View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Structure, Vol. 10, 249–258, February, 2002, 2002 Elsevier Science Ltd. All rights reserved. PII S0969-2126(02)00703-7 Solution Structure of the Calponin CH Domain and Fitting to the 3D-Helical Reconstruction of F-Actin:Calponin

Janice Bramham,1,6 Julie L. Hodgkinson,2 tropomyosin, caldesmon, and calponin. Basic calponin, Brian O. Smith,3 Dusan Uhrı´n,4 Paul N. Barlow,3,4 one of three genetic variants found almost exclusively in and Steven J. Winder5 smooth muscle, was first isolated from chicken gizzard 1 Department of Biochemistry smooth muscle as a 34 kDa protein that binds to filamen- University of Leicester tous actin (F-actin) [1]. It has been proposed that cal- Adrian Building ponin has a role both in muscle contraction and as a University Road structural component in the cytoskeleton of smooth Leicester LE1 7RH muscle cells. Calponin is associated with the contractile United Kingdom apparatus, along with actin, myosin, and caldesmon. It 2 Imperial College School of Science also occurs in the membrane skeleton that forms the Technology and Medicine interface between the contractile apparatus and the ex- National Heart and Lung Institute tracellular matrix, and it appears in the cytoskeleton, Dovehouse Street which surrounds and supports the contractile appara- London SW3 6LY tus, along with ␤-actin, filamin, and desmin [2]. United Kingdom In vitro studies have shown that calponin is an actin 3 Institute of Cell and Molecular Biology binding protein and a major regulator of muscle contrac- University of Edinburgh tion. The regulatory function arises from the ability of Mayfield Road calponin to modulate the interaction between myosin Edinburgh EH9 3JR and actin filaments by inhibition of the actomyosin Scotland ATPase [3]. In addition to its ability to bind to F-actin, 4 Department of Chemistry calponin is also reported to interact with myosin itself [4] Joseph Black Building and with a number of other cytoskeletal components, in- University of Edinburgh cluding tropomyosin [1], desmin intermediate filaments West Mains Road [5], tubulin [6], and phospholipids [7]. The reported inter- Edinburgh EH9 3JJ actions between calponin and several signal-transduc- Scotland ing proteins, such as the extracellular-regulated kinase 5 Institute of Biomedical and Life Sciences (ERK) and protein kinase C-⑀ (PKC⑀) [8,9], suggest fur- Division of Biochemistry and Molecular Biology ther potential functions for calponin as a signaling mole- Davidson Building cule or as an adaptor for the localization of these kinases. University of Glasgow The calponin molecule consists of four recognized Glasgow G12 8QQ domains: one at the N terminus, the calponin homology Scotland (CH) domain, and three repeating domains comprising the C terminus, the calponin-like repeats (CLR). The CH Summary domain has approximately 100 residues and was first identified in calponin from which its name is derived Calponin is involved in the regulation of contractility and [10]. Tandem pairs of CH domains are found in a number ␣ organization of the actin cytoskeleton in smooth muscle of actin binding proteins such as -actinin, dystrophin, cells. It is the archetypal member of the calponin homol- filamin, fimbrin and spectrin (reviewed by Stradal et al., ogy (CH) domain family of actin binding proteins that 1998 [11]). In these proteins, a role for the CH domain in includes cytoskeletal linkers such as ␣-actinin, spectrin, actin binding has been demonstrated: two CH domains and dystrophin, and regulatory proteins including VAV, positioned in tandem within a molecule are required to IQGAP, and calponin. We have determined the first form a high-affinity actin binding domain (ABD), thus structure of a CH domain from a single CH domain- potentially providing multiple sites for interaction with containing protein, that of calponin, and have fitted actin. The single CH domain of calponin is not part of the NMR-derived coordinates to the 3D-helical recon- an ABD and its precise role in actin binding has yet to struction of the F-actin:calponin complex using cryo- be determined [12]. Similarly, single copies of the CH electron microscopy. The tertiary fold of this single domain are found in other cytoskeletal and signaling CH domain is typical of, yet significantly different from, proteins where its function is unclear, including the Sac- those of the CH domains that occur in tandem pairs charomyces cerevisiae homolog Scp1, the smooth mus- to form high-affinity ABDs in other proteins. We thus cle protein SM22, the proto-oncogene Vav, and IQGAP provide a structural insight into the mode of interaction (reviewed by Stradal et al., 1998 [11]). between F-actin and CH domain-containing proteins. A number of in vitro studies have defined regions within the calponin molecule involved in its interactions Introduction with other proteins. One binding site for actin has been located between the CH domain and the first CLR in the The thin filaments of smooth muscle cells comprise actin C-terminal portion, in the region that is also required for filaments complexed with three other major proteins: Key words: actin binding, calponin-homology domain, NMR, struc- 6 Correspondence: [email protected] ture, calponin Structure 250

Figure 1. Analysis of the NOE and CSI Data for the Calponin CH Domain (A) Number of unambiguous NOE-derived distance restraints used in the structure calculations. Black, gray, and white bars correspond to sequential (i → i ϩ 1), medium range (i → i ϩ (2, 3, or 4)), and long range (i → i ϩ (Ͼ4)) NOEs, respectively.

(B) NOEs between sequential backbone amide protons (Hn) and between Hn and H␣ (the heights of the bars are proportional to the NOE intensity). (C) Consensus CSI derived from the chemical shifts of C␣,C␤, CO, and H␣. (D) Cartoon representation of the secondary structure elements of the calponin CH domain. inhibition of the actomyosin ATPase (the so-called actin tion structure of any CH domain. The structure is com- binding site 1, ABS1) [13]. A second actin binding site pared to those of the crystal structures of the actin (ABS2) lies within the region spanning the three C-ter- binding CH domains and the implications for the ob- minal CLRs [14, 15]. Two binding sites for Ca2ϩ binding served differential actin binding properties are dis- proteins, such as calmodulin and S100, have been iden- cussed. The structure has also been fitted to the cryo- tified: one within an N-terminal region encompassing EM reconstruction of F-actin decorated with calponin; residues 7–45 and the second in the region between a comparison with the F-actin:CH domain reconstruc- residues 52 and 144; residues 142–227 contain a tropo- tions of other CH domains is discussed. myosin binding site [16]. Additionally, phosphorylation sites have been identified on all three CLRs [17]. The Results and Discussion relative significance of these reported in vitro interac- tions of calponin in vivo and of its distribution in smooth Structure Determination muscle has yet to be ascertained. Calponin may function A 1 mM sample of 15N,13C-labeled CH domain (residues separately both as a regulator of smooth muscle con- 27–134) yielded high-quality NMR spectra such that, traction and as a structural protein in smooth muscle. using a combination of complementary triple-resonance The distribution of the sites of interaction throughout experiments, nearly all of the 1H, 15N, and 13C nuclei in this modular molecule implies that the individual do- the protein could be unambiguously assigned [22]. The 1 13 13 mains may have separate functions in vivo. chemical shifts of all the assigned H␣, C␣, C␤, and The crystal structures of several CH domains from 13CO provided a consensus Chemical Shift Index (CSI) ABDs comprising tandem pairs of CH domains have that is consistent with a predominantly helical structure been reported [18–21]. In this report we describe the (Figure 1C). Five ␣ helices were predicted with no evi- three-dimensional solution structure of the archetypal dence of ␤ strands, concordant with the previously re- CH domain from chicken gizzard calponin, determined ported crystal structures of CH domains. Initial struc- by multinuclear NMR spectroscopy. This is the first tures were calculated using 3347 distance restraints structural investigation of a CH domain from a protein derived from the integrated crosspeaks in the 13C-edited containing only a single CH domain and is the first solu- NOESY-HSQC spectrum and the 15N-resolved cross- Solution Structure of the Calponin CH Domain 251

Table 1. Structural Statistics of the Final Ensemble of Twenty Refined Structures of the CH Domain of Chicken Calponin Ensemble Closest to Mean Number of NOE restraints used: Intraresidue (i → i) 1034 1034 Sequential (i → i ϩ 1) 320 320 Medium range (i → i ϩ (2 to 4)) 234 234 Long range (i → i ϩ (Ͼ4)) 315 315 Ambiguous 672 672 Total 2575 2575 Average NOE restraint violation (A˚ ) 0.031 Ϯ 0.007a 0.033 Lennard-Jones energy (kJmolϪ1)b Ϫ207 Ϯ 49 Ϫ222 Coordinate rmsd (A˚ ): All residues* (backbone heavy atoms) 0.580 0.493 (all heavy atoms) 0.990 0.873 Residues in helices† (backbone heavy atoms) 0.448 0.381 (all heavy atoms) 0.828 0.708 Parameter rmsd from idealized geometry: Bond lengths (A˚ ) 0.0018 Ϯ 6 ϫ 10Ϫ4c 0.0019 Angles (Њ) 0.331 Ϯ 0.007c 0.329 Improper dihedrals (Њ) 0.233 Ϯ 0.076c 0.232 Ramachandran assessment (%)d: Most favored region 84.5 87.2 Additionally allowed region 9.1 6.4 Generously allowed region 5.1 6.4 Disallowed region 1.3 0.0 a The sum of the NOE violations divided by the total number of restraints, averaged over the ensemble. b The Lennard-Jones potential was not used at any stage in the refinement. c The rmsd for each structure, averaged over the ensemble. d Based upon location of all residues except glycines and prolines (using PROCHECK[46]). * Residues 33–132 † Helices: residues 33–41, 58–68, 81–97, 116–132

13 15 peaks in the C, N HSQC-NOESY-CH3NH spectrum dered terminal residues, is shown in Figure 3A. This [23]. After iteratively filtering the ambiguously assigned ensemble satisfies the experimentally derived distance restraints against the calculated structures and removal restraints with, on average, less than one NOE violation of the duplicate restraints, a total of 2575 distance re- greater than 0.4 A˚ per structure. The Ramachandran straints were used as the input data for the structure analysis of the main chain torsion angles (φ, ␸) of this calculations (Table 1). ensemble revealed that 84.5% of residues lie within the The structure of the expressed protein in solution is most favored region (see Table 1). well defined by the experimental data except for the four N-terminal residues (M27–T30) and the two C-terminal Calponin CH Domain Structure residues (T133 and K134) that are disordered. The reso- The tertiary fold of the CH domain of calponin is compact nances arising from the backbone amide protons of and globular (Figure 3B); it is based upon a core of four residues M27, Q29, and T30 (residue 28 is a proline) are main ␣ helices (I, III, IV, and VI) connected by long loops, absent from the NMR spectra implying that, at pH 7.0, with two short helical structures (II and V). Three of the these protons are in rapid exchange with the bulk water. core helices (III, IV, and VI) form a triple-helix bundle in This fact, together with the low number of interresidue which helices III and VI are approximately parallel with NOEs observed for the N-terminal residues (see Figure helix IV lying at an angle of approximately 40Њ. The 1A), is consistent with flexible, random-coil structure in N-terminal helix I lies perpendicular to helices III and VI this region. The 15N relaxation and 1H-15N heteronuclear and the two short helices (II and V) are located close NOE experiments indicate that the two C-terminal resi- together at the N termini of helices III and VI, respec- dues are also more mobile in solution relative to the rest tively. A 310-helical turn is also present in the loop be- of the domain (Figures 2A and 2B). tween helices IV and V. The compact fold of the structure There is a good correlation between the consensus is maintained by a network of hydrophobic interactions. CSI, the pattern of sequential and interresidue NOE in- In particular, helices I and IV are tethered at opposite teractions characteristic of ␣-helical structure (depicted sides of the domain by extensive hydrophobic interac- in Figure 1), and the resulting secondary structure ele- tions with both helixes III and VI, while the smaller helix ments. Four of the ␣ helices predicted by the CSI corre- II has a significant number of interactions with helices spond to helix I (R32–T42), helix IV (N81–G99), helix V III, IV, V, and VI. (E107–E113), and helix VI (H117-T133) in the final struc- Notable observations made in the NMR data can be ture. The remaining CSI-predicted ␣ helix corresponds rationalized by inspection of the refined structure. For to a region spanning, but considerably shorter than, both example, one H␤ from L60 exhibits an abnormal upfield helix II (F50–D56) and helix III (V58–Q68) (Figure 1C). chemical shift of Ϫ0.19 ppm in the NMR spectra. In the The final ensemble of structures, excluding the disor- refined structures residue L60 is located close to F91, Structure 252

Figure 2. NMR Relaxation Data for the Cal- ponin CH Domain 15 (A) NT1 and T2 relaxation times for the back- bone amides of the calponin CH domain, shown as triangles and circles, respectively. (B) 1H-15N heteronuclear NOEs for the back- bone amides. Error bars were calculated us- ing the spectral noise. (C) Rmsd of the C␣ for the ensemble of 20 structures.

within the hydrophobic core, with one H␤ lying above dues) is 0.45 A˚ compared to 0.58 A˚ for the whole domain the center of the aromatic ring of F91 and therefore (residues 33–132), again consistent with only a slight subject to nuclear shielding from the ring current of the lack of precision in the loop regions compared to the aromatic ring, resulting in the observed upfield shift. helices (Table 1). Apart from at the termini, the only The extended nature of loop structures results in few significantly higher values of the rmsds of the C␣ posi- short-range interresidue NOE interactions irrespective tions in the ensemble of structures are in the interhelical of the flexibility of the loop. In the calponin CH domain, loop regions immediately before helix II (residues G47– although there are generally fewer interresidue NOEs N49), between helices III and IV (Q73–K74) (Figure 2C). arising from the interhelical loop regions than from the These residues generally have few medium and very ␣ helices, there are sequential NOEs from all the residues few, if any, long-range NOE interactions and are conse- in loops and very few of these residues have no medium quently less well defined by the experimental data. In or long range NOE interactions (see Figure 1A). Conse- the loop region between helices III and IV, the “dip” in quently, and perhaps surprisingly, most of the loop re- the NOE values (V75–V79) and relatively higher T1 values gions in the calponin CH domain are only slightly less (V79 and Q80) may indicate that the lack of structural precisely defined by the experimental data than the ap- definition in this region is due to a genuine increased parently well ordered ␣ helices. The 15N relaxation data flexibility compared to the rest of the molecule (Figures and the 1H-15N heteronuclear NOEs for the backbone 2A and 2B). Also, in this region are found the only two amide groups show relatively consistent values through- nonterminal residues with no observable backbone am- out the domain, except for the disordered C terminus, ide resonances in the NMR spectra, probably due to indicating that there is little variability in flexibility rapid exchange with the bulk water (Q73 and N81). throughout the molecule (on the 109Ϫ1012 sϪ1 timescale). The superposition of the backbones of the ensemble of Comparison of CH Domain Structures structures highlights, in particular, the precise definition The tertiary fold of the calponin CH domain is very similar of the triple helical bundle formed by three of the core to those of the CH domains from the ABDs of the actin helices (III, IV, and VI). The mean rmsd of the backbone binding proteins dystrophin, utrophin, spectrin, and fim- heavy atoms of the four core helices (a total of 54 resi- brin, previously determined by X-ray crystallography Solution Structure of the Calponin CH Domain 253

Figure 3. Solution Structure of the CH Domain of Chicken Calponin (A) A stereo plot of the backbone traces of an ensemble of 20 structures (residues 31–133) superimposed on the backbone heavy atoms of the four major helices. (B) Cartoon representations of the structure closest to that of the mean of the ensemble, in two orthogonal orientations. The plots were produced using MOLSCRIPT [51] and Raster3D [52].

[18–21]. Figure 4 shows the sequences of these CH of fimbrin all lack the helical turns that are present within domains aligned on the basis of their secondary struc- the loop between helices III and IV in the structures of tural elements, highlighting the similarity in the helical all the remaining CH domains. The two fimbrin domains, regions. The four core helices (I, III, IV, and VI) and helix CH1.1 and CH2.1, also have significantly longer loops V are conserved in all the CH domain structures; the between helices I and II, and between helices II and IV, main differences lie out with these helices. The CH1.1 respectively. Calponin is unique in having a short 310 domain of fimbrin is the only other structure with a de- helix between helices IV and V. Helix I is shorter in cal- fined ␣ helix corresponding to helix II in the calponin ponin than in all the other structures, and the loop be- domain. Although all the other structures are also helical tween helices I and II is also much shorter. The noncon- in this region, the CH2 domains of the spectrin family served glycine at the C terminus of helix I in calponin, (␤-spectrin, dystrophin, and utrophin) contain a shorter G43, may permit the necessary conformational freedom

310 helix here while the remaining structures do not sat- in the following loop required to maintain the relative isfy the criteria for canonical helices here, using PRO- positions of helices I and II. MOTIF [24]. The CH domain of calponin, the CH1 do- Analysis of the roles of conserved residues within the mains of dystrophin and utrophin, and the CH2.1 domain structures may explain their conservation in CH do- Structure 254

Figure 4. Structure-Based Alignment of the Sequences of all the CH Domains Whose Structures Have Been Determined to Date

The secondary structural elements defined by PROMOTIF [24] are highlighted in yellow (␣ helix) and cyan (310 helix). Conserved residues are indicated as fully (green), strongly (‡), and weakly conserved (†). Residues unique to calponin that are otherwise conserved are highlighted in red. The numbering scheme is that of chicken calponin and the numbered cylinders represent the ␣ helices of the calponin CH domain. The names and residue numbers are as follows: CAP, chicken calponin (28–134); DYS1, human dystrophin (13–119); UTR1, human utrophin (31–135); DYS2, human dystrophin (135–243); UTR2, human utrophin (151–258); SPC2, human ␤-spectrin (173–280); FIM1.1, human fimbrin (121–241); FIM2.1, human fimbrin (268–375). mains. For instance, the length of helix III is maintained tion of the helices, with no two structures having rmsd by a proline or glycine at the C terminus (P69 in calponin) greater than 1 A˚ for the backbone atoms of the helices and a fully conserved, dipole-compensating aspartate in the pairwise comparison. The data is concordant with (D56), capping the N terminus. The negatively charged the sequence analysis of all known CH domains which, side chain of the fully conserved aspartate in the middle in summary, determined that all the proteins containing of helix V (D110) appears to interact with the backbone a single CH domain are grouped together in one family amides of the corresponding i-3 residue (E107), possibly and that there are two other distinguishable families— stabilizing the short helix V; alternatively, this conserved the fimbrin family and the spectrin family—both of which residue may also have a role in protein:protein interac- contain tandem pairs of CH domain comprising an actin tions. Such analyses can also be informative when com- binding domain (ABD) [11]. Proteins in the fimbrin family paring the structures of the different CH domains. In each contain two ABDs, i.e., two tandem pairs of CH particular, the fully conserved tryptophan in helix I (W37) domains. The spectrin family proteins, including ␤-spec- is in the same location in all the structures and is in the trin, ␣-actinin, dystrophin, and utrophin, contain only same orientation, stacked between residue i ϩ 4 in helix one ABD. Within the same family, the N-terminal CH I and a conserved large hydrophobic residue in helix VI, domains in the ABDs are more similar to each other in all the structures except the calponin CH domain and than, and are significantly different from, the C-terminal the fimbrin CH1.1 domain. In calponin, the side chains CH domains in the ABDs. The phylogenetic analysis thus of the corresponding stacking residues (A41 and A128) defines five classes of CH domain: the CH domains are smaller, allowing the tryptophan ring to adopt a from proteins containing single copies, including the different orientation. In the fimbrin CH1.1 structure, this calponins; the N-terminal (CH1) domains from the spec- residue is also orientated differently and, in this case, trin family; the C-terminal (CH2) domains from the spec- makes interdomain contacts to the fimbrin CH2.1 do- trin family; the N-terminal (CH1.1 and CH1.2) domains main within the dimer in the crystal structure [19]. from the fimbrin family; and the C-terminal (CH2.1 and Several residues are conserved in the sequences of CH2.2) domains from the fimbrin family. A further class all the CH domains except calponin (see Figure 4). The of CH domain-containing actin binding proteins has re- majority of these are in helix VI, including T118 (mostly cently been described although to date no structural K in the other domains), Q121 (I or L in the others), and data is available [25]. In the pairwise statistical analysis T123 (L, Y, or F in the others). The substitution of a large of the structures (Table 2), the lowest rmsds are between positively charged side chain with a threonine at the CH domains within these classes, indicating the agree- solvent-exposed position 118 (in calponin) might have ment with the sequence analysis. For example, the rmsd a significant effect on any potential protein-protein or between the CH2 domains of ␤-spectrin and dystrophin ligand interactions involving this residue. Both Q121 and is 0.3 A˚ . Similarly, CH domains within the same class T123 are involved in a number of hydrophobic interac- have similar interhelical angles (Table 2). The calponin tions, similar to the corresponding conserved hydropho- CH domain is significantly different to all the other do- bic residues in the other domains. The relatively smaller mains, in accord with its classification in a separate side chain of T123 is compensated for by larger side chains group. of residues that are close in space, such as F106 and F91. In helix IV, a strongly conserved, polar residue (N, Q, or Actin Binding E) is replaced by a glycine, G89, in the calponin domain. In addition to the phylogenetic classification distinguish- This does not appear to have any effect upon the tertiary ing the CH1 and CH2 domains, the CH1 domains are structure of the domain but again may have a significant functionally different from the CH2 domains. In isolation, effect on any potential interactions in this region. only the CH1 domain has the intrinsic ability to bind F-actin; however, a tandem pair of CH domains is re- CH Domain Families quired for the full actin binding potential of the ABD. In Table 2 shows the statistical comparison of the four the dystrophin and ␣-actinin ABDs, the binding sites core helices of all the CH domains whose structures proposed for actin are the N-terminal helices of the CH1 have been determined to date, illustrating the conserva- and CH2 domains (helix I) and in a region encompassing Solution Structure of the Calponin CH Domain 255

Table 2. Statistical Analysis of the CH Domain Structures, Performed Using the Backbone Atoms of the Four Core Helices of Each Structure

A. Rmsd Values of the Pairwise Comparison of All the CH Domains CAP CH DYS CH1 UTR CH1 DYS CH2 UTR CH2 SPC CH2 FIM CH1.1 FIM CH2.1 CAP CH — DYS CH1 0.79 — UTR CH1 0.79 0.25 — DYS CH2 0.72 0.60 0.57 — UTR CH2 0.76 0.65 0.62 0.19 — SPC CH2 0.76 0.62 0.58 0.30 0.31 — FIM CH1.1 0.78 0.75 0.68 0.69 0.74 0.70 — FIM CH2.1 0.83 0.65 0.70 0.58 0.59 0.67 0.72 — B. Interhelical Angles Determined Using PROMOTIF [24] I/III I/VI III/IV III/VI IV/VI CAP CH 78.5 80.0 40.4 16.0 45.4 DYS CH1 60.2 55.8 25.1 7.9 30.8 UTR CH1 59.5 55.1 28.3 4.4 32.5 DYS CH2 71.2 64.9 34.5 Ϫ10.3 37.3 UTR CH2 73.6 60.1 34.0 Ϫ16.4 40.6 SPC CH2 67.9 65.7 39.1 Ϫ2.8 39.8 FIM CH1.1 74.2 56.1 25.0 18.1 43.0 FIM CH2.1 80.9 69.4 17.0 13.0 29.7

helices V and VI [26–28]. In the isolated CH1 domains utrophin to the 3D-helical image reconstructions derived of dystrophin, utrophin, ␣-actinin, and ␤-spectrin, helix from their respective complexes with F-actin [32–34].

VI corresponds to a major site of interaction with F-actin In the fimbrin-like orientation the short 310 helix of the [29]. In the calponin CH domain structure, these regions calponin CH domain, helix V, is closest to the F-actin are the most divergent: helix I is shorter than in the other axis with helix VI also lying close to the actin interface structures and, as described above, helix VI is the least (Figure 5A), while in the utrophin-like orientation helix I conserved in sequence. is closest to the F-actin axis (Figure 5B). However, it We have previously shown, using a sedimentation was possible to rotate the calponin CH domain NMR assay, that intact calponin cosediments with F-actin [3]. structure within the plane of the asymmetry of the EM Similar in vitro experiments, however, indicated that the CH density envelope, hence the fits were not unique. In isolated CH domain of calponin and F-actin do not either orientation the calponin CH domain comes into cosediment (unpublished data). The lack of conserva- close proximity to sequential actin monomers in one tion in helix VI may explain why the isolated CH domain strand of the actin filament [termed the pointed end of calponin does not appear to bind strongly to F-actin. (ϩ1) and the barbed end (Ϫ1) monomers]. The closest Two actin binding regions have so far been located approaches on actin are to the bottom of subdomain 1 outside the limits of the CH domain in calponin, one of the pointed end monomer (residue 354) and the side immediately C-terminal of the CH domain [13] and a of subdomain 2 and top of subdomain 1 of the barbed second in the C-terminal repeat region [14, 15]. Never- end monomer (residues 41 and 92). To further constrain theless, the calponin CH domain is a major determinant the orientation of the fitted structure, the known binding for actin binding, serving to locate the calponin molecule sites for calponin on actin were taken into consideration. on the actin filament and adding considerably to the ABS1 (residues 145–163), the principal actin binding site overall affinity of calponin for actin [30]. in calponin, lies immediately C-terminal to the CH do- main and has been crosslinked to residues 326–355 of actin [30, 35]. Other crosslinking studies have also Fitting to the 3D-Helical Reconstructions delineated parts of subdomain 1 of actin as the principal of F-Actin:Calponin site of calponin interaction [15, 36, 37]. Furthermore, the The difference maps calculated between F-actin and possibility of interactions with subdomains 3 and 4 of calponin-decorated F-actin revealed a shield-like den- actin is excluded by tropomyosin binding as calponin sity that was identified, upon a variety of evidence, to displaces smooth muscle tropomyosin from the “off” comprise the calponin CH domain [31–33], with the other into the “on” position on F-actin [31]. The orientation regions of calponin being either extended or unstruc- of the C-terminal of the calponin CH domain toward tured and thus not clearly visible. Fitting the coordinates subdomain 1 of actin would satisfy the distance con- of the calponin CH domain structure to the EM difference straints imposed by the known binding location of the map of the CH domain was constrained by the overall calponin ABS1. The utrophin-like fit [34] clearly fulfils shape of the CH domain derived from the NMR structure. this requirement, whereas, in the fimbrin orientation, the Figures 5A and 5B show good fits in two orientations C terminus is oriented away from actin [32, 33] (compare that also locate likely actin-interacting sequences at the Figures 5A and 5B). The orientation of the utrophin CH surface of the CH domain reconstructed density enve- domain in the 3D-helical image reconstruction [34] lope interfacing actin. These orientations are consistent places the central helical region of the utrophin CH1 with the fitting of the CH domains from fimbrin and domain ABD extended down from the top of subdomain Structure 256

Figure 5. Side View of the Calponin CH Do- main Structure Fitted to the Difference Den- sity Map Derived from the 3D-Helical Im- age Reconstructions of Calponin-Decorated F-Actin The calponin CH domain, shown as backbone trace colored as Figure 3, has been fitted into the difference density (cyan mesh): (A) in the fimbrin-like orientation [33] and (B) in the utrophin-like orientation [34]. The pointed end monomer of actin is shown in green and the barbed end monomer is in red. Distances of the closest approaches to the C␣ of labeled residues (green) in the fitted atomic model of F-actin [53] are indicated in red (A˚ ).

1 toward residue 334 of actin (Ϫ1) (Figure 6B). Actin nus. The family of actin binding proteins that contain residue 334 has been implicated in the direct crosslink CH domains include the major cytoskeletal proteins with the calponin ABS1 [30]. Furthermore, visualization ␣-actinin, spectrin, dystrophin, and utrophin, all of which of the calponin-like repeat region (ABS2) by 3D-helical contain pairs of consecutive CH domains, and a number image reconstruction of decorated F-actin is consistent of regulatory proteins, including VAV, IQGAP, and cal- with such an orientation (J.L.H. and M. Gimona, unpub- ponin, which have only one CH domain. Several struc- lished data). Taken together, these data strongly sug- tures of actin binding domains (ABD) containing pairs gest that the calponin CH domain adopts the utrophin of CH domains have recently been solved [19–21]; how- orientation, with the C terminus pointing down (Figure ever, no structures exist from any single CH domain-con- 6A), and potentially describes a common mode of inter- taining protein. The role of the single CH domain in regula- action for CH domain-containing actin binding proteins, tory proteins such as calponin is still not clear; while whether containing one or two CH domains. dispensable for actin binding per se, it is nonetheless an important locator for the protein on the actin filament and Biological Implications is crucial to the regulatory function of calponin. Like other CH domains, the structure of the single Calponin is a 34 kDa actin binding protein involved in calponin CH domain is based upon four core ␣ helices the regulation of contraction and organization of the connected by long loops. The calponin structure, how- actin cytoskeleton in smooth muscle cells [38]. Calponin ever, is the most divergent of the solved CH domain has an important modulatory role on muscle contraction structures, in two of the core helices in particular, one through its ability to bind to F-actin and inhibit the cross- being considerably shorter and the other lacking several bridge cycling rate of myosin; this effect can be regu- residues that are otherwise conserved in CH domains. lated by the reversible phosphorylation of calponin [3, It is these features that put the calponin CH domain in 36]. The N-terminal half of the calponin molecule con- a unique class [11]. Furthermore, both of these helices tains a single domain of approximately 100 residues, correspond to proposed sites of interaction with actin termed the calponin homology (CH) domain [10], fol- in the ABD-containing proteins and may explain the dif- lowed by three so-called calponin repeats at the C termi- ferent actin binding ability of the isolated calponin CH

Figure 6. Relation of the Oriented Calponin CH Domain Structure to the Proposed Bind- ing Site for Calponin (ABS1) and Comparison with Utrophin-Induced Fit (A) Distances of the closest approaches of the fitted calponin CH domain C terminus (A˚ ; red labels) to ␣ helix 334–348 of actin pro- posed to bind ABS1 of calponin (green labels). (B) Part of the utrophin actin binding domain comprising CH1 and the interdomain helical linker (cyan) as modeled in the cryo-EM re- construction of utrophin with F-actin [34] in the context of two actin monomers; pointed end (green), barbed end (red), in one strand of the filament. Solution Structure of the Calponin CH Domain 257

domain. Using 3D image reconstruction of calponin- quality of the final ensemble of 20 refined structures was assessed decorated actin filaments, we have also determined the with PROCHECK [47] and PROMOTIF [24]. The latter was also em- orientation and position of the calponin CH domain on ployed to compare the structural properties of the different CH domain structures. F-actin. Not only does this provide a model from which The consensus Chemical Shift Index (CSI) was calculated, using to elucidate the molecular basis of calponin inhibition of all the 1H␣, 13C␣, 13C␤, and 13CO shifts assigned from the NMR experi- the acto-myosin ATPase, but also suggests a common ments, in the CSI program obtained from the University of Alberta mode of interaction between CH domains and F-actin, [48]. For the 15N relaxation and heteronuclear NOE experiments, only with the calponin CH domain serving to locate the cal- the nonoverlapping resonances of the backbone amide groups were analyzed. The T1 and T2 relaxation times were calculated by nonlinear ponin molecule on the actin filament thus augmenting least squares fitting. the actin binding affinity of its other actin binding sites. Electron Microscopy and Helical Reconstruction Experimental Procedures F-actin filaments decorated with calponin were reconstituted, and low-dose negative stain transmission electron microscopy was car- Cloning, Expression, and Purification ried out, as described previously [31]. Three-dimensional-helical DNA encoding residues 27–134 from chicken gizzard calponin was image reconstruction was carried out by standard methods using generated by the polymerase chain reaction from the cDNA and the Brandeis helical package [49]. Difference maps showing mass was inserted into the pET vector pSJW [39, 40]. The recombinant present only in the F-actin:calponin maps were calculated between protein was uniformly enriched with 15N and 13C by overexpression F-actin:calponin and F-actin alone as described previously [31]. in Escherichia coli and purified to homogeneity as described pre- Fitting of the coordinates of the NMR-derived solution structure of viously [22]. calponin to the reconstructed density was carried out using the program O [50]. NMR Spectroscopy and Structure Determination NMR spectra were acquired on a Varian INOVA 600 MHz spectrome- Acknowledgments ter using a 5 mm triple resonance probe equipped with triaxial gradi- mM sample of the 15N,13C-labeled This work was supported by a grant from the UK Biotechnology 1ف ents. Data were collected on a 2 CH domain (10% v/v H20, 1 mM sodium phosphate buffer, pH 7.0, and Biological Sciences Research Council (B.B.S.R.C.) Integration not adjusted for isotope effects, 291 K). All NMR data were pro- of Cellular Responses Initiative to P.N.B. and S.J.W. and by a Well- cessed using the Azara suite of programmes provided by W. Bou- come Trust grant (to S.J.W.). B.O.S. is a MRC Fellow and J.L.H. is cher and the Department of Biochemistry, University of Cambridge, funded by a Wellcome Trust grant. We are grateful to Professor UK, using maximum entropy methods for the three-dimensional M.P.Walsh for the chicken calponin cDNA and to Dr. E. Morris for experiments. ANSIG was employed to display the spectra and for advice in using the O program. picking, assignment and integration of the crosspeaks [41]. Using a combination of complementary triple-resonance experi- Received: August 31, 2001 ments and aided by the method of Grzesiek and Bax for the se- Accepted: December 18, 2001 quence-specific assignments [42], the majority of the 1H, 15N, and 13C nuclei were assigned [22]. NOE data for the structure calculation References were derived from two complementary NOESY experiments. A 13C- edited NOESY-HSQC, with spectral widths of 8000, 6400, and 5159 1. Takahashi, K., Hiwada, K., and Kokubu, T. (1986). Isolation and characterisation of a 34000-dalton calmodulin- and F-actin- Hz, and with 1024, 144, and 64 complex data points in F (1H), F 1 2 binding protein from chicken gizzard smooth muscle. Biochem. (1H), and F (13C), respectively, and a mixing time of 100 ms, was 3 Biophys. Res. Commun. 141, 20–26. analyzed to obtain the NOEs arising from all 13C-resolved protons. 2. North, A.J., Gimona, M., Cross, R.A., and Small, J.V. (1994). The recently reported 13C,15N HSQC-NOESY-CH NH experiment [23], 3 Calponin is localised in both the contractile apparatus and the with spectral widths of 8000, 6400, and 2500 Hz, and with 1024, cytoskeleton of smooth muscle cells. J. Cell Sci. 107, 437–444. 192, and 96 complex data points in F (1H), F (1H), and F (13C and 1 2 3 3. Winder, S.J., and Walsh, M.P. (1990). Smooth muscle calponin: 15N), respectively, and a mixing time of 100 ms, was analyzed. This inhibition of the actomyosin MgATPase and regulation by phos- experiment provided the NOE information from the 15N-resolved NH phorylation. J. Biol. Chem. 265, 10148–10155. protons and, simultaneously, the highly resolved 13C-resolved 4. Szymanski, P., and Tao, T. (1993). Interaction between calponin methyl protons. The increased resolution of the 13 C-resolved NOE and smooth muscle myosin. FEBS Lett. 331, 256–259. data from this latter experiment was used to confirm and to augment 5. Fujii, T., Takagi, H., Arimoto, M., Ootani, H., and Ueeda, T. (2000). 13 the assignment of the corresponding crosspeaks in the C-edited Bundle formation of smooth muscle desmin intermediate fila- NOESY-HSQC spectrum. ments by calponin and its binding site on the desmin molecule. A total of 3347 distance restraints were generated from 2367 J. Biochem. 127, 457–465. 13 crosspeaks in the C-edited NOESY-HSQC spectrum and from 980 6. Fujii, T., and Koizumi, Y. (1999). Identification of the binding 15 13 15 N-resolved crosspeaks in the C, N HSQC-NOESY-CH3NH spec- region of basic calponin on ␣ and ␤ tubulins. J. Biochem. 125, trum. Of these, 2529 crosspeaks were fully assigned manually and 869–875. the remaining 818 crosspeaks, unassigned in one or more dimen- 7. Fujii, T., Yamana, K., Ogoma, Y., and Kondo, Y. (1995). Interac- sion, were given assignment possibilities on the basis of their chemi- tion of calponin with phospholipids. J. Biochem. 117, 999–1003. cal shifts using “Connect” from AZARA. The crosspeaks were 8. Menice, C.B., Hulvershorn, J., Adam, L.P., Wang, C.-L.A., and grouped into four categories according to their relative peak intensi- Morgan, K.G. (1997). Calponin and mitogen-activated protein ties of strong, medium, weak, and very weak, and were designated kinase signalling in differentiated vascular smooth muscle. J. the corresponding distance restraint limits of 0–2.7 A˚ , 0–3.3 A˚ , Biol. Chem. 272, 25157–25161. 0–5.0 A˚ , and 0–6 A˚ , respectively. 9. Leinweber, B.D., Leavis, P.C., Grabarek, Z., Wang, C.-L.A., and The structure calculations were carried out using the PARALLHDG Morgan, K.G. (1999). Extracellular regulated kinase (ERK) inter- v5.1 forcefield, with the nonbonded energy function of PROLSQ action with actin and the calponin homology (CH) domain of [43] within the CNS-Solve program [44], modified to allow floating actin-binding proteins. Biochem. J. 344, 117–123. stereochemistry of, and to include active swapping of, prochiral 10. Castresana, J., and Saraste, M. (1995). Does Vav bind to F-actin centers [45]. The initial calculations used every distance restraint, through a CH domain? FEBS Lett. 374, 149–151. including duplicates. The ambiguous distance restraints were then 11. Stradal, T., Kranewitter, W., Winder, S.J., and Gimona, M. (1998). filtered iteratively, based upon the coordinates of an ensemble of CH domains revisited. FEBS Lett. 431, 134–137. refined structures, and redundant restraints (duplicates) were dis- 12. Gimona, M., and Winder, S.J. (1998). Single calponin homology carded (see [46]). Subsequently, a recalibration step was also intro- domains are not actin binding domains. Curr. Biol. 8, R674–R675. duced [46]. The final structure calculation used 2575 NOE-derived 13. Mezgueldi, M., Strasser, P., Fattoum, A., and Gimona, M. (1996). distance restraints of which 1903 were unambiguous (Table 1). The Expressing functional domains of mouse calponin: involvement Structure 258

of the region around alanine 145 in the actomyosin ATPase cium-binding proteins, and tropomyosin. J. Biol. Chem. 270, inhibitory activity of calponin. Biochemistry 35, 3654–3661. 8867–8876. 14. Gimona, M., and Mital, R. (1998). The single CH domain is neither 36. Winder, S.J., Kargacin, G.J., Bonet-Kerrache, A.A., Pato, M.D., and sufficient nor necessary for F-actin binding. J. Cell Sci. 111, Walsh, M.P. (1992). Calponin: localisation and regulation of smooth 1813–1821. muscle actomyosin MgATPase. Jap. J. Pharmacol. 58, 29–34. 15. Mino, T., Yuasa, U., Nakamura, F., Naka, M., and Tanaka, T. 37. Bonet-Kerrache, A., and Mornet, D. (1995). Importance of the (1998). Two distinct actin-binding sites of smooth muscle cal- C-terminal part of actin in interactions with calponin. Biochem. ponin. Eur. J. Biochem. 251, 262–268. Biophys. Res. Commun. 206, 127–132. 16. Gimona, M., and Small, J.V. (1996). Calponin. In Biochemistry 38. Winder, S.J., Allen, B.G., Clement-Chomienne, O., and Walsh, of Smooth Muscle Contraction, M. Ba´ ra´ ny, ed. (New York: Aca- M.P. (1998). Regulation of smooth muscle actin-myosin interac- demic Press), pp. 91–103. tion and force by calponin. Acta Physiol. Scand. 164, 415–426. 17. Winder, S.J., Allen, B.G., Fraser, E.D., Kang, H.-M., Kargacin, 39. Winder, S.J., and Kendrick-Jones, J. (1995). Protein production G.J., and Walsh, M.P. (1993). Calponin phosphorylation in vitro in three different expression vectors from a single PCR product. and in intact muscle. Biochem. J. 296, 827–836. Anal. Biochem. 231, 271–273. 18. Djinovic Carugo, K., Banuelos, S., and Saraste, M. (1997). Crys- 40. Winder, S.J., and Kendrick-Jones, J. (1996). Protein production tal structure of a CH domain. Nat. Struct. Biol. 4, 175–179. in three different expression vectors from a single PCR product 19. Goldsmith, S.C., Pokala, N., Shen, W., Fedorov, A.A., Matsu- (vol. 231, p. 271, 1995). Anal. Biochem. 236, 190. daira, P., and Almo, S.A. (1997). The structure of an actin-cross- 41. Kraulis, P.J., Domaille, P.J., Campbell-Burk, S.L., van Aken, T., linking domain from human fimbrin. Nat. Struct. Biol. 4, 708–712. and Laue, E.D. (1994). Solution structure and dynamics of Ras 20. Keep, N.H., Winder, S.J., Moores, C.A., Walke, S., Norwood, p21 GDP determined by heteronuclear three- and four-dimen- F.M.L., and Kendrick-Jones, J. (1999). Crystal structure of the sional NMR spectroscopy. Biochemisrty 33, 3515–3531. actin-binding region of utrophin reveals a head-to-tail dimer. 42. Grzesiek, S., and Bax, A. (1993). Amino-acid type determination Structure 7, 1539–1546. in the sequential assignment procedure of uniformly 13C/15N- enriched proteins. J. Biomol. NMR 3, 185–204. 21. Norwood, F.L.M., Sutherland-Smith, A.J., Keep, N.H., and Ken- 43. Linge, J.P., and Nilges, M. (1999). Influence of non-bonded pa- drick-Jones, J. (2000). Crystal structure of the dystrophin actin- rameters on the quality of NMR structures: a new force field for binding domain. Structure 8, 481–491. NMR structure calculation. J. Biomol. NMR 13, 51–54. 22. Bramham, J., Smith, B.O., Uhrı´n, D., Barlow, P.N., and Winder, 44. Bru¨ nger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, S.J. (2001). 1H, 15N and 13C resonance assignments of the N P., Grosse-Kunstleve, R.W., Jiang, J.-S., Kuszewski, J., Nilges, terminal region of calponin. J. Biomol. NMR 19, 189–190. M., Pannu, N.S., et al. (1998). Crystallography and NMR system 23. Uhrı´n, D., Bramham, J., Winder, S.J., and Barlow, P.N. (2000). (CNS): a new software system for macromolecular structure Simultaneous CT-C-13 and VT-N-15 chemical shift labeling: ap- determination. Acta Crystallogr. D54, 905–921. plication to 3D NOESY-CH3NH and 3D C-13, N-15 HSQC- 45. Wakefield, R.I., Smith, B.O., Nan, X., Free, A., Soteriou, A., Uhrin, NOESY-CH3NH. J. Biomol. NMR 18, 253–259. D., Bird, A.P., and Barlow, P.N. (1999). The solution structure 24. Hutchinson, E.G., and Thornton, J.M. (1996). PROMOTIF—a of the domain from MeCP2 that binds to methylated DNA. J. program to identify and analyse structural motifs in proteins. Mol. Biol. 291, 1055–1065. Protein Sci. 5, 212–220. 46. Nilges, M., Macias, M.J., O’Donaghue, S.I., and Oschkinat, H. 25. Olski, T.M., Noegel, A.A., and Korenbaum, A. (2001). Parvin, a (1997). Automated NOESY interpretation with ambiguous dis- 42 kDa focal adhesion protein, related to the ␣-actinin superfam- tance restraints: the refined NMR solution structure of the pleck- ily. J. Cell Sci. 114, 525–538. strin homology domain from beta-spectrin. J. Mol. Biol. 269, 26. Levine, B.A., Moir, A.J.G., Patchell, V.B., and Perry, S.V. (1990). 408–422. The interaction of actin with dystrophin. FEBS Lett. 263, 47. Laskowski, R.A., MacArthur, M.W., Moss, D.S., and Thornton, 159–162. J.M. (1993). PROCHECK: a program to check the stereochemi- 27. Levine, B.A., Moir, A.J.G., Patchell, V.B., and Perry, S.V. (1992). cal quality of protein structures. J. Appl. Crystallogr. 26, Binding sites involved in the interaction of actin with the 283–291. N-terminal region of dystrophin. FEBS Lett. 298, 44–48. 48. Wishart, D.S., and Sykes, B.D. (1994). The 13C Chemical-Shift 28. Corrado, K., Mills, P.L., and Chamberkain, J.S. (1994). Deletion Index: a simple method for the identification of protein second- analysis of the dystrophin-actin binding domain. FEBS Lett. 344, ary structure using 13C chemical shift data. J. Biomol. NMR 4, 255–260. 171–180. 29. Winder, S.J., Hemmings, L., Maciver, S.K., Bolton, S.J., Tinsley, 49. Owen, C.H., Morgan, D.G., and DeRosier, D.J. (1996). Image J.M., Davies, K.E., Critchley, D.R., and Kendrick-Jones, J. analysis of helical objects: the Brandeis Helical Package. J. (1995). Utrophin actin binding domain: analysis of actin binding Struct. Biol. 116, 167–175. and cellular targeting. J. Cell Sci. 108, 63–71. 50. Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991). 30. Mezgueldi, M., Fattoum, A., Derancourt, J., and Kassab, R. Improved methods for binding protein models in electron den- (1992). Mapping of the functional domains in the amino-terminal sity maps and the location of errors in these models. Acta Crys- region of calponin. J. Biol. Chem. 267, 15943–15951. tallogr. A47, 110–119. 31. Hodgkinson J.L., EL-Mezgueldi, M., Craig, R., Vibert, P., Mar- 51. Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both ston, S.B., and Lehman, W. (1997). 3D image reconstruction of detailed and schematic plots of protein structures. J. Appl. Crys- reconstituted smooth muscle thin filaments containing cal- tallogr. 24, 946–950. ponin: visualisation of interactions between F-actin and cal- 52. Merritt, E.A., and Bacon, D.J. (1997). Raster3D photorealistic ponin. J. Mol. Biol. 273, 150–159. molecular graphics. Methods Enzymol. 277, 505–524. 32. Hanein, D., Volkmann, N., Goldsmith, S., Michon, A.M., Lehman, 53. Lorenz, M., Popp, D., and Holmes, K.C. (1993). Refinement of W., Craig, R., DeRosier, D., Almo, S., and Matsudaira, P. (1998). the F-actin model against X-ray fiber diffraction data by the use An atomic model for the interaction of the calponin-homology of a directed mutation algorithm. J. Mol. Biol. 234, 826–836. domains and F-actin: implications for filament crosslinking and regulation. Nat. Struct. Biol. 5, 787–792. Accession Numbers 33. Hanein, D., Matsudaira, P., and DeRosier, D.J. (1997). Evidence for a conformational change in actin induced by fimbrin (N375) The chemical shifts of 1H, 15N, and 13C have been deposited at the binding. J. Cell Biol. 139, 387–396. BioMagResBank database (http:\\www.brmb.wisc.edu) under acces- 34. Moores, C.A., Keep, N.H., and Kendrick-Jones, J. (2000). Struc- sion number 4880. Atomic coordinates for the final set of 20 refined ture of utrophin actin binding domain bound to F-actin reveals structures have been submitted to the Protein Data Bank under binding by an induced fit mechanism. J. Mol. Biol. 297, 465–480. accession code 1h67. 35. Mezgueldi, M., Mendre, C., Calas, B., Kassab, R., and Fattoum, A. (1995). Characterisation of the regulatory domain of gizzard calponin. Interactions of the 145–163 region with F-actin, cal-