Solution Structure of the Calponin Homology (CH) Domain from the Smoothelin-like 1 Protein

The SMTNL1 protein contains a single type-2 calponin homology (CH) domain at its C terminus that shares sequence identity with the smoothelin family of smooth muscle-specific proteins. In contrast to the smoothelins, SMTNL1 does not associate with F-actin in vitro, and its specific role in smooth muscle remains unclear. In addition, the biological function of the C-terminal CH-domains found in the smoothelin proteins is also poorly understood. In this work, we have therefore determined the solution structure of the CH-domain of mouse SMTNL1 (SMTNL1-CH; residues 346-459). The secondary structure and the overall fold for the C-terminal type-2 CH-domain is very similar to that of other CH-domains. However, two clusters of basic residues form a unique surface structure that is characteristic of SMTNL1-CH. Moreover, the protein has an extended C-terminal α-helix, which contains a calmodulin (CaM)-binding IQ-motif, that is also a distinct feature of the smoothelins. We have characterized the binding of apo-CaM to SMTNL1-CH through its IQ-motif by isothermal titration calorimetry and NMR chemical shift perturbation studies. In addition, we have used the HADDOCK protein-protein docking approach to construct a model for the complex of apo-CaM and SMTNL1-CH. The model revealed a close interaction of SMTNL1-CH with the two Ca2+ binding loop regions of the C-terminal domain of apo-CaM; this mode of apo-CaM binding is distinct from previously reported interactions of apo-CaM with IQ-motifs. Finally, we comment on the putative role of the CH-domain in the biological function of SMTNL1.

Calponin is a key regulator of smooth muscle contraction (reviewed in Refs. 1 and 2), and the calponin homology (CH) 5 domain was identified in the N-terminal portion of this protein as an ϳ110-amino acid region that contributed to its actinbinding properties (3). CH-domains have since been identified in a number of cytoskeletal and signaling proteins (4 -6). The CH-domain has a highly conserved structure that is associated with diverse biological functions. Although the various CHdomains share relatively little amino acid sequence identity, a number of strictly conserved hydrophobic residues give rise to an almost invariant hydrophobic core (see Fig. 1). Thus, all of the CH-domain structures that have been determined to date are very similar. Despite a common overall fold, different CHdomains serve to interface with a wide variety of proteins involved in cytoskeletal dynamics and/or signal transduction. Therefore, the divergence in CH-domain function is thought to result from discrete sequence elements that are exposed on the protein surface. CH-domains have been classified into several families (summarized in Ref. 5). The type-1 and type-2 CHdomains are normally arranged in tandem and are found in many actin-binding proteins, including members of the spectrin, ␣-actinin, dystrophin, and fimbrin protein families. Single CH-domains are found in several proteins, such as calponin and IQGAP, and are usually classified as type-3 CH-domains. The type-2 CH-domain can also exist as an isolated CH-domain, and it is found in a few proteins, including smoothelins, MICALs, and RP/EBs (6).
Calponin and other CH-domain proteins may regulate smooth muscle contractility via the thin filament regulatory system. In a previous report, Borman et al. (7) identified a novel ϳ60 kDa protein that was phosphorylated by cGMP-dependent protein kinase during cGMP-induced Ca 2ϩ densensitization in ileal smooth muscle. This protein was shown to contain a single type-2 CH-domain at its C terminus, which shared sequence similarity with the smoothelin family of smooth muscle-specific proteins (reviewed in Ref. 8). The 459-residue protein, initially called CHASM (calponin homology-associated smooth muscle), is termed SMTNL1 (smoothelin-like 1). However, unlike the smoothelins, it did not associate with actin filaments in vitro, and hence the specific role of SMTNL1 in smooth muscle relaxation remains undefined. Also, the interaction of smoothelin proteins with actin is mediated by additional N-terminal actin-binding domains such that their CHdomain was neither necessary nor sufficient for actin binding (9). Therefore, the biological role of the C-terminal type-2 CHdomain in the smoothelin protein family remains unclear.
Close inspection of their amino acid sequences reveals that the CH-domains of the smoothelins and SMTNL1 contain a putative calmodulin (CaM)-binding IQ-motif sequence (Fig. 1). It has been reported that CaM can bind to many other CHdomain proteins, even though these generally do not posses IQ-motifs. These proteins include calponin (10,11), spectrin (12), dystrophin (13), and filamin A (14), in which CaM modulates the Ca 2ϩ dependence of actin binding (13)(14)(15). Some CHdomain proteins, such as IQGAP, also contain IQ-motifs, though these are located outside of the CH-domain region of the protein (16). In this study, we provide the first insight into the structure of the C-terminal type-2 CH-domain from SMTNL1 (SMTNL1-CH) and discuss its structural characteristics with respect to previously reported CH-domains. We also demonstrate the binding of apo-CaM to SMTNL1-CH via the IQ-motif sequence and calculate a docking model for the SMTNL1-CH⅐apo-CaM complex using the HAD-DOCK protein-protein docking program (17).

EXPERIMENTAL PROCEDURES
Expression and Purification of SMTNL1-CH-The full-length mouse SMTNL1 cDNA was generated from IMAGE clone 3593616 as described previously (7). A fragment of SMTNL1 encoding the CH-domain of SMTNL1 (residues 346 -459) was then amplified by standard PCR techniques and subcloned into the pGEX-6P1 vector (GE Healthcare) using BamHI/ NotI sites. The construct was verified by DNA sequencing.
The GST-SMTNL1-CH fusion protein was produced in Escherichia coli, strain BL21(DE3) in LB medium. Uniformly 15 N-labeled and 15 N, 13 C-labeledGST-SMTNL1-CH proteins were prepared in M9 medium containing 0.5 g/liter 15 NH 4 Cl and 1 g/liter [ 13 C 6 ]glucose (or unlabeled glucose). The fusion proteins were isolated using glutathione-Sepharose 4B resin and cleaved "on-column" by treatment with PreScission Protease (GE Healthcare). The eluted protein contained the cloning artifact "GPLGS" at its N terminus. The SMTNL1-CH construct was concentrated and exchanged into 1 mM sodium phosphate buffer with an Amicon centrifugal filter (Millipore).
Expression and Purification of Calmodulin (CaM)-Chicken CaM was expressed from the pET30b(ϩ) vector in E. coli strain BL21(DE3) grown in LB medium as described previously (18). Uniformly 15 N-labeled CaM was prepared in M9 medium containing 0.5 g/liter 15 NH 4 Cl. CaM proteolytic fragments, the N-terminal domain of CaM (CaM-nt; residues 1-77), and the C-terminal domain of CaM (CaM-ct; residues 78 -148) were produced as described previously (19).
sample was also prepared in 99.99% D 2 O. The samples used for residual dipolar coupling (RDC) measurements also contained 300 mM KCl, 20 mM bis-Tris (pH 6.9), and 10 mg/ml filamentous phage Pf1 (Asla Biotech Ltd.). All samples to monitor the interaction between SMTNL1-CH and apo-CaM also contain an additional 1 mM EDTA.
Isothermal Titration Calorimetry (ITC) Measurements-All ITC experiments were performed on a MicroCal VP-ITC microcalorimeter. Solutions of ϳ0.5 mM CaM, CaM-nt, or CaM-ct with 20 mM HEPES (pH 7.0) and 1 mM EDTA or 5 mM CaCl 2 were sequentially injected into a sample cell containing 20 M SMTNL1-CH in the same buffer. All buffers contained 1 mM 2-mercaptoethanol to prevent intermolecular disulfide bonding in SMTNL1-CH. The concentration of each protein was determined using their predicted molar extinction coefficients (cm Ϫ1 M Ϫ1 ): CaM, ⑀ 280 ϭ 2560; CaM-ct, ⑀ 280 ϭ 2980; CaM-nt, ⑀ 259 ϭ 742; and SMTNL1-CH, ⑀ 280 ϭ 18450. All titrations were performed at 30°C, and the data were fit to a one-site binding model (MicroCal Origin software) to obtain dissociation constants (K d ).
NMR Measurements-All NMR experiments were carried out at 20°C on Bruker Avance 500 or 700 MHz NMR spectrometers equipped with triple resonance inverse Cryoprobes with a single z-axis gradient. Sequential assignments of HN, N, CO, C-␣ and C-␤ resonances of SMTNL1-CH were achieved using two-dimensional 15 (20). All NOESY experiments including the three-dimensional 15 N NOESY-HSQC, three-dimensional 13 C NOESY-HSQC, and two-dimensional NOESY were measured with a mixing time of 100 ms. { 1 H}-15 N heteronuclear NOE experiments were acquired on a 700-MHz spectrometer with a recycle delay of 5 s (21). The experiments were repeated three times and averaged. H-N RDC measurements were performed using the in-phase/antiphase 1 H, 15 N HSQC experiment (22). The chemical shift perturbation (CSP) studies were performed by monitoring the 1 H, 15 N HSQC spectra of 15 N-labeled SMTNL1-CH and 15 N-labeled apo-CaM by adding unlabeled apo-CaM and unlabeled SMTNL1-CH, respectively. The CSP value was then evaluated as a weighted average chemical shift difference of 1 H and 15 N resonances, using the equation CSP ϭ ͌(⌬HN) 2 ϩ (⌬N/5) 2 (23).
Structure Calculations-The initial SMTNL1-CH structure was calculated with CYANA version 2.0 (27) using distance restraints obtained from the automatic NOE assignment protocol. Dihedral angle restraints (, ) were predicted with TALOS (28), and hydrogen bond restraints were based on secondary structure from a chemical shift index for the C-␣ and CЈ atoms. Further structural refinement with the addition of RDC restraints were performed by XPLOR-NIH (29). Initial estimates for the axial component of the molecular alignment tensor (Da) and the rhombicity (R) were obtained on the lowest energy structure calculated by CYANA using PALES (30). Finally, the 30 lowest energy structures from a total of 200 were selected and analyzed.
Docking Model for the SMTNL1-CH⅐Apo-CaM Complex-Based on the CSP data, a docking model for the SMTNL1-CH⅐apo-CaM complex was calculated with the HADDOCK2.0 (high ambiguity driven biomolecular docking) program in conjunction with CNS (31). The HADDOCK program is originally designed for the generation of protein⅐protein complex structures based on available experimental data (17). According to the outcome of the CSP and ITC data, only the C-terminal domain of apo-CaM was employed in this calculation. The 30-structure ensemble for SMTNL1-CH (determined in this study) and the 10-structure ensemble for the C-terminal domain of apo-CaM (residues 82-148; Protein Data Bank code 1F71) (32) were used as the starting structures. The average solvent accessibilities per residue in the ensemble structures

Complex Structure of the CH-domain from SMTNL1
were calculated by NACCESS. The high solvent accessible residues (Ͼ40%) that exhibited a CSP Ͼ0.04 for SMTNL1-CH and Ͼ0.025 for the C-terminal domain of apo-CaM were designated as active residues (see below). The residues neighboring those active residues with a high solvent accessibility (Ͼ40%) were designated as passive residues. The active residues in SMTNL1-CH were Gly 4 , Lys 6 , Asn 7 , Asp 93 , Lys 95 , Thr 99 , Glu 103 , Arg 106 , and Gly 112 . The associated passive residues were Ser 5 , Ser 94 , Cys 96 , Gln 102 , Val 109 , Gln 110 , and Lys 111 . For the C-terminal domain of apo-CaM, the active residues were Lys 94 , Asp 95 , Gly 96 , Tyr 99 , Asp 131 , Gly 134 , and Gln 135 , and the passive residues were Asn 97 , Ser 101 , Gly 132 , Asp 133 , and Asn 137 . The positions of the active residues for both proteins are indicated on their structure (see below). In the first stage of the calculation, an initial set of 1000 rigidbody docking models was generated. The 200 lowest energy complex models were then selected and submitted for a second stage of calculations with semi-flexible simulated annealing. The 50 lowest energy models in the second stage were refined in water solvent and clustered using a 3.0 Å r.m.s.d. cutoff criterion. Finally, the 15 lowest energy complex models were selected from the most populated cluster with the lowest HAD-DOCK score and used for the analysis. All molecular graphics were created with MOLMOL (33).

RESULTS
Structure Determination of SMTNL1-CH-Using the HNCACB, HN(CO)CACB, HN(CA)CO, and HNCO experiments, all amide resonances except for Gly 1 , Pro 47 , Pro 52 , and Pro 92 were assigned in the 1 H, 15 N HSQC spectrum (Fig. 2). The side-chain resonance assignments were mainly obtained from the C(CO)NH TOCSY, H(CCO)NH TOCSY, and HBHA-(CBCACO)NH spectra. Because SMTNL1-CH contains a total of 16 aromatic residues, the (HB)CB(CGCD)HD and (HB)CB-(CGCDCE)HE experiments were also carried out to obtain unambiguous resonance assignments of the aromatic side chains. Consequently, 94.6% of the total 1 H resonances were assigned and used in the structure calculation. The initial structures were calculated using the torsion angle dynamics and the automatic NOE assignment protocol with CYANA. At the first stage, a total of 3521 NOE signals were manually identified on the three-dimensional 15 N and 13 C NOESY-HSQC spectra as  well as the two-dimensional NOESY spectrum acquired in D 2 O. 3261 NOE signals were automatically assigned by CYANA, and this generated 2136 distance restraints. The distribution of the distance restraints was determined as a function of residue number (Fig. 3a) Fig. 4, a and (34), and 84% of the residues were found in favored regions of the Ramachandran plot; the remaining residues were all found in the additionally allowed regions ( Table 1). The SMTNL1-CH is a globular molecule containing five ␣-helices and one 3 10 -helix with no ␤-sheet structure (Fig. 4b).
Characterization of the Apo-CaM Interaction with SMTNL1-CH-The binding of apo-CaM with SMTNL1-CH was first characterized by ITC experiments. Apo-CaM exhibited an exothermic interaction with SMTNL1-CH (K d ϭ 2.7 ϫ 10 Ϫ6 M) in the presence of EDTA, whereas no heat of binding was detected in the presence of Ca 2ϩ (Fig. 6a). We also performed ITC experiments with the proteolytic C-and N-terminal fragments of apo-CaM. CaM-ct exhibited an exothermic interaction with a similar K d (4.0 ϫ 10 Ϫ6 M) to that observed with intact CaM. However, CaM-nt did not bind to SMTNL1-CH (Fig. 6b).
CSP values, determined from the 1 H, 15 N HSQC spectrum of SMTNL1-CH following the addition of unlabeled CaM, were plotted as a function of residue number (Fig. 7a). Amides that experience CSP Ͼ0.04 could be mapped to residues located either on the IQ-motif sequence of SMTNL1-CH or to regions that were in close proximity (Fig. 7c). Similarly, the CSPs were monitored in the 1 H, 15 N HSQC spectrum of apo-CaM by adding unlabeled SMTNL1-CH into the NMR sample (Fig. 7b). Residues with CSP values Ͼ0.025 were only found in the C-terminal domain of CaM and were located around its two Ca 2ϩ binding loops (Fig. 7d). The backbone structure of 15 refined complex models and a ribbon representation of the lowest energy model for the interaction of SMTNL1-CH with the C-terminal domain of apo-CaM were generated (Fig. 8). The average backbone r.m.s.d. for the 15 complex models was 0.95 Ϯ 0.33 Å. In Fig. 8b, the side chains that form intermolecular hydrogen bonds at the interface of the docking model are also indicated. Those side chains were defined as belonging to Lys 6 , Lys 95 , Glu 103 , Arg 106 , and Lys 115 on SMTNL1-CH that form a hydrogen bond to the side chains of Asp 95 , Asn 97 , Tyr 99 , Asp 133 , and Asp 131 on apo-CaM, respectively.

DISCUSSION
We have undertaken a structural analysis of the C-terminal type-2 CH-domain of the smoothelin-like 1 protein to assist in our understanding of the biological function of this novel member of the smoothelin family of smooth muscle-specific proteins. Herein, we illustrate that the CH-domain of SMTNL1 maintains a helix-rich, globular structure that is typical of that found for the CH-domains of many other proteins (Fig. 4b). A number of conserved hydrophobic residues, including the two Trp residues in SMTNL1-CH, contribute to a hydrophobic core that is conserved in the structure of other CH-domains ( Figs. 1 and 4b). The average backbone { 1 H}-15 N NOE value in the well folded region (residues 5-111) is Ͼ0.8 without any considerable flexible regions, suggesting a very rigid protein conformation (Fig. 3a). This observation is similar to the reported { 1 H}-15 N NOE data for the type-3 CH-domain of calponin (38). SMTNL1-CH shares high sequence identity with the CH-domain of smoothelin, and not surprisingly, the backbone r.m.s.d. is 1.15 Å when the SMTNL1-CH structure was superimposed on the recently deposited CH-domain structure of smoothelin (Fig. 5a). Despite the relatively low level of sequence identity (37.6%, 34.2%, and 22.2% to the CH-domains of actinin 1, spectrin, and MICAL-1, respectively), the threedimensional structures were very similar to one another (Fig. 5,  b, c, and d). The most remarkable difference in the structure of SMTNL1-CH as compared with the other CH-domains is the extended C-terminal ␣-helix (helix VI) (Fig. 5). The unique basic cluster "KTKKK" at the tail of helix VI bestows a highly basic surface to the SMTNL1-CH molecule (Figs. 1 and 4c). Another basic surface is formed by residues 59 -61 and 88 (Figs. 1 and 4d) and is also unique to the CH-domains of the smoothelin family proteins. Unlike the majority of the CH-domains, which are generally located at the N terminus of proteins, helix VI of SMTNL1-CH is located at the C terminus of the SMTNL1 protein and is exposed to the solvent. The recently reported solution structure of the MICAL-1 type-2 CH-domain (37) also demonstrated a fairly long helix at its C terminus. However, this CH-domain is located in the middle of the MICAL-1 protein, and the basic KTKKK cluster is absent (39).
A most interesting new finding from this work is the presence of a CaM-binding IQ-motif on helix VI of SMTNL1-CH that is also conserved in the other smoothelin family members (Fig. 1b). The typical IQ-motif, with consensus sequence IQXXXRGXXXR, was first characterized in the heavy chain of many myosin motor proteins as multiple tandem repeats (40) targeting either CaM or CaM-related molecules (41,42). The IQ-motif sequence with minor substitutions has also been identified in many CaM-binding proteins such as unconventional myosins, PEP-19, and IQGAP-like proteins (42)(43)(44). We tested whether CaM was capable of binding to SMTNL1-CH, and the K d value obtained by isothermal titration calorimetry was 2.7 ϫ 10 Ϫ6 M for apo-CaM (Fig. 6a). This is in line with many previously examined apo-CaM target protein interactions. For example, the reported K d values are 10 Ϫ5 M for neurogranin (45) and PEP-19 (46) and 10 Ϫ7 M for neuromodulin (47) and IQGAP1 (48), whereas K d values vary from 10 Ϫ5 to 10 Ϫ8 M for the IQ-motifs in myosin V (IQ1-6) (49). These apo-CaM complexes are distinct from the typical Ca 2ϩ ⅐CaM complexes that represent a tight binding (K d ϳ 10 Ϫ8 -10 Ϫ9 M) that lead to enzyme activation. The formation of apo-CaM complexes prevents CaM from diffusing in the cytoplasm. The localization of CaM at specific sites in the resting cell is believed to facilitate a direct response to an influx of Ca 2ϩ (50,51). The 1 H, 15 N HSQC titration experiments showed the formation of a 1:1 complex with slow exchange on the NMR time scale (data not shown). The stoichiometry derived from ITC experiments was Ͻ1 (Fig.  6a), and this discrepancy was probably caused by the difficulty of determining the binding stoichiometry of the relatively weak interaction by ITC experiments under the conditions of low protein concentrations (ϳ20 M). Although most IQ-motifs are known to bind to apo-CaM, some proteins and peptides containing IQ-motifs are also capable of binding Ca 2ϩ ⅐CaM (42). Our ITC experiments suggested either extremely weak or no binding of Ca 2ϩ ⅐CaM with SMTNL1-CH (Fig.  6a). Consistent with the outcome of these calorimetry studies, in 1 H, 15 N HSQC titration experiments with Ca 2ϩ ⅐CaM we only observed a few signals that belong to the N terminus of SMTNL1-CH that showed minor perturbations in the fast exchange regime on the NMR time scale, which might be caused by slight changes in the NMR sample conditions (data not shown). We therefore conclude that only apo-CaM binds efficiently to SMTNL1-CH. This behavior closely resembles that of apo-CaM binding to neuromodulin, neurogranin, and IQGAP1, in which Ca 2ϩ ⅐CaM poorly interacts with these proteins (42,43,48). Hence, in the remainder of our discussion, we will focus on the novel interaction identified between apo-CaM and SMTNL1-CH. The CSP study and the 1 H, 15 N HSQC experiments clearly demonstrated that the interface of apo-CaM binding was located around the IQ-motif sequence of SMTNL1-CH (Fig.  7c). Previous models of the crystal structure of myosin V complexed with the CaM-like protein, Mlc1p, together with fluorescence resonance energy transfer distance measurements, indicated that CaM bound to IQ4 and IQ6 using only its C-terminal domain, whereas the N-terminal domain remained free in solution (52). In SMTNL1-CH, the CaM-ct showed a similar binding affinity to that observed with intact CaM, whereas the CaM-nt did not show binding (Fig. 6b). However, the amount of heat (⌬H) generated by intact apo-CaM binding to SMTNL1-CH was almost eight times higher than that by CaM-ct (Fig. 6b). On the other hand, the clear CSPs observed in the 1 H, 15 N HSQC spectrum of apo-CaM that were induced by binding to SMTNL1-CH only locate to the C-terminal domain of apo-CaM (Fig. 7, b and d), which is consistent with the ITCdata, and only small CSPs are observed in the N-terminal domain. These results suggest that apo-CaM binds to SMTNL1-CH mostly through its C-terminal domain, whereas the N-terminal domain possibly contributes minor interactions. As the magnitude of the CSP observed in both spectra is relatively small, there is no significant alteration of the structures of both proteins that is induced by the binding. Based on these data, we can therefore construct a docking model for the SMTNL1-CH complex with the C-terminal domain of apo-CaM using the HADDOCK program. The interaction appears to be electrostatic where several acidic residues (including Asp 95 , Asp 131 , and Asp 133 ) of the Ca 2ϩ binding loops of the C-terminal domain of apo-CaM form intermolecular hydrogen  bonds to basic residues (including Lys 6 , Lys 95 , Arg 106 , and Lys 115 ) of SMTNL1-CH. In this calculation, the potential active residues, Lys 117 and Lys 118 , that are located on the disordered C-terminal region of SMTNL1-CH structure were not employed. However, considering their relatively large CSP values, as well as their close proximity to the CaM protein, Lys 117 and/or Lys 118 are also likely to form additional electrostatic interactions to CaM. The recent crystal structure of apo-CaM complexed with IQ-motifs from myosin V suggested that the C-terminal domain of CaM adopted a semi-open conformation to grab the first half of the IQ-motif through a hydrophobic interaction (53). In this complex, the Ile residue at the first position and a hydrophobic residue at the fifth position (Ile or Val) of the IQ-motif sequence make important hydrophobic contacts to the semi-open hydrophobic patch of the C-terminal domain of apo-CaM (Fig. 1). However, in SMTNL1-CH, the first Ile residue, Ile 101 , is buried in the protein structure and is therefore not available to contribute such a hydrophobic interaction. In addition, the hydrophobic residue at position five of the myosin V IQ-motif (Ile or Val) is replaced with Tyr in SMTNL1-CH (Fig. 1). Many side chains of the C-terminal domain of apo-CaM that form major electrostatic contacts to SMTNL1-CH, including residues Asp 95 , Asn 97 , Asp 131 , and Asp 133 , also serve as the Ca 2ϩ binding ligands, suggesting that this interaction cannot be made once CaM binds Ca 2ϩ ions. This is in agreement with our ITC and NMR results, in which only apo-CaM seems capable of binding to SMTNL1-CH (e.g. Fig. 6a).
The addition of SMTNL1 can induce relaxation of permeabilized ileal smooth muscle strips (7), and our recent measurements of muscle force using SMTNL1 fragments suggest that the removal of SMTNL1-CH is critical for the relaxant potential of the protein. 7 On the other hand, the mechanism whereby SMTNL1 can induce muscle relaxation is not understood. Although at this moment, the role of the SMTNL1-CaM interaction has not yet been defined, we speculate that the association of apo-CaM with SMTNL1 has physiological relevance. There are several mechanisms whereby a functional SML-1apo-CaM interaction could modulate the contractility of smooth muscle. 1) The colocalization of CaM with the CH-domain of SMTNL1 could be required for the formation of ternary complexes with other target proteins of SMTNL1. 2) CaM binding might influence the phosphorylation efficiency of SMTNL1 or, alternatively, phosphorylation may influence CaM binding efficiency under conditions of Ca 2ϩ desensitization. The cGMP-dependent protein kinase phosphorylation site (7) is located outside of the CH-domain but in close proximity to the CaM binding site.
3) The C-terminal basic cluster of SMTNL1 that also exists in smoothelin could interact with other acidic molecules, and because of the 7 J. A. MacDonald, unpublished observations. FIGURE 7. Identification of the interfaces for SMTNL1-CH-apo-CaM interaction. a, CSP induced by apo-CaM binding to SMTNL1-CH is plotted as a function of the residue number. b, CSP induced by SMTNL1-CH binding to apo-CaM is plotted as a function of the residue number. The locations of the N-and C-terminal domains of CaM are indicated. c, amide proton atoms of residues with CSP Ͼ 0.04 are mapped as spheres on the SMTNL1-CH structure. The region that contains the IQ-motif-like sequence is circled. d, amide proton atoms with CSP Ͼ 0.025 are mapped as spheres on the structure of the C-terminal domain of apo-CaM. The spheres are colored the same in panels c and d. The residues that are employed as active residues in the HADDOCK calculation are shown in blue, whereas the others are shown in green.
overlap with the CaM binding site, CaM might compete with those targets. 4) As SMTNL1 interacts with common acidic residues on CaM for Ca 2ϩ binding (54), it could modulate the Ca 2ϩ association and/or dissociation of CaM, as has also been discussed in the case of PEP-19 (55,56). 5) Finally, as we mentioned, binding of apo-CaM to proteins, followed by dissociation of Ca 2ϩ ⅐CaM, has often been described as a method for intracellular localization of CaM (50). Such localization would influence the kinetics for Ca 2ϩ ⅐CaM regulated processes in vivo.
As CaM binding is one of the miscellaneous functions of the CH-domain of calponin, SMTNL1-CH might also be a multifunctional domain. As we have described in this manuscript, the C-terminal type-2 CH-domain has distinct structural properties from the other types of CH-domains. Therefore, identification of additional proteins that interact with the CH-domains of smoothelin and SMTNL1 in smooth muscle cells is necessary for a more complete understanding of the function of the C-terminal type-2 CH-domain.