3-D image reconstruction of reconstituted smooth muscle thin filaments containing calponin: Visualization of interactions between F-actin and Calponin1

J. Mol. Biol. (1997) 273, 150±159

3-D Image Reconstruction of Reconstituted Smooth Muscle Thin Filaments Containing Calponin: Visualization of Interactions between F-actin and Calponin J. L. Hodgkinson1*, M. EL-Mezgueldi1, R. Craig2, P. Vibert3 S. B. Marston1 and W. Lehman4 1

Imperial College School of Medicine, National Heart and Lung Institute, Dovehouse Street, London SW3 6LY, UK 2

Department of Cell Biology University of Massachusetts Medical School, 55 Lake Avenue North, Worcester MA 01655, USA 3

Rosenstiel Basic Medical Sciences Research Center Brandeis University, Waltham MA 02254, USA 4

Department of Physiology Boston University School of Medicine, 80 East Concord Street, Boston, MA 02118 USA

Calponin is a putative thin ®lament regulatory protein of smooth muscle that inhibits actomyosin ATPase in vitro. We have used electron microscopy and three-dimensional reconstruction to elucidate the structural organization of calponin on actin and actin-tropomyosin ®laments. Calponin density was clearly delineated in the reconstructions and found to occur peripherally along the long-pitch actin-helix. The main calponin mass was located over sub-domain 2 of actin, and connected axially adjacent actin monomers by binding to the ``upper'' and ``lower'' edges of sub-domains 1 of each actin. When the reconstructions were ®tted to the atomic model of F-actin, calponin appeared to contact actin near the N terminus and at residues 349 to 352 close to the C terminus of subdomain 1 on one monomer. It also touched residues 92 to 95 of subdomain 1 on the axially neighboring actin and continued up the side of this monomer as far as residues 43 to 48 of sub-domain 2. These positions are consensus binding sites for a number of actin-associated proteins and are also near to sites of weak myosin interaction. Calponin did not appear to block strong myosin binding sites on actin. In contrast to the calponin mass which appeared monomeric in reconstructions, tropomyosin formed a continuous strand of added density along F-actin. When added to tropomyosin-containing ®laments, calponin caused a shift of tropomyosin away from sub-domain 1 towards sub-domain 3 of actin, exposing strong myosin-binding sites that were previously covered by tropomyosin. This structural effect is unlike that of troponin and therefore inhibition of actomyosin ATPase by calponin and troponin cannot be strictly analogous. The location of calponin would allow it to directly compete or interact with a number of actin-binding proteins. # 1997 Academic Press Limited

*Corresponding author

Keywords: actin; a-actinin; calponin homology (CH)-domain; ®mbrin; tropomyosin

Introduction

Dedicated to the memory of our friend and colleague Fredric S. Fay. Present address: P. Vibert, Biology Department, Brookhaven National Laboratory, Upton, NY 11973, USA. Abbreviations used: CH-domain, calponin homology domain; em, electron microscopy; F-actin, ®lamentous actin; SD, standard deviation. 0022±2836/97/410150±10 $25.00/0/mb971307

Vertebrate smooth muscle contractility is regulated primarily via phosphorylation of the myosin ®laments (see Kamm & Stull, 1985; Sellers & Adelstein, 1986, for reviews). Fine-tuning of this regulation by the actin-binding proteins, calponin and caldesmon, present in smooth muscles but absent from skeletal muscle, has also been proposed (see Sellers & Sobue, 1991; Chalovich, 1992; Winder & Walsh, 1993; Dabrowska, 1994; ELMezgueldi, 1996; Gimona & Small, 1996; Lehman # 1997 Academic Press Limited

F-actin ±Calponin Structure

et al., 1996; Marston & Huber, 1996 for reviews). Both caldesmon and calponin have been investigated by biochemical, immunolocalization and physiological studies, and have been shown to be capable of inhibiting actomyosin ATPase (ibid.) and ®lament sliding (Okagaki et al., 1991; Haeberle et al., 1992; Shirinsky et al., 1992; Fraser & Marston, 1995) in vitro. However, the differing conclusions that have been reached from these studies have not so far produced complete agreement on the roles played by either protein in smooth muscle contraction in vivo. In vitro biochemical data on the inhibitory properties of calponin have been quite variable. Although the content of calponin in vivo is considerable (one calponin: seven actin monomers; Lehman & Kaminer, 1984; Takahashi et al., 1986; Lehman, 1991), most recent biochemical studies suggest that still higher levels of calponin are needed to inhibit actomyosin ATPase (see Gimona & Small, 1996) and that this inhibition may be low at physiological salt concentrations under in vitro conditions (EL-Mezgueldi et al., 1996). In addition, whereas calponin is approximately equimolar with tropomyosin in vivo, inhibition of actomyosin ATPase by calponin is independent of tropomyosin in vitro (Winder & Walsh, 1990). The uncertainties about calponin function are underscored by contradictory results concerning proposed mechanisms for releasing the ATPase inhibition (Winder & Walsh, 1990; BaÂraÂny et al., 1991; BaÂraÂny & BaÂraÂny, 1993). Despite these uncertainties, however, recent observations on skinned smooth muscle cell preparations suggest that removal of calponin (Malmquist et al., 1997) or exogenous addition of

151 calponin or calponin peptides can greatly in¯uence contractility (Itoh et al., 1994; Horowitz et al., 1996; Obara et al., 1996). Several studies indicate that calponin may not be homogeneously distributed throughout smooth muscle cells. Immunocytochemical studies by North et al. (1994a,b), supported by Mabuchi et al. (1996), suggest that calponin is mainly restricted to b-actin in the cytoskeletal domain of smooth muscle cells. It is either absent, or present in reduced amounts, in contractile domains, regions replete with caldesmon-containing a- or g-actin localized near myosin ®laments. These results are consistent with in vitro studies in which two classes of thin ®laments could be isolated immunologically (Lehman, 1991). If calponin is con®ned mainly to the smooth muscle cytoskeleton, it is dif®cult to envision how it could modulate contractility directly, although it could play a role in the organization of the cytoskeleton. However, other studies suggest that calponin is more uniformly distributed (Walsh et al., 1993). The structural interactions between calponin and F-actin have not been well characterized. Hydrodynamic measurements and electron microscopy of puri®ed calponin indicate that it is approximately 16 nm long with a diameter of about 2.6 nm (Stafford et al., 1995). However, little is known about the structural organization and binding of calponin along thin ®laments. Here, we have used electron microscopy and three-dimensional reconstruction of F-actin ± calponin complexes to determine the contacts made by calponin on thin ®laments. Fitting these reconstructions to the atomic model of F-actin, we have identi®ed clusters of

Figure 1. Electron micrographs of negatively stained F-actin and F-actin ± calponin complexes. a, Actin alone; note well de®ned, helically arranged subunits. b, Actin saturated with calponin; here ®laments appear to be larger in diameter than in a and there is a background of unbound calponin accounting for the staining differences in the two samples. Scale bar represents 50 nm.

152

F-actin±Calponin Structure

amino acids on actin involved in the interaction. We show that calponin does not directly cover strong myosin binding sites on actin, but could impede the approach of myosin to those sites. We also show that it binds at a site on actin to which several other actin-binding proteins also bind. Possible functional roles for calponin are discussed in light of these and previous results.

Results F-actin ®laments were complexed with calponin under near saturating conditions to optimize detection and visualization of the protein. At low ionic strength, the resulting ®laments tended to aggregate into large bundles which were unsuited for analysis. Bundle formation was reduced by raising the ionic strength to 100 mM KCl (Figure 1b), under which conditions quantitative SDS-PAGE of sedimented actin± calponin ®laments indicated almost equimolar binding (1.5 actin: 1 calponin, mol:mol). Binding of calponin was also evident from electron micrographs of negatively stained ®laments, whose diameters were wider than those of F-actin or F-actin ± tropomyosin controls (Figure 1). Density maps of F-actin ± calponin and F-actin ± tropomyosin± calponin were calculated from the averaged Fourier transform layer-line data obtained from the negatively stained ®laments. When compared to similar maps generated from pure F-actin or F-actin± tropomyosin controls, reconstructions of ®laments containing calponin showed the presence of considerable extra density (Figures 2 and 3). The extra density appeared as a compact mass on the outer domain of actin, contacting sub-domain 1 of one actin on its ``upper'' side, bridging over the shallow aspect of subdomain 2, and connecting to sub-domain 1 of the next actin along the long pitch helix on its ``lower'' side (Figures 2 and 4). There was no evidence for signi®cant density running longitudinally and continuously along the ®lament that might have been due to calponin lying in an extended con®guration, although narrow extended portions of calponin may not have been resolved. The position and size of the extra density accounted in part for the increased diameter of the negatively stained ®laments (Figures 2 and 3). However, it seemed not to account for all of the calponin mass, as the size of calponin in reconstructions was similar only to that of a single actin sub-domain (Figures 2 b,d and 3), which is much smaller than would be expected from calponin's molecular mass (compare the size of 30 kDa calponin density to that of a 42 kDa actin monomer). The relatively low calponin density may have been a result of disorder of parts of the molecule or of partial occupancy on F-actin. In addition, our reconstructions were contoured to approximate the predicted volume of actin, and insigni®cant low

Figure 2. Surface views of reconstructions of: a, F-actin, b, F-actin-calponin, c, F-actin-tropomyosin, d, F-actintropomyosin-calponin. Characteristic helically arranged, bilobed actin monomers are seen in the F-actin control (a), and these are partially obscured by the addition of calponin and/or tropomyosin. Note the extra density arising from calponin, which bridges over sub-domain 2 of actin in b and d (marked by asterisks). Tropomyosin strands (marked by arrows) are seen in c and d following the long-pitch actin-helix. Note the different positions of tropomyosin in c and d. In the absence of calponin, tropomyosin adopts a position on the outer domain of actin (Ao); addition of calponin causes a shift onto the inner domain of actin (Ai). Maps in a to d were calculated from the average amplitudes and phases along layer-lines of Fourier transforms of 18, 12, 15 and 7 ®laments, respectively. Phase residuals (  SD), a measure of the accuracy of the alignment between corresponding layer-line data sets and the average for the data, were 46.9(5.5) , 55.9(4.8) , 46.4(4.9) , 55.1(5.8) . The up-down phase residuals (
 SD), a measure of ®lament polarity comparing phase residuals between data sets and a reference average where the reference is ®rst pointed in one and then in the opposite direction, were 28.6(5.9) , 21.3(5.7) , 34.5(5.6) , 26.2(6.0) . Layer-line data extended to a resolution of 1/2.5 to 3.0 nm and no data were collected beyond 1/ 2.0 nm.

density edges of actin were omitted. The effect of this is to partially compromise visualization of the calponin density, which is localized at the edge of actin and is also of low density. To be able to separate overlapping low density edges of actin and calponin, difference maps were generated by subtracting densities associated with pure F-actin from those in F-actin± calponin complexes, thereby isolating the calponin contribution. Difference density maps constructed by comparing F-actin ± calponin and F-actin clearly delineated the boundaries of the calponin density and con®rmed the location suggested above (Figures 3 and 4). The calponin density in such maps is augmented (Figure 3 g and h) and is signi®cant at con®dence levels greater than 99.95% (Figure 3i).

F-actin ±Calponin Structure

Figure 3. a,c,e, Transverse sections (``z-sections'') through maps of three-dimensional reconstructions. Since adjacent actin monomers on either side of the ®lament axis are staggered, sectioning through the center of sub-domains 1 and 3 of one monomer will result in sectioning through sub-domains 2 and 4 of the other monomer. a, F-actin (sub-domains 1 to 4 are labelled 1,2,3,4, respectively). c, F-actin ± calponin; note the extra density contributed by calponin and associated with sub-domain 2 (marked by bold arrow); also note the bending of subdomain 1 towards sub-domain 3 by comparing this map to that of pure F-actin in a. e, F-actin ± tropomyosin ± calponin; comparable extra density noted in c is again evident (marked by bold arrow) as well as density associated with tropomyosin strands on sub-domains 3 and 4 (marked by narrow arrows). b,d,f, Maps showing the signi®cance of the contributing densities in sections a,c,e, respectively, and indicating those that are signi®cantly different from zero at con®dence levels greater than 99%. Each map pair (a,b; c,d; e,f) shows a near-perfect ®t demonstrating the reliability of the data shown in a,c,e. g, Difference density attributed to calponin, created by subtracting densities associated with map a from those in map c. The major difference is associated with the mass of calponin (marked by bold arrow). Smaller differences near the surface of actin close to the junctions

153 Besides the additional mass contributed by calponin, difference maps revealed changes in the structure of actin itself. These were located at sites on actin distant from the calponin binding site and were a result of a bending of sub-domain 1 towards sub-domain 3 in the direction of the central axis of the ®lament (compare Figure 3a and c). This apparent conformational change is evident from the signi®cant positive difference densities associated with actin per se observed in the difference maps discussed above (Figure 3g,i). The positive density changes in actin, a manifestation of the new positions adopted by its sub-domains, were detected on the back side of sub-domain 1 of each actin monomer and to a lesser extent near the junction between sub-domain 2 and sub-domain 3 of neighboring monomers in the genetic helix (Figures 3 g and h). All changes in actin structure were signi®cant at a con®dence level of 99.5%; many were signi®cant at a level greater than 99.95% (Figure 3i). The position of the calponin density suggested that it might affect tropomyosin binding to the outer domain of actin observed in actin-tropomyosin controls (Figure 2; cf. Hodgkinson et al., 1997). We therefore added calponin to reconstituted actin-tropomyosin ®laments. Comparison of actintropomyosin-calponin maps with actin-tropomyosin controls showed that calponin caused a shift in position of the smooth muscle tropomyosin from its position on the outer domain of actin to one on the inner domain (compare Figure 2c and d). The position and shape of calponin in these ®laments were similar to those in the ®laments lacking tropomyosin, and the presence of calponin caused the same conformational change in actin as above (data not shown). The location of the shield-like calponin density over sub-domain 2 of actin was best visualized by superimposing the three-dimensional reconstruction on the atomic model of F-actin (yellow in Figure 4) and highlighting the calponin difference density determined above (magenta in Figure 4). It was clear from this ®tting that calponin most closely approached amino acid residues associated with the N terminus and a loop (residues 349 to 352) projecting from the C terminus of sub-domain 1 on one actin monomer, and also amino acid residues 92 to 95 of sub-domain 1 on the axially neighboring actin and those continuing up the side of this monomer to as far as residues 43 to 48 of sub-

between adjacent actin monomers along the genetic actin-helix are also noted (cf. Figure 4). h, Difference densities (®lled black) overlaid on the actin map, showing position of calponin (marked by bold arrow) and the effect of calponin on actin structure. The bending of actin sub-domains results in a positive difference density on the back side of sub-domain 1 and at the junction of subdomains 2 and 3 (h). i, Densities in map g that are statistically signi®cant at a con®dence level greater than 99.5% (Student's t-test).

154 domain 2 (Figure 4c). We saw no evidence that calponin covers regions of actin thought to be involved in strong myosin binding (Rayment et al., 1993). Instead, calponin appeared to be positioned directly in contact with or over sites of proposed weak actin-myosin interactions. We attempted to dock correspondingly scaled models of S-1 onto the exposed strong binding site in our actin-calponin model using the program O. We found that there was a partial collision between the edges of calponin and S-1 mass distal to the surface of actin. The conformational changes in actin brought about by addition of calponin and noted above were also localized by superimposition of positive difference densities on the molecular model of F-actin (Figure 4). They can be most easily visualized by viewing the back-side of one monomer of the model (Figure 4d). The bending of sub-domain 1 results in an increase in mass distributed over C-terminal amino acids in the model (red in Figure 4d). The corresponding bending of subdomain 3 was not detected at the contour levels chosen.

Discussion We have localized the binding site of calponin on the periphery of F-actin using electron microscopy and three-dimensional reconstruction, and have resolved and visualized the regularly associating actin-binding segment of calponin itself. Calponin forms a shield-like density over sub-domain 2 of each actin monomer and connects adjacent actin monomers along the bottom and the top edges of respective sub-domains 1. Our results are consistent with earlier crosslinking studies indicating that calponin-binding is localized near the C terminus of actin on sub-domain 1 (Mezgueldi et al., 1992; Bonet-Kerrache & Mornet, 1995). The appearance of the calponin density and its location on F-actin are extremely similar to those of the microvillar actin-crosslinking protein, ®mbrin, determined independently by cryo-electron microscopy and image analysis (Hanein et al., 1997). Interestingly, N375, the actin-binding region of ®mbrin, contains a pair of 12 kDa domains each with amino acid sequences homologous to that of the N-terminal end (residues 28 to 142) of calponin (Castresana & Saraste, 1995; Hanein et al., 1997); one of the so-called calponin homology (CH)domains in ®mbrin contacts actin and the second, even though lying in tandem with the ®rst, does not (Hanein et al., 1997). Since calponin itself contains only one CH-domain (Castresana & Saraste, 1995), it was crucial to compare the structural data on ®mbrin and calponin systematically for similarities and differences. Fitting corresponding maps of the F-actin ±®mbrin and F-actin± calponin complexes to each other (Hanein & W. L., unpublished observations) demonstrated that calponin and the CH-domain of ®mbrin lysing closest to actin superimpose, contact actin at the same site and are

F-actin±Calponin Structure

oriented identically. Hence, the densities associated with calponin in our reconstructions essentially represent the CH-domain of the protein, i.e. approximately 40 to 50% of the whole molecule. The remaining protion of the protein presumably is either disordered or unstructured. Biochemical studies show that both a-actinin and gelsolin also interact with actin at the same sites as calponin and ®mbrin, and share both sequence and structural homology with them (Castresana & Saraste, 1995; McGough et al., 1994; McGough & Way 1995; Hanein et al., 1997; cf. Matsudaira, 1991). Calponin, caldesmon and troponin-I compete with each other for actin-binding (Makusch et al., 1991; Szymanski et al., 1997), consistent with the proposition that calponin shares binding sites common to other actin-associated proteins. These observations indicate that one possible function of calponin could be to block the binding of these or related proteins on actin by competition. For example, inhibition of gelsolinlike severing by competitive binding might serve to maintain the integrity of the smooth muscle cytoskeleton. Alternatively, competition or interactions with a-actinin or with ®lamin might be necessary for building or remodeling the smooth muscle cytoskeleton and therefore adjusting its mechanical properties. The inhibition of actomyosin ATPase by calponin observed in vitro may result from either direct blocking of myosin binding (EL-Mezgueldi & Marston, 1996) or a conformational change in Factin (Noda et al., 1992; Borovikov et al., 1996a,b), and it is known to be tropomyosin-independent (Winder & Walsh, 1990). We have shown that calponin cannot regulate actomyosin ATPase by a mechanism comparable to that of troponin (cf. Winder & Walsh, 1993; EL-Mezgueldi & Marston, 1996), since calponin, unlike troponin, does not ®x tropomyosin over sites of strong myosin binding. In fact, calponin, like caldesmon (Vibert et al., 1993; Hodgkinson et al., 1997) causes tropomyosin to move away from these sites while not appearing to occupy them itself. Therefore any possible mechanism of regulation of actomyosin ATPase by calponin would have to differ qualitatively from the steric-blocking mechanism operating in striated muscle. Biochemical studies suggest that inhibition of actomyosin ATPase by calponin in vitro would be most effective if actin were saturated with the protein (Gimona & Small, 1996, and references within). Given the position we have determined for calponin, it could regulate actin± myosin interaction by blocking weak myosin binding sites on actin considered to be necessary for initiating actomyosin ATPase (Rayment et al., 1993; Miller et al., 1995). It could also obstruct access of myosin heads to sites of strong myosin binding (even if these sites were themselves not covered), and thus prevent interactions necessary for ATPase-linked force-transduction. However, the stoichiometry and distribution of calponin on smooth muscle

155

F-actin ±Calponin Structure

thin ®laments is not well established and we therefore cannot make de®nitive judgments about such mechanisms. We also cannot rule out the possibility that calponin could block strong sites of

myosin binding directly, as suggested by others (EL-Mezgueldi & Marston, 1996; Borovikov et al., 1996a,b), since we have not resolved the entire calponin molecule. It should be noted, in addition,

Figure 4 (legend on page 156)

156 that calponin and its homolog, ®mbrin, cause a virtually identical conformational change in F-actin, affecting the same cluster of amino acids on subdomain 1 (cf. Hanein et al., 1997). At this point, we cannot be certain of the consequences of this on actomyosin ATPase, and can only reiterate the conclusions of Hanein and colleagues (Hanein et al., 1997) that this structural change may be a passive consequence of protein binding but could also profoundly in¯uence actin-ligand binding. The importance of these apparent changes, whether they are propagated along F-actin cooperatively (cf. Egelman and Orlova, 1995), and the levels of calponin required to effect them deserve further study. Different actin isoforms are highly conserved (we used skeletal muscle a-actin in our studies); isoform variability, in fact, occurs mainly at the acidic N-terminal six amino acids of actin (Vanderkerckhove & Weber, 1978; Lehman et al., 1996). What determines the segregation of different actin ®laments and the apparent targeting of calponin to b-actin is unknown. Given that the best estimates of the molar ratio of calponin to total smooth muscle cellular actin are approximately 1:7 (Lehman & Kaminer, 1984; Takahashi et al., 1986; Lehman, 1991) and that cytoskeletal b-actin comprises roughly 10% of total cellular actin (Vandekerckhove & Weber, 1978), calponin, if primarily associated with this isoform (North et al., 1994a,b; Mabuchi et al., 1996), should saturate it. If so, the presence of calponin might limit the interaction of myosin with cytoskeletal bactin, and therefore favor its association with either a- or g-contractile actin. It should be recognized that other proteins with calponin-homology domains are typically cytoskeletal in nature, and some link actin to intermediate ®laments (Yang et al., 1996). Such linkage also seems plausible in the cytoskeletal domain of smooth muscle, which

F-actin±Calponin Structure

contains cytoskeletal b-actin, intermediate ®laments, and probably most of the calponin (North et al., 1994a,b; Mabuchi et al., 1997). Recent studies suggest that calponin may indeed bind to intermediate ®laments as well as actin ®laments (Wang & Gusev, 1996; Mabuchi et al., 1997). If calponin is present in the contractile domain as well as the cytoskeletal domain (Walsh et al., 1993; North et al., 1994b), then calponin might equally act as an intracellular strut linking intermediate ®laments and contractile actin. If these interpretations are correct, then the inhibition of contractility of intact smooth muscle preparations by exogenously added calponin could be due to artifactually introduced constraints being placed on previously switched-on contractile thin ®laments (Itoh et al., 1994; Obara et al., 1996). Conversely, experimental removal (Malmqvist et al., 1997) or inactivation (Horowitz et al., 1996) of calponin in cytoskeletal domains may facilitate myosin interactions with previously constitutively switched-off ®laments. The hypotheses above assume that calponin± actin interactions are inhibitory, static and largely con®ned to a discrete smooth muscle cytoskeleton. However, little is known about in vivo calponin ± actin interactions and whether such interactions can be modulated. Calponin may, in fact, redistribute within the cell in response to stimuli (Parker et al., 1994). Calponin, particularly if found in the contractile domain, thus may play a more dynamic role than suggested above, for example, in the control of crossbridge cycling by dephosphorylated myosin, thereby regulating tension maintenance during the latch phase (Malmqvist et al., 1997). As with other proteins with CH-domains, calponin itself may be involved in modelling the cytoskeleton or in regulating contraction and/or cellular signal transduction pathways.

Figure 4. Fitting calponin densities onto the atomic model of F-actin (Lorenz et al., 1993). a, Calponin densities (magenta), derived from the difference densities between F-actin and F-actin ± calponin, were ®rst superimposed on the envelope formed by ®ve monomers of the F-actin em-map (blue). b, The atomic model of F-actin (depicted as a yellow a-carbon chain, Lorenz et al., 1993) was then ®tted to the reconstructed densities and substituted for the actin em-envelope, showing the position of calponin (magenta) on the periphery of the actin. Contour levels in a and b were chosen so that calponin densities and more axially located densities attributed to actin conformational changes did not superimpose with each other. The polarities of the actin monomers here are the same as in Figure 2. c, An enlargement of the area surrounding the calponin midway between two actin monomers on the right side of Figure 4b showing a more detailed view of the actin ± calponin interface. Calponin appears to be closely associated with the N terminus and a loop projecting from the C terminus (residues 349 to 352) of one actin monomer and with residues 92 to 95 and those extending to residues 43 to 48 of the actin monomer beneath it. These sites appear to coincide with those used by myosin to make weak initial contacts on actin. All sites on actin thought to be involved in strong myosin binding appear to remain exposed in the maps generated. Amino acid residues mentioned above are numbered. Small magenta densities closer to the ®lament axis seen in a and b represent difference densities attributed to changes in actin conformation. To localize these in more detail, a single monomer in b was rotated to show the back-side of the molecule where the changes occurred (d); here the upper actin monomer in Figure 4c was rotated clockwise about its vertical axis by 40 ), calponin was removed from the Figure photographically. Positive difference densities (red) were centered approximately over residue 361 of actin near the actin C terminus (residue 375). This positive density is a manifestation of the new positions occupied by actin sub-domain 1 following the calponin-induced conformational change. For orientation, an asterisk indicates the center of sub-domain 2.

157

F-actin ±Calponin Structure

Materials and Methods Protein preparation Skeletal muscle F-actin (a-actin) was prepared from rabbit back muscle using the Drabikowski & Gergely (1964) modi®cation of the original Straub (1942) method, and smooth muscle tropomyosin was prepared from sheep aorta by the Eisenberg & Kielley (1974) adaptation of the Bailey (1948) protocol. Calponin was prepared from chicken gizzard muscle as described by Takahashi et al. (1986) and Mezgueldi et al. (1996). Protein purity was assessed and ®lament composition quanti®ed by sedimentation and SDS-PAGE of protein pellets, as previously described (Hodgkinson et al., 1997). Electron microscopy and helical reconstruction Thin ®laments were reconstituted by combining 11 mM F-actin in the presence and absence of 2.5 mM tropomyosin and/or 11 mM calponin in a buffer consisting of 100 mM KCl, 5 mM MgCl2, 5 mM Pipes-dipotassium salt, 10 mM NaN3, 1 mM DTT (pH 7.0). Samples were left to incubate for 20 minutes at room temperature before rapid dilution 12.5-fold with additional buffer containing 11 mM calponin. Excess calponin in the dilution buffer was used to prevent any possible dissociation of the protein from actin at the low protein concentrations used for electron microscopy. Diluted samples were immediately applied to carbon-coated electron microscope grids and negatively stained with 1% (w/v) uranyl acetate. Electron micrograph images were recorded at 60,000 magni®cation under low dose conÊ 2 ) on a Philips CM120 electron microditions (12eÿ/A scope. Micrographs were digitized on an Eikonix model 1412 CCD camera at a pixel size corresponding to 0.67 nm in the ®laments (Vibert et al., 1993). Regions of ®laments suitable for helical reconstruction were selected on the basis of ®lament straightness and lack of aggregation, uniformity of staining, and freedom from astigmatism. Slightly curved ®laments were straightened by ®tting a cubic spline and then re-interpolating the image (Egelman, 1986). Helical reconstruction was carried out using standard methods (DeRosier & Moore, 1970; Amos & Klug, 1975; Owen et al., 1996) as described previously (Vibert et al., 1993, 1997). Statistical analysis was carried out by point by point comparison using a Student's t-test (Milligan & Flicker, 1987; Trachtenberg & DeRosier, 1987). Fitting of the a-carbon chain of the atomic resolution F-actin model (Lorenz et al., 1993) to our reconstructions was carried out as described by McGough et al. (1994) using the program O (Jones et al., 1991).

Acknowledgements This work was funded by a Wellcome Trust Career Development (045449/z/95/z) fellowship to J. L. H., NIH grants awarded to W. L. (HL36153) and F. S. Fay (Program Project grant HL47530), a NSF grant to P. V. (MCB04746), a Wellcome Trust grant to M. E-M. (040937/115) and a NIH Shared Instrumentation grant to R. C. (RR08426). We are especially grateful to Drs Dorit Hanein and David DeRosier for sharing their data and thoughts on ®brim ±actin interactions. We are also indebted to Dr Hanein for ®tting her maps of F-actin ± ®mbrin to the maps presented here. We also thank Ms

Marie Picard Craig for arranging Figures 2 and 4 and Mr Norberto Gherbesi for photographic work. We were privileged to have been colleagues of Dr Fredric Fay, who encouraged us to carry out these studies, and who was an inspiration to us all.

References Amos, L. & Klug, A. (1975). Three-dimensional image reconstructions of the contractile tail of T4 bacteriophage. J. Mol. Biol. 99, 51± 73. Bailey, K. (1948). Tropomyosin: a new asymmetric protein component of the muscle ®bril. Biochem. J. 43, 271± 279. BaÂraÂny, M. & BaÂraÂny, K. (1993). Calponin phosphorylation does not accompany contraction of various smooth muscles. Biochim. Biophys. Acta, 229± 233. BaÂraÂny, M., Rokolya, A. & BaÂraÂny, K. (1991). Absence of calponin phosphorylation in contracting or resting arterial smooth muscle. FEBS Letters, 279, 65± 68. Bonet-Kerrache, A. & Mornet, D. (1995). Importance of the C-terminal part of actin in interactions with calponin. Biochem. Biophys. Res. Commun. 206, 127 ± 132. Borovikov, Y. S., Khoroshev, M. I. & Chacko, S. (1996a). Comparison of the effects of calponin and a 38 kDa caldesmon fragment on formation of the strongbinding state in ghost muscle ®bers. Biochem. Biophys. Res. Commun. 223, 240± 244. Borovikov, Y. S., Horiuchi, K. Y., Avrova, S. V. & Chacko, S. (1996b). Modulation of actin conformation and inhibition of actin ®lament velocity by calponin. Biochemistry, 35, 13849± 13857. Castresana, J. & Saraste, M. (1995). Does Vav bind to Factin through a CH domain?. FEBS Letters, 374, 149 ± 151. Chalovich, J. (1992). Actin mediated regulation of muscle contraction. Pharmacol. Ther. 55, 95± 148. Dabrowska, R. (1994). Actin and thin-®lament-associated proteins in smooth muscle. In Airways Smooth Muscle: Biochemical Control of Contraction and Relaxation (Raeburn, D. & Giembycz, M. A., eds), pp. 31 ± 59, Birkhauser Verlag, Basel, Switzerland. DeRosier, D. J. & Moore, P. B. (1970). Reconstruction of three-dimensional images from electron micrographs of structures with helical symmetry. J. Mol. Biol. 2, 355±369. Drabikowski, W. & Gergely, J. (1964). The effect of the temperature of extraction and of tropomyosin on the viscosity of actin. In Biochemistry of Muscle Contraction (Gergely, J., ed.), pp. 125± 131, Little, Brown and Co., Boston, USA. Egelman, E. (1986). An algorithm for straightening images of curved ®lamentous structures. Ultramicroscopy, 19, 367± 374. Egelman, E. H. & Orlova, A. (1995). Allostery, cooperativity, and different structural states in F-actin. J. Struct. Biol. 115, 159± 162. Eisenberg, E. & Kielley, W. W. (1974). Troponin-tropomyosin complex. Column chromatographic separation activity of the three active troponin components with and without tropomyosin present. J. Biol. Chem. 249, 4742±4748. EL-Mezgueldi, M. (1992). Calponin. Int. J. Biochem. Cell Biol. 28, 1185± 1189. EL-Mezgueldi, M. & Marston, S. B. (1996). The effects of smooth muscle calponin on the strong and weak

158 myosin binding sites of F-actin. J. Biol. Chem. 271, 28161± 28167. Fraser, I. D. C. & Marston, S. B. (1995). In vitro motility analysis of smooth muscle caldesmon control of actin-tropomyosin ®lament movement. J. Biol. Chem. 270, 19688± 19693. Gimona, M. & Small, J. V. (1996). Calponin. In Biochemistry of Smooth Muscle Contraction (BaÂraÂny, M., ed.), pp. 91± 103, Academic Press, San Diego, USA. Haeberle, J. R., Trybus, K. M., Hemric, M. E. & Warshaw, D. M. (1992). The effects of smooth muscle caldesmon on actin ®laments motility. J. Biol. Chem. 267, 23001± 23006. Hanein, D., Rost, L. E., Matsudaira, P. & DeRosier, D. J. (1997). 3D structures of F-actin decorated with two different actin-binding domains. Biophys. J. 72, A240. Hodgkinson, J. L., Marston, S. B., Craig, R., Vibert, P. & Lehman, W. (1997). 3D image reconstruction of reconstituted smooth muscle thin ®laments: Effects of caldesmon. Biophys. J. 72, 2398± 2404. Horowitz, A., CleÂment-Chomienne, O., Walsh, M. P., Tao, T., Katsuyama, H. & Morgan, K. (1996). Effects of calponin on force generation by single smooth muscle cells. Am. J. Physiol. 270, H1858± H1863. Itoh, T., Suzuki, S., Suzuki, A., Nakamura, F., Naka, M. & Tanaka, T. (1994). Effects of exogenously applied calponin on Ca2‡ regulated force in skinned smooth muscle of the rabbit mesenteric artery. P¯uÈgers Arch. 427, 301± 308. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for the building of protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 110 ± 119. Kamm, K. E. & Stull, J. T. (1985). The function of myosin and myosin light chain kinase phosphorylation in smooth muscle. Annu. Rev. Pharmacol. Toxicol. 25, 593± 620. Lehman, W. (1991). Calponin and the composition of smooth muscle thin ®laments. J. Muscle Res. Cell Motil. 12, 221 ± 224. Lehman, W. & Kaminer, B. (1984). Identi®cation of a thin ®lament linked 35,000 dalton protein in vertebrate smooth muscle. Biophys. J. 45, 109a. Lehman, W., Vibert, P., Craig, R. & BaÂraÂny, M. (1996). Actin and the Structure of Smooth Muscle Thin Filaments. In Biochemistry of Smooth Muscle Contraction (BaÂraÂny, M., ed.), pp. 47 ± 62, Academic Press, San Diego, USA. Lorenz, M., Popp, D. & Holmes, K. C. (1993). Re®nement of the F-actin model against x-ray ®ber diffraction data by the use of a directed mutation algorithm. J. Mol. Biol. 234, 826 ± 836. Mabuchi, K., Li, Y., Tao, T. & Wang, C-L. A. (1996). Immunocytochemical localization of caldesmon and calponin in chicken gizzard smooth muscle. J. Muscle Res. Cell Motil. 17, 243 ± 260. Mabuchi, K., Li, B., Ip, W. & Tao, T. (1997). Interaction of calponin with desmin. Biophys J. 72, A189. Makuch, R., Birukov, K., Shirinsky, V. & Dabrowska, R. (1991). Functional interrelationship between calponin and caldesmon. Biochem. J. 280, pp. 33 ± 38. Malmqvist, U., Trybus, K. M., Yagi, S., Carmichael, J. & Fay, F. S. (1997). Slow cycling of unphosphorylated myosin is inhibited by calponin, this keeping smooth muscle relaxed. Proc. Natl Acad. Sci. USA, 94, 7655± 7660. Marston, S. B. & Huber, P. A. J. (1996). Caldesmon. In Biochemistry of Smooth Muscle Contraction (BaÂraÂny,

F-actin±Calponin Structure M., ed.), pp. 77± 90, Academic Press, San Diego, USA. Matsudaira, P. (1991). Modular organization of actin crosslinking proteins. Trends Biochem. Sci. 16, 87 ± 92. McGough, A. & Way, M. (1995). Molecular model of an actin ®lament capped by a severing protein. J. Struct. Biol. 115, 144 ± 150. McGough, A., Way, M. & DeRosier, D. J. (1994). Determination of the alpha-actinin binding site on actin ®laments by cryo-electron microscopy and image analysis. J. Cell Biol. 126, 433 ± 443. Mezgueldi, M., Fattoum, A., Derancourt, J. & Kassab, R. (1992). Mapping the functional domains in the amino-terminal region of calponin. J. Biol. Chem. 267, 15943± 15951. Miller, C. J., Cheung, P., White, P. & Reisler, E. (1995). Actin's view of the actomyosin interface. Biophys. J. 68, 50s ± 56s. Milligan, R. A. & Flicker, P. F. (1987). Structural relationships of actin, myosin and tropomyosin revealed by cryo-electron microscopy. J. Cell Biol. 105, 29 ± 39. Noda, S., Ito, M., Watanabe, S., Takahashi, S. & Maruyama, K. (1992). Conformational changes of actin induced by calponin. Biochem. Biophys. Res. Commun. 185, 481 ± 487. North, A. J., Gimona, M., Cross, R. A. & Small, J. V. (1994a). Calponin is localized in both the contractile apparatus and the cytoskeleton of smooth muscle cells. J. Cell Sci. 107, 437± 444. North, A. J., Gimona, M., Lando, Z. & Small, J. V. (1994b). Actin isoform compartments in chicken gizzard smooth muscle cells. J. Cell Sci. 107, 445± 455. Obara, K., Szymanski, P. T., Tao, T. & Paul, R. J. (1996). The effects of calponin on isometric force and shortening velocity in permeabilized taenia coli smooth muscle. Am. J. Physiol. 270, C481± C487. Okagaki, T., Higashi-Fujime, S., Ishikawa, R., TakamaOhmuro, H. & Kohama, K. (1991). In vitro movement of actin ®laments on gizzard smooth muscle myosin: requirement of phosphorylation of myosin light chain and effects of tropomyosin and caldesmon. J. Biochem. 109, 858 ± 866. Owen, C. H., Morgan, D. G. & DeRosier, D. J. (1996). Image analysis of helical objects: the Brandeis helical package. J. Struct. Biol. 116, 167± 175. Parker, C. A., Takahashi, K., Tao, T. & Morgan, K. (1994). Agonist-induced redistribution of calponin in contractile vascular smooth muscle cells. Am J. Physiol. 267, C1262± C1270. Rayment, I., Holden, H. M., Whittaker, M., Yohn, C. B., Lorenz, M., Holmes, K. C. & Milligan, R. A. (1993). Structure of the actin-myosin complex and its implications for muscle contraction. Science, 261, 58± 65. Sellers, J. R. & Adelstein, R. S. (1986). Regulation of contractile activity. In The Enzymes (Boyer, P. D. & Krebs, E. G., eds), pp. 381± 418, Academic Press, Orlando, USA. Shirinsky, V. P., Birukov, K. G., Hettasch, J. M. & Sellers, J. R. (1992). Inhibition of the relative movement of actin and myosin by caldesmon and calponin. J. Biol. Chem. 267, 15886± 15892. Smith, C. W., Pritchard, K. & Marston, S. B. (1987). The mechanism of calcium regulation of vascular smooth muscle by caldesmon and calmodulin. J. Biol. Chem. 262, 116±122. Sobue, K. & Sellers, J. R. (1991). Caldesmon, a novel regulatory protein in smooth muscle and nonmuscle actomyosin systems. J. Biol. Chem. 266, 12115± 12118.

159

F-actin ±Calponin Structure Stafford, W. F., Mabuchi, K., Takahashi, K. & Tao, T. (1995). Physical characterization of calponin. J. Biol. Chem. 270, 10576± 10579. Straub, F. B. (1942). Actin. In Studies from the Institute of Medical Chemistry, University of Szeged, vol. 2, pp. 3 ± 16, Szeged, Hungary. Szymanski, P., Grabarek, Z. & Tao, T. (1997). Correlation between calponin and myosin sub-fragment 1 binding of F-actin and ATPase inhibition. Biochem. J. 321, 519± 523. Takahashi, K., Hiwada, K. & Kokubu, T. (1986). Isolation and characterization of a 34 kDa calmodulin and F-actin binding protein from chicken gizzard smooth muscle. Biochem. Biophys. Res. Commun. 141, 20 ± 26. Trachtenberg, S. & DeRosier, D. J. (1987). Three-dimensional structure of the frozen hydrated ¯agellar ®lament of Salmonella typhimurium. J. Mol. Biol. 195, 581± 601. Vanderkerckhove, J. & Weber, K. (1978). At least 6 different actins are expressed in a higher mammal. J. Mol. Biol. 126, 783 ±802. Vibert, P., Craig, R. & Lehman, W. (1993). Three-dimensional reconstruction of caldesmon-containing

smooth muscle thin ®laments. J. Cell Biol. 123, 313 ± 321. Vibert, P., Craig, R. & Lehman, W. (1997). Steric model for activation of muscle thin ®laments. J. Mol. Biol. 266, 8 ± 14. Walsh, M. P., Carmichael, J. D. & Kargacin, G. J. (1993). Characterization and confocal imaging of calponin in gastrointestinal smooth muscle. Am. J. Physiol. 265, 1371± 1378. Wang, P. & Gusev, N. (1996). Interaction of smooth muscle calponin and desmin. FEBS Letters, 392, 255± 258. Winder, S. J. & Walsh, M. P. (1990). Smooth muscle calponin. Inhibition of actomyosin MgATPase and regulation by phosphorylation. J. Biol. Chem. 265, 19148± 10155. Winder, S. J. & Walsh, M. P. (1993). Calponin: thin ®lament-linked regulation of smooth muscle contraction. Cellular Signalling, 5, 677 ± 686. Yang, Y., Dowling, J., Yu, Q-C. & Kouklis, P. (1996). An essential cytoskeletal linker protein connecting actin micro®laments to intermediate ®laments. Cell, 86, 655 ± 665.

Edited by W. Baumeister (Received 30 April 1997; received in revised form 20 July 1997; accepted 21 July 1997)