-lobe in Ca 2ϩ -sCaM1, which in turn restores its ability to activate NOS. Taken together our data provide unique insights into the relationship between the structural differences and the selective target activation properties of plant sCaM isoforms.
Fluorescence Spectroscopy-Steady-state ANS fluorescence was recorded on a Varian Cary Eclipse fluorescence spectrophotometer (Varian Inc., Victoria, Australia), with an excitation wavelength of 370 nm, and the fluorescence emission was recorded from 400 to 600 nm. The excitation and emission slit widths were 2.5 nm and 5 nm, respectively, and a volume of 1 ml was used. The samples contained 60 M ANS and 20 M sCaM1, sCaM4, aCaM, or S1V144M in a buffer containing 20 mM HEPES (pH 7.5), 1 mM dithiothreitol, 100 mM KCl, and 1 mM EDTA for the apo-form or 1 mM CaCl 2 for the Ca 2ϩ form. The concentrations of sCaM4 and aCaM were measured using a molar extinction coefficient of 2560 M Ϫ1 cm Ϫ1 at 280 nm. The concentration of sCaM1 and S1V144M was measured by using a molar extinction coefficient of 1450 M Ϫ1 cm Ϫ1 at 280 and by using the Bio-Rad protein assay kit.
Isothermal Titration Calorimetry Experiments-The ANS-CaM interactions were monitored by ITC at 20, 25, and 30°C in 20 mM HEPES (pH 7.5), 100 mM KCl, and either 5 mM CaCl 2 or 3 mM EDTA/EGTA. 5 mM ANS was titrated into a sample cell containing 50 -70 M protein. sCaM1, sCaM4, and S1V144M were incubated overnight at room temperature in the buffer containing an additional 5 mM dithiothreitol and then desalted into the buffer without dithiothreitol using Bio-Rad Econo-Pac 10DG column before the ITC experiments. All data were analyzed using the "one set of sites" model supplied in the MicroCal Origin software to determine association constant (K a ) as described in studies of methionine-modified CaMs (29). 5 NMR Measurements-All NMR experiments for structure determination were performed at 30°C on Bruker Avance 500-or 700-MHz NMR spectrometers equipped with triple resonance inverse Cryoprobes with a single axis z-gradient. Sequential assignments of HN, N, CO, C␣, and C␤ resonances of sCaMs were achieved using two dimensional ( 1 H, 15 (31). { 1 H}-15 N NOE experiments were acquired on 700-and 500-MHz for sCaM1 and sCaM4, respectively, using a recycle delay of 5 s (32), and each experiment was repeated three times. Chemical shifts in all spectra were referenced using 2,2-dimethyl-2-silapentane 5-sulfonate to obtain the 1 H, 15 N, and 13 C chemical shifts as described by Wishart et al. (33). All spectra were processed using the program NMRPipe (34) and analyzed using the NMRView software (35).
Structure Calculations-Initial structures were calculated with CYANA (36) version 2.0 using distance restraints obtained from the automatic NOE assignment protocol, hydrogen bond restraints based on secondary structure from the chemical shift index, and dihedral angle restraints predicted by TALOS (37). Further structural refinements with the addition of RDC restraints were performed by XPLOR-NIH (38). Initial estimations for the axial component of the molecular alignment tensor (Da) and the rhombicity (R) were obtained for the structure calculated by CYANA using PALES (39). Finally, the 20 lowest energy structures from a total of 200 calculated were selected and analyzed. All molecular graphics are created using MOL-MOL (40).

RESULTS
Structure Determination-Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4 both generated well dispersed high quality 1 H, 15 N-HSQC spectra ( Fig. 2). Sequential assignments of the main-chain resonances were obtained from the HNCACB, HN(CO)CACB, HN(CA)CO, and HNCO experiments. Consequently, all amide resonances except for one Pro and a few of the N-terminal residues were successfully assigned in the 1 H, 15 N-HSQC spectrum of both Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4 (Fig. 2). The sidechain resonance assignments were mainly obtained from the C(CO)NH-TOCSY, H(CCO)NH-TOCSY, and HBHA(CO)NH spectra. The (HB)CB(CGCD)HD and (HB)CB(CGCDCE)HE experiments were also used for the assignment of aromatic 1 H resonances. Consequently, 96.4% and 98.5% of the total 1 H resonances are assigned and used for the structure calculations of Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4, respectively. The NOE signals used to generate distance restraints were collected from the three-dimensional 15 N-and 13 C-NOESY-HSQC experiments as well as the two-dimensional-NOESY experiment recorded in D 2 O solution. Initially, a total of 3408 and 2808 NOE signals were manually identified, and 3008 and 2727 signals were successfully assigned by CYANA 2.0 with an automatic NOE assignment protocol, which generated 2019 and 1942 distance restraints for Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4, respectively. The calculated structures together with hydrogen-bond restraints, standard Ca 2ϩ -ligand distances, and the TALOS dihedral angles for both sCaM1 and sCaM4 quite closely resembled the previously determined structures for Ca 2ϩ -aCaM, where the N-and C-lobe of the protein are tethered by a short disordered linker region (residues 77-79), and no NOE restraints were found between the two lobes of the protein. { 1 H}-15 N-NOE experiments for both sCaM1 and sCaM4 show small heteronuclear NOE values (ϳ0.4) in their central linker regions, suggesting that this part of each protein is highly flexible in contrast to the N-and C-lobes themselves, which have NOE values typical for folded globular domains (ϳ0.7) (Fig. 3). Therefore, in the later stages of structure refinement, the structures were separately refined against the N-lobe (residues 1-79) and C-lobe (residues 80 -148/149) of the sCaMs using the program XPLOR-NIH, with the addition of H-N RDC restraints. The RDC values for sCaM1 ranged between Ϫ24 and 21 Hz with a digital resolution of 2.5 Hz. The complete set of RDC values determined for sCaM4 ranged from Ϫ16 to 16 Hz with a digital resolution of 1.7 Hz. All of the restraints used for the final structure calculations of the four structures are summarized in Table  1. The backbone r.m.s.d. values of the final 20 structures of sCaM1 were 0.47 Ϯ 0.06 Å and 0.32 Ϯ 0.06 Å for the N-lobe (residues 5-76) and C-lobe (residues 81-147), respectively. Those for sCaM4 in the corresponding regions are 0.44 Ϯ 0.07 Å and 0.36 Ϯ 0.06 Å, respectively. The experimentally determined RDC values provide an excellent correlation with the best-fit RDC values calculated with the final structures of sCaM1 and sCaM4. The correlation (R)/quality (Q) factors of sCaM1 were 1.00/0.007 and 1.00/0.011 for the N-and C-lobe, respectively. These values for sCaM4 were 1.00/0.026 and 1.00/ 0.018 for the N-and C-lobe, respectively.
Structural Comparisons of sCaM1, sCaM4, and aCaM- Fig.  4 shows the superimposed ribbon diagrams of the lowest energy structures of sCaM1 and sCaM4 for the N-and C-lobe. These structures are as expected, based on sequence homology, very similar to one another. The backbone r.m.s.d. values in the helical regions are 0.88 Å and 1.50 Å for the N-and C-lobe, respectively. Consistent with this, the R/Q factors from the correlations between experimentally measured 1 H-15 N RDC values of Ca 2ϩ -sCaM1 and the best-fit RDC values calculated with sCaM4 structures were 0.99/0.163 and 0.88/0.477 for the Nand C-lobe, respectively. The disagreement in the C-lobes seems to be mainly caused by a difference in the location of the H-helices (Fig. 4b). The structures were also compared with an available high resolution (1 Å) crystal structure (PDB entry 1EXR (14)) and to NMR structures (PDB entries 1J7O and 1J7P (12)) of aCaM. The correlations of the RDC values that were measured for sCaM1 and sCaM4 to these aCaM structures and the resulting backbone r.m.s.d. values are summarized in Table  2. The N-lobes of both sCaM1 and sCaM4 show better agree- ment with the N-lobe solution structure of aCaM determined by NMR than with the high resolution crystal structure, as determined by examining both the RDC correlations and the backbone r.m.s.d values. In Fig. 5a, all N-lobe structures of CaM isoforms are overlaid on their A-and D-helices, so that the differences in the positions of the B-and C-helices are empha-sized. As has been discussed, the crystal structure of Ca 2ϩ -aCaM differs from the Ca 2ϩ -aCaM structures determined by NMR in terms of the arrangement of helices (12), where the solution NMR structure adopts a more "closed" conformation. Similar closed conformations are found in our N-lobe structures of both Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4. The differences in the helix angles among all three NMR structures are small, Ͻ6°for both the Band C-helix ( Fig. 5a and Table 3). The C-lobes however seem to have a larger variability in the arrangements of its helices among the various CaM structures (Fig. 5b). Similar to the N-lobe, the C-lobe of Ca 2ϩ -aCaM adopts a more closed conformation in the NMR structure than what is seen in the crystal structure. In Fig. 5b, when the E-and H-helices are overlaid, the difference in the helix angles between Ca 2ϩ -sCaM4 and Ca 2ϩ -aCaM are relatively small, 4.6°and 11.4°for the F-and G-helix, respectively. However, those differences between Ca 2ϩ -sCaM4 and Ca 2ϩ -sCaM1 are more obvious, 19.5°a nd 16.0°for the F-and G-helix, respectively. Backbone r.m.s.d. values also show that the C-lobe of Ca 2ϩ -sCaM4 is relatively similar to that of the NMR-determined Ca 2ϩ -aCaM structure. On the other hand, the C-lobe of Ca 2ϩ -sCaM1 is more closely related to the crystal structure of Ca 2ϩ -aCaM ( Table 2). The backbone r.m.s.d. values in the latter case are 1.14 and 0.90 Å to the NMR and the crystal structure, respectively. Based on these observations, it seems that the C-lobe of sCaM1 adopts a more open conformation in solution and has a larger hydrophobic target binding pocket than either Ca 2ϩ -sCaM4 or Ca 2ϩ -aCaM.
Characterization of the Exposed Hydrophobic Surfaces on Ca 2ϩ -CaM Isoforms-To confirm that Ca 2ϩ -sCaM1 has a larger hydrophobic pocket than Ca 2ϩ -sCaM4 or Ca 2ϩ -aCaM,    MAY 23, 2008 • VOLUME 283 • NUMBER 21

Solution Structures and Activation Properties of Plant CaMs
we studied the binding of the hydrophobic probe ANS to each protein using steady-state fluorescence spectroscopy and ITC measurements. The fluorescence intensity of ANS is known to be strongly dependent on the polarity of its environment, and in aqueous solution the fluorescence intensity of ANS is extremely small (41)(42)(43). Fig. 6 shows steady-state ANS fluorescence spectra with sCaM1, sCaM4, or aCaM, in the presence of either Ca 2ϩ or EDTA. The apo-forms of sCaM1, sCaM4, and aCaM all give rise to very small enhancements in the fluorescence inten-sity of ANS, suggesting that the hydrophilic surfaces of the apo-CaMs bind poorly to ANS. From this observation, we can conclude that ANS is a suitable probe to provide a semiquantitative estimate of the exposed hydrophobic surface area in the Ca 2ϩ -CaMs. The fluorescence intensity enhancements seen for Ca 2ϩ -sCaM4 and Ca 2ϩ -aCaM are relatively similar at 33-and 26-fold, respectively, and are considerably smaller than the fluorescence intensity enhancement upon addition of Ca 2ϩ -sCaM1, which is ϳ70-fold. ITC experiments showed that ANS binding to Ca 2ϩ -sCaM1, Ca 2ϩ -sCaM4, and Ca 2ϩ -aCaM was relatively weak (K a ϭ 10 3 -10 4 M Ϫ1 ) and exothermic in each case (Fig. 7). However, Ca 2ϩ -sCaM1 gave rise to a much larger heat of binding to ANS than the other two Ca 2ϩ -CaMs (Fig. 7). The overall affinity of Ca 2ϩ -sCaM1 for ANS was also higher than for Ca 2ϩ -sCaM4 or Ca 2ϩ -aCaM at all temperatures tested (Table 4). No interaction between the apo-CaMs and ANS was detected by ITC, consistent with the very weak interactions observed in the fluorescence spectroscopy experiments.
Further characterization of ANS binding was achieved by NMR titration of each Ca 2ϩ -CaM with 0.0, 0.4, 0.8, 1.2, 1.6, 2.0, and 2.8 molar equivalent of ANS. NMR CSP studies are well suited to provide detailed information on protein-ligand binding, particularly for relatively weak interactions (44). Fig. 8 shows the CSPs induced in the backbone amide resonances of Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4 upon binding to ANS. In Ca 2ϩ -sCaM1, many signals that belong to both the N-and C-lobe were perturbed when 0.4 molar equivalents of ANS were added (Fig. 8a). However, with Ca 2ϩ -sCaM4, a characteristic sequential ANS binding was observed, where the first molar equivalent of ANS perturbs only residues in the N-lobe of sCaM4, and the second ANS equivalent perturbs mostly C-lobe residues. Similar sequential ANS binding was also observed with Ca 2ϩ -aCaM (Fig. 8c). From these fluorescence, ITC, and NMR data, it is apparent that the C-lobe of Ca 2ϩ -sCaM1 binds ANS more   strongly than the C-lobes of Ca 2ϩ -sCaM4 or Ca 2ϩ -aCaM, which is consistent with the presence of a larger hydrophobic pocket in the solution structure of the C-lobe of Ca 2ϩ -sCaM1. Effect of the Val-144 3 Met Substitution on the Structure of Ca 2ϩ -sCaM1-To further investigate the relationship between the larger hydrophobic pocket that exists in the C-lobe of Ca 2ϩ -sCaM1 and the inability of this protein to activate NOS, we generated the sCaM1 mutant, S1V144M (see Fig. 1). The complete NMR main-chain assignment of Ca 2ϩ -S1V144M was achieved, and the amide CSP caused by the substitution was analyzed (Fig. 9). The impact of the single Val-144 3 Met mutation was unexpectedly large, spanning all four helices in the C-lobe of Ca 2ϩ -sCaM1. The steady-state fluorescence spectrum of ANS showed a remarkably reduced enhancement upon binding to Ca 2ϩ -S1V144M compared with wild-type Ca 2ϩ -sCaM1 (Fig. 6). A similar trend was also found in the ITC data in which Ca 2ϩ -S1V144M creates a much smaller heat of binding to ANS with a slightly weaker affinity at all three temperatures that we have tested (Fig. 7 and Table 4). Consistent with these results, the NMR titration of Ca 2ϩ -S1V144M with ANS showed a sequential ANS binding similar to that observed with Ca 2ϩ -sCaM4 and Ca 2ϩ -aCaM (Fig. 8d). We also detected the presence of a significant conformational alteration in Ca 2ϩ -S1V144M using backbone H-N RDC NMR experiments. As expected from the CSP data, the RDC data for the Ca 2ϩ -S1V144M N-lobe still show an excellent correlation with the N-lobe structures of both Ca 2ϩ -aCaM and wild-type Ca 2ϩ -sCaM1. However, the reduced R factor and increased Q factor of the C-lobe of the mutant protein compared with the wildtype Ca 2ϩ -sCaM1, together with the increased R factor and the reduced Q factor compared with Ca 2ϩ -aCaM (Table 5), provide evidence that the C-lobe of Ca 2ϩ -S1V144M more closely resembles the solution structure of Ca 2ϩ -aCaM.
Taken together, the NMR and biophysical data suggest that the single Val-144 3 Met mutation altered the structure of the C-lobe of Ca 2ϩ -sCaM1 to adopt a more closed conformation with a smaller hydrophobic target-binding pocket, similar to the C-lobe conformation of Ca 2ϩ -aCaM and Ca 2ϩ -sCaM4.

DISCUSSION
The solution structures of Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4 were found to be dumbbell-shaped structures with two lobes (N-and C-lobe) connected by a highly flexible linker similar to other CaM isoforms (11). In the N-lobe, no significant differences are found among all the solution structures (Fig. 5a), yet mutations in this region of sCaMs are known to cause changes in enzyme activation (24,25). Because none of the amino acid substitutions change the tertiary structure of the N-lobe of the sCaM isoforms, we can conclude that the alterations of surface charges at the positions of residues 30 and 40 must be directly responsible for the different activation profiles reported for these two sCaM isoforms with NAD kinase and MLCKs (24,25).
Even though the sequence identity between the C-lobes of Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4 (80%) is higher than the sequence identity between the N-lobes of each protein (77%), the former shows a larger difference in the arrangement of the helices. From the overlay of the backbone structure of the C-lobe of Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4, it is clear that the position of the H-helix has changed (Fig. 4b).
Consequently, the C-lobe of Ca 2ϩ -sCaM1 was found to be more opened, having a larger hydrophobic target-binding pocket than that of Ca 2ϩ -sCaM4 (Fig. 5b). On the other hand, the helix arrangement and hydrophobic pocket of Ca 2ϩ -sCaM4 are very similar to Ca 2ϩ -aCaM. This outcome is particularly surprising, because the C-lobes of sCaM1 and aCaM have a higher amino acid sequence identity (87%) than the N-lobes of sCaM4 and aCaM (77%). One often assumes that a higher level of sequence similarity will lead to a closer structural agreement; however, our results show exactly the opposite for these three highly related proteins. The presence of the larger hydrophobic pocket in the C-lobe of Ca 2ϩ -sCaM1 was supported by our ANS binding experi-ments, where both the fluorescence intensity enhancement and the affinity for ANS were higher with Ca 2ϩ -sCaM1 than with Ca 2ϩ -sCaM4 or Ca 2ϩ -aCaM (Figs. 6 and 7 and Table 4). Our NMR titration studies revealed that Ca 2ϩ -sCaM4 and Ca 2ϩ -aCaM bound ANS in a sequential manner, whereas both the Nand C-lobe of Ca 2ϩ -sCaM1 bound ANS simultaneously (Fig. 8). Taking this into account, we propose that the more open conformation of the C-lobe of Ca 2ϩ -sCaM1 may allow ANS to fit much deeper into its hydrophobic pocket, thereby leading to higher affinity binding.
Because mutation of Val-144 in sCaM1 to Met is known to restore the ability of this protein to activate NOS to ϳ60% of the sCaM4 activation level (26), we also studied this mutant form of sCaM1 (S1V144M). Fluorescence and ITC experiments indicate that Ca 2ϩ -S1V144M binds ANS in a manner closely resembling Ca 2ϩ -sCaM4 or Ca 2ϩ -aCaM, which suggests that Ca 2ϩ -S1V144M has a smaller hydrophobic pocket than wildtype Ca 2ϩ -sCaM1. Consistent with this, the backbone H-N RDC analysis of Ca 2ϩ -S1V144M clearly revealed that the single Va1144 3 Met substitution altered the structure of the C-lobe FIGURE 8. CSPs upon ANS binding to Ca 2؉ -sCaM1, Ca 2؉ -sCaM4 Ca 2؉ -aCaM and Ca 2؉ -S1V144M monitored by ( 1 H, 15 N)-HSQC NMR spectra. The HSQC spectra of Ca 2ϩ -sCaM1 (a), Ca 2ϩ -sCaM4 (b), Ca 2ϩ -aCaM (c), and Ca 2ϩ -S1V144M (d) acquired with 0.0 (black), 0.4 (red), and 2.0 (blue) molar equivalents of ANS are overlaid. Some well separated signals that are perturbed when 0.4 molar equivalent of ANS was added are indicated by black arrows, which are labeled with the residue numbers, whereas those for the signals that were perturbed only when more than one molar equivalent of ANS was added are colored in cyan. of Ca 2ϩ -sCaM1 to a conformation similar to that of Ca 2ϩ -aCaM (Table 5). We note that the chemical shift differences (Fig. 9) as well as the changes in the H-N RDC values (data not shown) caused by the single substitution were found over all four helices of the C-lobe. Thus, our results suggest that the smaller hydrophobic pocket of the C-lobe of Ca 2ϩ -S1V144M can form the same hydrophobic contacts to the CaMBD of NOS as Ca 2ϩ -sCaM4 or Ca 2ϩ -aCaM, thereby leading to enzyme activation. We have previously reported that the thermodynamic parameters for the binding of the two sCaM isoforms to a synthetic peptide corresponding to the CaMBD of cerebellar NOS were somewhat different (28). Therefore, from these results, we propose that the wide open conformation of Ca 2ϩ -sCaM1 is likely responsible for its failure to activate NOS.
Even though the enthalpy and entropy of binding are different, the binding constants for the cerebellar NOS peptide are almost the same for Ca 2ϩ -sCaM1 and Ca 2ϩ -sCaM4 (28); therefore, a difference in affinity cannot explain the different activation properties. Here, we propose two possible explanations. First, similar to our observations with ANS, the large hydrophobic pocket of Ca 2ϩ -sCaM1 could allow the anchoring residues of NOS to bind more deeply than with Ca 2ϩ -sCaM4 and Ca 2ϩ -aCaM, which may cause unfavorable conformational changes in the neighboring regions of the NOS enzyme, thereby prohibiting activity. Secondly, to make proper hydrophobic contacts between the wide open conformation of Ca 2ϩ -sCaM1 and the CaMBD of NOS, Ca 2ϩ -sCaM1 may have to adopt a different orientation of its N-and C-lobes in the complex compared with Ca 2ϩ -sCaM4 and Ca 2ϩ -aCaM. This would cause a difference in the charge distribution on the surface of the sCaM⅐CaMBD complex structure, which in turn would cause unfavorable repulsion/interactions for other parts of the NOS enzyme. In the case of Ca 2ϩ -aCaM, it has also been demonstrated that such additional interactions with the target protein through the surface structure of Ca 2ϩ -aCaM are required to remove the Ca 2ϩ -aCaM bound inhibitory region from the catalytic cleft of the protein to give rise to activation of skeletal muscle MLCK (45).
In this work, we have discussed the structural differences of two Ca 2ϩ -bound soybean CaM isoforms and Ca 2ϩ -aCaM, and their relationship to their different target-enzyme activation properties. Consistent with the similarity in their amino acid sequences the two plant sCaM isoforms have very similar tertiary structures and dynamic properties. This agrees with previous studies that have shown that they can bind in a similar manner to most of the target enzymes. However, in solution, a variety of open conformations with a different size of the hydrophobic target-binding pocket in the C-lobe exists among the sCaM isoforms, and these lead to different hydrophobic contacts for the CaMBDs of some target enzymes, including NOS. We note that many CaMBDs use a Trp side chain to anchor to the C-lobe of CaM, whereas NOS does not (10). Furthermore, the resulting charge distribution on the protein surface of the sCaM⅐CaMBD complexes, which would form additional contacts with the target protein, would also be essential. As we have discussed, this is apparently the case for the activation of NAD kinase and MLCKs. Therefore, future structural investigations of CaM complexes should aim to focus on these additional interactions, which would require studies with intact targetproteins/domains rather than short synthetic peptides. Such studies would allow for a better understanding of the role of plant CaM isoforms in fine-tuning the calcium-dependent response to various stimuli.

TABLE 5 Correlations between experimentally measured backbone 1 H-15 N RDC values of S1V144M and best-fit RDC values calculated with aCaM and sCaM1 structures
All values were calculated using the same regions defined in Table 2. Changes in the R and Q values of S1V144M from those calculated with a wild-type sCaM1 are also shown in the table as ⌬R and ⌬Q, respectively.