Structural modules in actin-binding proteins: towards a new classification

Biochimica et Biophysica Acta 1448 (1999) 323^348

Review

Structural modules in actin-binding proteins: towards a new classi¢cation Marleen Van Troys, Joe«l Vandekerckhove, Christophe Ampe * Flanders Interuniversity Institute for Biotechnology, Department of Biochemistry, Faculty of Medicine, University of Gent, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium Received 8 September 1998; accepted 14 October 1998

Abstract The number of actin binding proteins for which (part of) the three-dimensional structure is known, is steadily increasing. This has led to a picture in which defined structural modules with actin binding capacity are shared between different actin binding proteins. A classification of these based on their common three-dimensional modules appears a logical future step and in this review we provide an initial list starting from the currently known structures. The discussed cases illustrate that a comparison of the similarities and variations within the common structural actin binding unit of different members of a particular class may ultimately provide shortcuts for defining their actin target site and for understanding their effect on actin dynamics. Within this concept, the multitude of possible interactions by an extensive, and still increasing, list of actin binding proteins becomes manageable because they can be presented as variations upon a limited number of structural themes. We discuss the possible evolutionary routes that may have produced the present array of actin binding modules. ß 1999 Elsevier Science B.V. All rights reserved. Keywords: Actin; Actin-binding protein; Classi¢cation; Structure^function relationship; Evolutionary relationship

Contents 1.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

324

2.

Evolutionary aspects of actin and actin-binding proteins . . . . . . . . . . . . . . . . . . . . . . . . . .

324

3.

Structure^function relationships of actin-binding proteins: towards a new classi¢cation . . . 3.1. Scenario 1: structurally similar actin-binding units have evolved to modules that recognise di¡erent parts of the actin molecule and either have di¡erent actin-binding properties or work together in generating the actin-binding site . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Scenario 2: structurally similar actin-binding units target to similar parts of the actin molecule though with a di¡erent or enhanced e¡ect on actin dynamics . . . . . . . . . . . .

327

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

342

4.

* Corresponding author. Fax: +32 (9) 264-5337; E-mail: [email protected] 0167-4889 / 99 / $ ^ see front matter ß 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 4 8 8 9 ( 9 8 ) 0 0 1 5 2 - 9

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

333 336

324

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348 4.1. What are the possible evolutionary strategies that have led to the structure^function relationships observed within the current actin system? . . . . . . . . . . . . . . . . . . . . . . .

342

Note added in proof: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

343

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

343

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

343

1. Introduction Actin is the main component of the micro¢lament system in all eukaryotic cells and central in their basic functions ranging from muscle contraction, cell crawling, cytokinesis, cytoplasmic organisation to intracellular transport. It is one of the most conserved proteins in eukaryote evolution, e.g. actin of human and of the protist Euplotes crassa still share 61% identity or 78.2% similarity and chordate muscle actin and lamprey (Lampetra £uviatilis) actin di¡er only in four out of 375 amino acids rendering actin even more conserved in evolution than histone H4 [1]. This extreme conservation of the actin structure basically re£ects two important characteristics of the molecule: ¢rst, the capability to self-assemble into polymers (F-actin) and second, the property to interact with a multitude of actin-binding proteins regulating this assembly or using this assembly as a scaffold. While elucidation of the three-dimensional structure of monomeric actin (G-actin) [2] and the subsequent modelling of the ¢lament [3] was informative for the identi¢cation of sequences responsible for actin^actin contacts, the location of subdomains or regions of actin involved in binding to the various regulatory proteins is only starting to be unravelled. The atomic structures of three of these proteins: gelsolin segment 1 [4], bovine pro¢lin [5] and DNase I [2] have been solved in complex with actin. In addition, part of or the complete structures of about 25 actin-binding proteins are known today. Combined with the wealth of data derived from biochemical and molecular biology approaches, the emerging picture, although far from complete, appears to be that only a limited part of actin's surface is being recognised. Consequently, in view of the large number of actin-binding proteins with both diverse and overlapping functions in regulating actin dynamics, the following questions arise. First, how do all these di¡er-

ent proteins interact with similar parts of actin and second, have actin-binding sites evolved from a limited number of structural themes? Identifying these common themes will ultimately lead to a new classi¢cation, thereby replacing the traditional classi¢cation according to function. In this review, after a brief survey of the evolutionary aspects of both actin and actin-binding proteins, we will discuss recent results that contribute to answering the above questions, we focus on a few already identi¢ed actin-binding units and expatiate upon the variations found in di¡erent proteins using the same unit by tracing the di¡erent scenarios of actin interaction developed in these actin-binding proteins. 2. Evolutionary aspects of actin and actin-binding proteins Actin is a very ancient molecule and is generally believed to ¢nd its origin at the onset of eukaryotic Table 1 Sequence similarity (%) between actins from evolutionary divergent species Actin

Hs

Hs Xl Dm Dd Sc At

Xl

Dm

Dd

Sc

At

99.5

98.9 98.7

97.1 96.8 96.8

94.4 93.9 93.3 93.1

92.5 92.0 92.0 93.9 90.7

Hs, Homo sapiens; Xl, Xenopus laevis; Dm, Drosophila melanogaster; Dd, Dictyostelium discoidium; Sc; Saccharomyces cerevisiae; At, Arabidopsis thalliania. Sequences were aligned using Pile Up (gap weight 12, gap-length weight 4), GCG Wisconsin version 9, and distances were calculated (threshold of comparison 2 denominator: length of shorter sequence without gaps, PAM 250 amino acid substitution matrix).

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

325

Fig. 1. Representation of the three-dimensional structure of actin with a nucleotide and divalent cation (shown in blue) bound in the central cleft [2] ; K-helices are shown in red and L-strands in yellow. The four subdomains and the NH2 - (N) and COOH- (C) terminus are indicated.

life 1.5^2 billion years ago [6]. As most families of actin-binding proteins are present in all branches of the eukaryotic kingdom, they also must originate from early eukaryotes. The large amount of sequence information available, forms, in combination with the structural data, the necessary basis to reconstruct past events that have led to the contemporary actin system, thereby helping us to get a clearer view on this complex array of interacting protein families. At present, over 180 unique full-length actin sequences (and an additional 42 partial) are listed in the Swiss Prot database (release 36). These actins are from a wide variety of organisms ranging from protists, fungi, plants to invertebrate and vertebrate animals. As sequence conservation is very high (illustrated for a small set of actins from evolutionary divergent species in Table 1), aligning these se-

quences is straightforward. For a recently published alignment of numerous actins and a phylogenetic tree, we refer to Sheterline et al. [7]. The highest number of amino acid substitutions, (very short) deletions or insertions are found in actins from lower eukaryotes, and to a lesser extent from plants. Reports on actins di¡erent in some functional characteristics such as DNase I- or phalloidin-binding usually concern these family members. Fig. 1 shows the tertiary actin structure originally determined by Kabsch et al. [2] using the K-skeletal muscle actin^DNase I complex. Later on, McLaughlin et al. [4] and Schutt et al. [5] respectively determined the K-actin^gelsolin segment 1 and the L-actin^pro¢lin structures showing only minor di¡erences in the actin structure at the interfaces with the actinbinding proteins, at the NH2 - and COOH-termini or

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

326

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

in the rotation of the domains. Actin appears as a relatively £at, two-lobed molecule with a nucleotide and a divalent cation bound in the central cleft between the two lobes. Each lobe-shaped domain is divided in two subdomains. Primary structure alignments reveal parts of the actin sequence that are more prone to variation than others and mapping these regions onto the three-dimensional structure of the actin monomer may lead to an interpretation in terms of ligand interaction or ¢lament assembly. The highest conservation is found in the interior of the protein and in extensive parts of the two lower subdomains 1 and 3. In recent years it has become evident that the strong sequence conservation of this interior part of the protein does not merely re£ect a `sca¡olding' function, but tells a tale of ancestry. This core appears to be an ancient nucleotide-binding pocket also found in other eukaryotic and prokaryotic proteins with very diverse functions [8^10]. For a description of the common fold (generally termed the `actin fold') of this superfamily and a detailed comparison of the di¡erent family members we refer to [11,12]. This conservation within the actin superfamily clearly is informative in elucidating common properties, such as the conformations of open and closed states coupled to interdomain motions around hinge regions or the mechanisms of nucleotide hydrolysis. Conversely, within the parts divergent from other superfamily members, the degree of conservation between the di¡erent actins may depend on whether these sequences are involved in general functions, e.g. in F-actin assembly, or not. Indeed, when involved in actin^actin contacts, the sequences are largely conserved. Based on the ¢lament model of Holmes et al. [3], the regions contributing to longitudinal actin ¢lament contacts are located in subdomain 3 (residues 166^169, 286^289, 322^325), subdomain 4 (residues 202^204, 243^245), subdomain 1 (residue 375) and in subdomain 2 (residues 40^45). Of these actin^actin contacts the latter, also termed the DNase-binding loop, is least conserved; it contains plenty of amino acid exchanges, especially in actins from protists, fungi and plants. Several studies demonstrated that this region of actin is locally very £exible [4,13^15] and this capacity on its own may be su¤cient to compensate for the higher mutational tolerance. In accordance, actin diver-

gent in this region, e.g. from Tetrahymena [16] and Trypanosoma [17] has been shown to self-assemble or copolymerise with skeletal muscle actin. The higher variability in regions not belonging to the core or not involved in self-assembly may re£ect isoform-speci¢c function or adaptation to speci¢c ligands. One example is formed by the extreme NH2 terminus. All actins contain acidic residues in this part of their sequence, but di¡erences do occur and were in fact used to classify vertebrate actins [18,19]. 1 H-NMR spectroscopy suggested that in F-actin, the NH2 -terminus protrudes from the ¢lament as a highly mobile sticky `¢shing-rod' [20]. Consequently, sequence variation within this arm may modulate the a¤nity of the respective actins for target actin-binding proteins. Indeed, the acidic NH2 -terminus, although shown not to be essential in yeast [21], has been implicated in the interaction with several actin-binding proteins (e.g. [22^24]). For instance, the presence and number of these acidic residues was demonstrated to be important in weak binding of myosin to actin ¢laments, in its transition to rigor binding and in force generation [25]. This is also re£ected in the higher sliding e¤ciency of skeletal muscle actin observed in in vitro motility assays compared to that of non-muscle isoforms; the former isoform always carries four negative charges in its NH2 -terminus [26]. Table 2 Sequence similarity (%) between di¡erent co¢lins (upper right) and between their respective actin-binding sitesa (lower left) Co¢lin

Hs

Hs Xb Dm Dd Sc At

88.1 66.7 59.5 73.8 69.1

Xb

Dm

Dd

Sc

At

82.6

46.6 44.6

50.4 50.4 54.0

52.5 51.1 54.6 57.6

53.2 50.4 59.0 58.4 55.4

61.9 57.1 69.1 64.3

69 64.3 73.8

69.1 73.8

73.8

Hs, homo sapiens; Xb, Xenopus borealis ; Dm, Drosophila melanogaster; Dd, Dictyostelium discoidium; Sc, Saccharomyces cerevisiae; At, Arabidopsis thalliania. Sequences were aligned using Pile Up (gap weight 12, gap-length weight 4), GCG Wisconsin version 9, and distances were calculated (threshold of comparison 2 denominator: length of shorter sequence without gaps, PAM 250 amino acid substitution matrix). a In each case, we used the sequence present as L-strand^K-helix^L-strand in the crystal structure and known to be important for actin interaction (see later Fig. 3A^C).

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

However, actin-binding proteins do not necessarily only interact with the more variable regions on actins. Gelsolin segment 1 and pro¢lin, for example, contact actin by wedging between the highly conserved subdomains 1 and 3. Also, the contact site of DNase I in subdomain 4 (residues 203^207) has been conserved between di¡erent actins. In these cases, the actin-binding proteins exert their e¡ect by overlapping with or mimicking of actin^actin contacts. It can be expected that, where in actin ^ as is evident from above ^ mutations have been selected against for reason of multifunctionality (i.e. ATPase activity, self-association and ligand interaction), actin-binding proteins will have evolved more freely. Indeed, when considering one actin-binding family across the whole phylogenetic tree, overall variation is usually much higher. Mammalian pro¢lins show less than 30% identity to pro¢lins from lower eukaryotes and plants [27]. Cyclase-associated protein (CAP) homologues in mammals, yeast and Dictyostelium are approximately 40% identical [28]. Amino acid sequences of members of the co¢lin family from species of di¡erent phyla are only 45^59% similar (Table 2, upper part); this is excluding drebrins, Dictyostelium coactosin and yeast Abp1p which are even more divergent [29]. Similarly, low eukaryote members of the gelsolin family display about 42^58% similarity to the NH2 -terminal half of human gelsolin. In having stayed 72^80% identical between vertebrate and invertebrate species [30], the L-thymosins form an exception, most probably due to their small size (43 amino acids) and consequently high percentage of essential residues. However, when only the region important in establishing the actin contact is considered, similarity within one family of actin-binding proteins is often higher. We present two case examples in Tables 2 and 3. One concerns co¢lins from evolutionary divergent species, the other the homologous segment 1 domains from members of the gelsolin family (sensu latiore) from the same or di¡erent species. As pointed out above, co¢lins from species of di¡erent phyla share about 45^59% similarity (Table 2, upper right). A major determinant in actin recognition in these proteins is a long K-helical structure [31,32] £anked by L-strands important for sca¡olding the helix [33] (see also below Fig. 3B). The similarity of

327

this region (L-strand^K-helix^L-strand) is signi¢cantly higher (57^74%, Table 2, lower left) than the one of the intact protein (45^59%, Table 2, upper right). One can make the same exercise for segment 1 from gelsolin and gelsolin-related proteins. These proteins are built up from either three, ¢ve or six repeated segments [34^37]. In general, segment 1 is the most conserved domain1 and important in capping actin ¢laments [38]. From the crystal structure of the gelsolin segment 1^actin complex [4] the residues important for contacting actin can be derived. Again, a major determinant is formed by an K-helix £anked by sca¡olding L-strands2 (see also Fig. 3A). With one exception, the similarity of this region (Table 3, lower left) is higher than the similarities of the entire segment 1 domain (Table 3, upper right, usually 10^15% higher). For a number of actin-binding proteins, the actin interaction is, however, but one of many functions and consequently the latter will also have imposed evolutionary pressure on parts of the molecule. Among the 35 known pro¢lins, only 18 residues are more than 80% conserved, eight of these form the polyproline-binding site and only two are involved in actin-binding [27]. Similarly, an extensive alignment of 82 myosin head domains reveals a higher degree of conservation within the nucleotide-binding pocket than in the actin-binding site [39]. 3. Structure^function relationships of actin-binding proteins: towards a new classi¢cation Up to now, the actin-binding proteins have usually been classi¢ed according to their functional properties and/or e¡ect on actin organisation, leading to groupings such as actin monomer-binding proteins, capping proteins, crosslinking proteins, membrane-

1 An exception to this is segment 2 from CapG and adseverin which have a higher similarity to each other and to segment 2 of gelsolin than their respective segments 1 (data not shown). The same applies for the similarity of segment 3 of CapG with gelsolin and adseverin. 2

Three other gelsolin segment 1 residues outside this region contact actin, i.e. Phe-49, Asp-50 and Phe-149 [4].

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

328

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

Table 3 Sequence similarity (%) between segment 1 of di¡erent gelsolins (upper right) and between their respective actin-binding sitesa (lower left) Gelsolin segment 1

Pp Fragm.

Pp Fragm. Lt Gels. Mm Vil. Mm Gels. Mm CapG Mm Adsev. Ha Gels. Dm Gels.

76.2 66.7 53.9 69.1 69.1 81.0 69.1

Lt Gels.

Mm Vil. Mm CapG

Mm Gels.

Mm Adsev.

Ha Gels.

Dm Gels.

64.6

56.5 51.9

58.8 55.1 66.7 55.2

59.3 53.7 60.2 58.1 75.9

65.7 59.1 57.1 53.3 65.7 62.9

60.8 57.0 59.8 59.1 68.2 62.6 71.4

73.8 71.8 76.2 69.1 76.2 71.4

69.2 85.7 78.6 73.8 73.8

53.3 53.3 57.1 66.7 71.8 64.1 71.8

83.3 73.8 78.6

73.8 73.8

83.3

Pp, Physarum polycephalum; Lt, Lumbricus terrestris ; Mm, Mus musculus ; Ha, Homerus americanus; Fragm., fragmin; Gels., gelsolin; Adsev., adseverin; Vil., villin. Sequences were aligned using Pile Up (gap weight 12, gap-length weight 4), GCG Wisconsin version 9, and distances were calculated (threshold of comparison 2 denominator: length of shorter sequence without gaps, PAM 250 amino acid substitution matrix). a In each case, we used the sequence present as L-strand^K-helix^L-strand in the crystal structure and known to be important for actin interaction (see later Fig. 3A^C).

associated proteins and so forth. However, in recent years, some actin-binding proteins, e.g. co¢lin, gelsolin and pro¢lin [38,40^42], have been shown to possess multifunctionality in their actin interaction obliging their classi¢cation into more than one group. In addition, it is evident that proteins from several of these groups, e.g. lateral binding proteins, actin severing proteins and crosslinking proteins, may all contact F-actin at similar sites. The calculated accessible surface per subunit in the actin ¢lament3 is î (compared to 19856 A î for an actin mono17052 A mer) and whereas this is still substantial, the currently identi¢ed di¡erent types of actin-binding proteins (more than 60) are liable to display some overlap in their target sites or may at least share some common binding features. In this respect, no less than eight actin-binding proteins have already been shown to bind the actin ¢lament on overlapping sites, designated as `binding hot spots' by McGough [44] (see Fig. 2 in [44]). As already hinted at by Pollard [45], a classi¢cation based upon the structural aspects of the actinbinding part of these proteins possibly forms a suitable and simplifying alternative. An initial grouping, mainly based on aligning primary structures, was at

3 Calculated using Biosym software [43] and using the actin ¢lament model proposed by Holmes et al. [3] as modi¢ed by Lorentz et al. [15].

that time limited to linear actin-binding motifs [46]. However, today, the three-dimensional structures of more and more actin-binding proteins are being solved at atomic resolution, allowing the identi¢cation of a number of structural actin-binding modules on which this new classi¢cation will be based. In a number of cases, these modules are not only common to members within one of the traditional families, but also appear to be used by actin-binding proteins that belonged to a di¡erent family if a classi¢cation based on function is used. Therefore, links are established between families that were previously considered to be unrelated. These ¢ndings strengthen a new tendency in the ¢eld that evolution has produced only a limited number of actin-binding units. These have been `incorporated' into di¡erent protein backgrounds and evolved to render proteins with either: (1) di¡erent e¡ects on actin dynamics; or (2) a similar e¡ect, but under a di¡erent regulation or coupled to a domain with an unrelated function. Together this gave rise to the strong diversity and the (apparent) redundancy characteristic of the current system of actin-binding proteins. Evidently, a modular build-up is far from being a unique feature of actin-binding proteins; domain shu¥ing and the resulting combinatorial advantage has been the basis of protein evolution and examples of mosaic proteins are numerous: SH2, SH3, WD, PH, FH, and PTB domains, the leucine zipper and ankyrin repeat are only a few well-known examples

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

of widely distributed domains observed in various combinations in eukaryotic proteins [47]. However, using this concept, one looks di¡erently at the complexity of the actin system and ultimately we believe it will provide an extensive simpli¢cation, certainly in view of the fact that only 1000 [48] to 8000 [49] fold types are estimated to be in use in the currently existing proteins. Doolittle [47] even suggests that less than 20 ancient domain types may have been su¤-

329

cient to generate the total current protein fold inventory. Consequently, more unexpected structural relationships between actin-binding proteins may be forthcoming. Most actin-binding protein families are found across the whole phylogenetic tree, indicating that their ancestral genes were formed before or at the start-point of the radiation of the eukaryotic kingdom. Evidently, the interfamily relationships will also be ancient and only obvious from a comparison

Fig. 2.

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

330

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

Fig. 2 (continued).

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

331

Fig. 2. Representation of the three-dimensional structures of actin-binding modules (see also Table 4, same order). All structures (except A) are taken from the Protein Data Base (Brookhaven National Laboratory) and the respective accession codes are listed. The drawings were generated using MOLSCRIPT [85]. K-Helices and L-strands are shown in red and yellow, respectively (except in D where the two all-K-CH domains are drawn in orange and red); the NH2 - (N) and COOH- (C) termini are indicated. In E, the threedimensional structure of the non-actin-binding protein galactose oxidase [84] is shown to illustrate the described fold (left) whereas the hypothetical related fold for the two scruin domains is shown on the right [86].

of tertiary structures, which, unfortunately, for many of the actin-binding proteins are still lacking. The strategy in setting up this classi¢cation will be to look for common structural features and only in a second step to understand how similar modules may have been adopted to function in more speci¢c ways. Table 4 and Fig. 2 provide an overview of the currently recognised actin-binding modules (references in Table 4): the gelsolin fold (common to the gelsolin and co¢lin family), the pro¢lin structure (overall similar to the gelsolin fold but based on a di¡erent topology), the villin headpiece F-actin-binding module (whether the active fold of thymosin L4 (TL4) is possibly related to the headpiece fold is discussed below), the calponin homology domain

(present in both well-known actin-binding proteins and in proteins involved in signal transduction), the scruin L-propeller fold (also connected with actinbinding in Drosophila Kelch and Physarum actin^ fragmin kinase), the myosin head domain (in which an ancient ATP-binding core, shared with microtubule motors and small G-proteins, has been recognised), and the hisactophilin fold (a L-barrel con¢guration strikingly similar to the non-homologous proteins interleukin-1L and ¢broblast growth factor [82] and also present as a four-fold repeated domain in the actin bundling protein fascin (A. Fedorov, L. Fedorov, S. Ono, F. Matsumura and S. Almo, personal communication). We also include DNase I in the list, although it is not yet clear whether its high

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

332

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

Table 4 Overview of the currently recognised actin-binding modules Actin-binding module, typical topology

Present in

References of 3D-structure determination or prediction

Gelsolin fold

c gelsolin segments 1^6 villin segments 1^6 adseverin segments 1^6a severin segments 1^3 fragmin segment 1^3a gCap39 segments 1^3a c destrin or actin depolymerizing factor/co¢lin/ actophorin/depactina 2U in twin¢lina

S1^K-actin [4], S1^6 [50] [51,52], S1 S2 [53]

[33,54,55] [56]a

Pro¢lin fold

Acanth. pro¢lin I and II human pro¢lin I bovine pro¢lin I and IIa plant pro¢lin

[57,58] [59] [5,60]a Arabidopsis [27] [61], Birch

Headpiece fold, 3 short helices surrounding tightly packed hydrophobic core

c vilin headpiece

[62]

dematin headpiecea Dictyostelium protovillin headpiecea c Drosophila quail headpiecea c Supervillin headpiecea

[63]a [64]a [65]a [66]a

c 2 in L-spectrin, in K-actinin, in dystrophina and utropina 2 in ¢lamin, ABP-120a 2 in tandem in ¢mbrin c 1 in calponin, SM-22a , neuronal proteina c 1 in Vav, IQGAPa

[67] K-actinin decorated F-actin [68]

Scruin, L-propeller/superbarrel

c 2Uin Limulus scruin c Drosophila Kelch c C-terminus of Physarum actin-fragmin kinasea

[73,74], scruin decorated F-actin, [75]a [76]a [77]a

Myosin motor domain

c Chicken muscle myosin c all members of the 13 myosin classes c myosin coreWkinesin and related microtubule motors

[78] [39]a [79^81]

Hisactophilin, L-barrel

c Dictyostelium hisactophilin c 4Uin fascin

[82] A. Fedorov, F. Matsumura and S. Almo, personal communication

DNase I, central core of 2, 6-stranded, L-sheets with 4 K-helices on either site

bovine DNase I

[2,83]

Calponin homology domain

[69] ¢mbrin decorated F-actin [70] calponin decorated F-actin [71] [72]a

a

The three-dimensional structure of these proteins has not yet been determined. They are incorporated on the basis of sequence homology and/or structure prediction.

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

a¤nity interaction to actin bears physiological relevance (the three last proteins in this list are not further discussed). As stated above, too few three-dimensional structures are elucidated as yet to present a ¢nal classi¢cation of actin-binding proteins. However, the present data do allow to trace how the currently known actin-binding modules have been adapted for their diverse regulation of actin dynamics in different actin-binding proteins. We present two scenarios which apply for most actin-binding proteins, although new scenarios or variations on a theme may pop up. 3.1. Scenario 1: structurally similar actin-binding units have evolved to modules that recognise di¡erent parts of the actin molecule and either have di¡erent actin-binding properties or work together in generating the actin-binding site Our ¢rst example here concerns the di¡erent segments of gelsolin and its related proteins. As already mentioned above, these proteins have a segmental build-up, probably arisen by multiple gene duplication. The recent elucidation of the gelsolin three-dimensional structure by Burtnick and colleagues [50] provided ultimate proof to the generally accepted assumption that the six homologous repeats of gelsolin all have the same fold. Those residues that, as sequence alignments show, are most conserved, appear important in forming the core of the domains and underlie the formation and packing of the secondary elements. Burtnick et al. [50] observed that the high degree of sequence identity in these structurally important regions within other members of the gelsolin family allows to assume similar folds and geometry's for their corresponding domains. Despite this common basis, di¡erent segments (S1, S2 and S4 in the case of gelsolin [87,88]; and S1, S2 and S5 in the case of adseverin [89]) display di¡erent actinbinding properties. Gelsolin S1 and S4 (S5 in adseverin) bind actin monomers, S1 can additionally cap ¢laments, whereas S2 only associates with the side of actin ¢laments [38]. The three-dimensional structure of gelsolin segment 1, solved as a complex with K-actin [4] (Fig. 2A), shows that the actin interaction of segment 1 is mainly mediated through several hydrophobic resi-

333

dues belonging to a long K-helix (residues 95^112 in plasma gelsolin) that packs on one side of a central L-sheet. The residues participating in the actin interaction are from the surface side of the helix and from the loops connecting the helix with the Lstrands [4]. The sequence `L-K-L', assigned DK1 E in the study of Burtnick et al. [50], in its particular conformation can be considered to form the actinbinding structural motif of segments 1 of these proteins (Fig. 3A). The topology diagram of gelsolin (S1 and S2 in Fig. 3A, Fig. 2 in [50]) shows that this same structural motif is present in all gelsolin segments. Recent results obtained in our laboratory demonstrate that, next to regions in the NH2 -terminus of S2 (identi¢ed by Sun et al. [92] and Kwiatkowski et al. [36]), the DK1 E module of segment 2 is also involved in establishing the actin contact. Using a peptide mimetic, we showed that the sequence 198^227 that spans L-strand D and K-helix K1 (of which in the peptide only K1 is probably adopting the correct fold), is involved in actin-binding [93]. This peptide inhibits F-actin-binding by segments 2^3 and severing by total gelsolin, stressing that the peptide binds to the segment 2 target site of gelsolin on the actin ¢lament. Consequently, positionally similar parts of segments 1 and 2 participate in the actin contact or, di¡erently put, segments 1 and 2 possess the same actin-binding structural motif. However, it is very evident they do not bind similarly to actin. Pope et al. [87] showed that S1 and S2^3 can interact with actin simultaneously. We showed that a peptide corresponding to the actin-binding helix of segment 1 does not interfere with S2^3-binding whereas the one from segment 2 does [32]. The actin^gelsolin S1 co-crystal [4] revealed that segment 1 binds in a cleft between subdomains 1 and 3 at the barbed end of the actin molecule, thereby mainly contacting the lower front face and bottom edge of subdomain 1 (orientation as in Fig. 1). The-binding site of segment 2 is not yet clearly de¢ned but it is generally modelled to bind along the ¢lament, contacting two subunits within one ¢lament strand of the double helix [68,94,95]. Together, these combined di¡erentialbinding modes of segments 1 and 2 are believed to mediate the severing activity of gelsolin and related proteins. Sequence comparison of the DK1 E module,

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

334

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

and, in particular, of the surface residues of the actin-binding helices K1 , demonstrates on what this difference in actin target site for the modules of S1 and S2 is likely based (Fig. 3D). The S1 hydrophobic residues Ile-103 and Val-106, essential in the S1^actin contact, are not conserved in the K-helix of S2 of which the surface side is overall more (positively) charged. In conclusion, what we are looking at in comparing S1 and S2 is the result of evolutionary sequence variation within the surface region of an identical structural module; a result that proved to be stable due to the gain in function (from only capping or severing to capping and severing). Actin-binding by the Limulus protein scruin forms a second example. This 102-kDa protein consists of two homologous domains (32% identity [75]) connected by a calmodulin-binding linker region [96]. Scruin binds actin protomers and is essential in the formation of the Limulus acrosomal process, a tightly packed bundle of approximately 100 scruindecorated actin ¢laments [97]. Helical reconstrucî resolution) from this tions of single ¢laments (13A bundle [73,74] demonstrated that the two scruin domains bind adjacent actin monomers along the one start helical actin ¢lament. The scruin amino acid sequence revealed the presence of six 50-amino acid repeats in each domain module, characteristic of the `superbarrel' structural fold found in the sialidase family of proteins [75,76] (Fig. 2E). Other actin-binding proteins are, by sequence homology, predicted to have the same fold: (1) the Physarum actin^fragmin kinase [77] that phosphorylates actin on subdomain 4 in the actin^fragmin complex [98]; and (2) the Drosophila Kelch gene product (ORF1) that is suggested to function in packing actin ¢lament bundles in the actin-rich ring canals through which nurse cells are

connected with the developing oocyte [99^101]. In scruin, each hypothetical six-bladed L-propeller-fold is suggested to form one actin-binding side (see below) and by combining cysteine modi¢cation with peptide mimetics, Sun et al. [86] suggested that the region around Cys-837 is part of the actin-binding site of the second domain. By correlating electron microscopy images of single scruin-decorated ¢laments to the Holmes F-actin model, Schmid and colleagues [74] determined the actin sequences to which each of the scruin domains binds. One domain binds the front face (as in Fig. 1) of subdomain 3 of a ¢rst actin subunit, whereas the second contacts the back of subdomain 1 of the next subunit along the genetic helix. Unfortunately, the high resolution structure of the scruin protein itself or in contact with actin are not yet available to elucidate the actin-binding interfaces of each of the two scruin domains. But evidently, these two homologous actin-binding structural modules of scruin contact di¡erent parts of the actin molecule. However, as stated by Schmid et al. [74] and in view of the duplicative origin of actin subdomains 1 and 3 [2], the sequences contacted by each scruin repeat in the different actin subdomains are related, namely the helix^loop^L-sheets 107^137 and 308^330, respectively. Thus, after gene duplication, these modules in scruin appear to have evolved to bind di¡erent, though related actin target sequences. A third example is found within the family of actin crosslinking proteins, such as L-spectrin, K-actinin, ¢mbrin, ¢lamin, cortexillin I and II, ABP-120, dystrophin and utrophin which have an actin-binding domain of approximately 27 kDa in common ([102^104] and references therein). The global buildup of these proteins is extremely modular: most contain, next to the actin-binding domain, a dimerising C

Fig. 3. Topology diagrams of gelsolin segment 1, gelsolin segment 2 (A) [50], yeast co¢lin (B) [33] and bovine pro¢lin (C) [5]. The start and end of the proteins or protein domains are indicated, K-helices are shown as circles, L-strands as triangles. The notations for the latter (A, AP, B, C, CP, D, E) are based on those of gelsolin as used by Burtnick et al. [50]. The part corresponding to the actinbinding structural DK1 E module is boxed (dotted line). The long helix K1 with actin-binding activity is shown in orange in each protein. The topology diagrams were generated using the TOPS programme [90] and the atlas of topology cartoons [91] (http:/tops.ebi. ac.uk/tops). (Bottom panel) Helical wheel presentation of the actin-binding K-helix (corresponding to K1 in the topology diagrams) of gelsolin segment 1, gelsolin segment 2 and vertebrate co¢lin. The basic residues conserved between the latter two are indicated and coloured blue, as well as their non-conserved counterparts in gelsolin segment 1. In green are shown the hydrophobic residues of the gelsolin segment 1 actin-binding helix crucial in the actin interaction and their non-conserved counterparts in gelsolin segment 2 and co¢lin.

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

335

336

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

domain that itself is either a repeat of an all-K domain (`spectrin' repeat [105]) or all-L domain (Ig-like [106]) or which contains a myosin-tail-like coiled-coil forming region [107]. In addition, some proteins from this family have regulatory domains that bind calcium [108]. Fimbrin does not dimerise, but contains a tandem pair of the þ 275 residue long actinbinding domain [109]. Sequence analysis [109] and image reconstruction from electron microscopy [110] revealed that the common actin-binding domain actually consists of two homologous repeats. The identity of these repeats within one protein or between proteins is relatively low: 15^20% (35^45% similar), but su¤cient to suggest a similar tertiary fold. This was recently con¢rmed through the elucidation of the three-dimensional structure of one of these repeats of L-spectrin [67] and two consecutive repeats of ¢mbrin [69]. These data demonstrate that each repeat is an independently folded, structurally homologous domain that consists of four major K-helices connected by long loops, some of which form less regular helical segments (Fig. 2D). Goldsmith et al. [69] show that the two repeats are connected by an extension of the last helix of the ¢rst repeat. They suggest that the orientation between the two repeats will be similar in related crosslinking proteins (see above) as the residues that interact within the interface of the repeats, are conserved. The independent nature of the repeat is additionally con¢rmed by the fact that this module is present in a single form in the NH2 -terminus of proteins of the calponin family [111] (reviewed in [112]) and interestingly, as well in a number of proteins involved in cdc42 and rac signalling [72]. From the homology with calponin stems the name for this repeat: calponin homology or CH domain. Electron microscopy data and low-resolution helical reconstructions [68,70], supported by earlier in vitro [113] and in vivo [114] results, indicate that the CH^CH dimer of ¢mbrin and K-actinin docks on the actin ¢lament in the concave surface between two actin subunits in the same ¢lament strand, contacting subdomains 1 and 2 of the lower and subdomain 1 of the upper subunit. The single CH domain was reported to target calponin to this same site on the ¢lament [71], but biochemical, as well as in vivo, analysis demonstrated that in calponin, the CH domain alone is not su¤cient to establish the F-actin

contact [115] and additional parts of the calponin molecule participate in the interaction [115^118]. Alternatively, proteins carrying a single CH domain may require dimerisation for actin-binding as was recently suggested for the signalling proteins IQGAP 1 [119] and Vav [120]. How do the properties of the CH domain ¢t under the above heading? The actual actin-interacting residues of both CH domains in the CH^CH dimer in the crosslinking proteins are not de¢ned. Both biochemical and structural data lend credibility to the fact that both evolutionary related CH domains participate in the actin contact (note that this refers to contacting one ¢lament and that crosslinking implies the binding of two separate CH^CH repeats), but not in the same way and probably not to the same extent. Indeed, Way et al. [121] showed that the ¢rst CH domain of K-actinin can bind F-actin, but with a ten-fold lower a¤nity than the tandem CH repeat. In analogy, Goldsmith et al. [69] report that the second CH domain binds F-actin, albeit weakly. This makes a synergistic role of each of the two CH domains in generating a tight F-actin interaction by the crosslinking proteins the most probable scenario, implying that the domain duplication has added extra features. Similarly, in calponin the combined structural and functional data suggest that the single CH domain works in synergy with other parts of the molecule [115]. 3.2. Scenario 2: structurally similar actin-binding units target to similar parts of the actin molecule though with a di¡erent or enhanced e¡ect on actin dynamics The most striking example of structural similarity connected to a grosso modo functional similarity is found for the F-actin-binding segment 2 of the gelsolin family and proteins of the co¢lin family. The latter display a mass that approximates that of a single gelsolin repeat. Co¢lin and its related proteins bind both G- and F-actin, but their most important functional e¡ect is to accelerate the rate-limiting step of the actin polymerisation cycle. Namely, they promote the dissociation of ADP subunits from the pointed end of the actin ¢lament and thereby speed up the turnover of ¢laments [41]. Lappalainen and Drubin [122] con¢rmed that a defect in co¢lin in

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

yeast leads to a slower depolymerisation of actin ¢laments in vivo. Hatanaka and colleagues [54] were the ¢rst to report the high structural similarity between gelsolin segments and destrin (a mammalian co¢lin homologue). Their observation was con¢rmed by the elucidation of the three-dimensional fold of the co¢lin isoforms from yeast [33] and Acanthamoeba castellanii [55]. This similarity between gelsolin segments and co¢lin had passed unnoticed when only the primary structures were available as these are only about 18% identical [55]. However, as illustrated in Fig. 3B, the topology of their folds is identical, except for the NH2 - and COOH-terminal part and the linker between L-strands B and C. Comparing gelsolin segments 1 and 2 demonstrates, however, that these regions, which di¡er between co¢lin and the gelsolin segments, are also divergent within gelsolin segments, strengthening the relation between the gelsolin and co¢lin family. Inevitably, the question arises whether this similarity is the result of convergent or divergent evolution. The low sequence identity seems to support the former option, although a study by Rost [123] proves that this does not form a conclusive argument as this author shows that both scenarios can lead to sequence identity as low as 10%. More importantly, the topology diagrams (Fig. 3A,B) clearly illustrate that the DK1 E-actin-binding module of gelsolin segments (see above) is also present in the co¢lin proteins. In vitro assays with mammalian and yeast co¢lin or with Acanthamoeba actophorin have shown that residues of the K1 -helix and of the D-K1 -loop are indeed participating in the actin interaction of these proteins ([23,31,32,124,125] and Van Troys et al., unpublished results). Because the K1 -helix in gelsolin S1 [4], gelsolin S2 [93] (see above) and in co¢lin interacts with actin, the question arose whether a functional resemblance exists between co¢lin and either of these gelsolin segments. As pointed out above, gelsolin segment 1 binds to the barbed end of ¢laments, whereas segment 2 binds along the side of the ¢lament. We recently demonstrated that the DK1 E structural module in co¢lin functions in a way analogous to the one of the Factin-binding segment 2 of gelsolin [32]. Competition experiments with either intact co¢lin and gelsolin segments or with chemically synthesised peptides that correspond to the DK1 sequence of co¢lin or

337

gelsolin S2 indicate that the actin-binding module of co¢lin targets this protein to the same or a very similar binding site on the actin ¢lament to which also the gelsolin segment 2 binds [32]. This is supported by a conservation of positive charges, which in co¢lin are implicated in the actin contact, on the surface site of the gelsolin S2- and co¢lin K1-helices (Fig. 3D) [32]. Using cryo-electron microscopy and image analysis, the binding site of gelsolin segment 2 was shown to lie between two neighbouring subunits along the ¢lament axis contacting subdomains 1 and 2 of the lower and subdomain 1 of the upper subdomain [94,95]. Recently, using the same approach, it was con¢rmed that also co¢lin binds to this site [126]. Although the presence of the similar actin-binding module appears to target proteins of the gelsolin (via S2) and co¢lin families to a similar site on the ¢lament, their overall subsequent e¡ect on actin dynamics appears not the same. As described above, gelsolin severs and caps ¢laments, whereas co¢lin promotes a faster depolymerisation of their ADPcharged parts. The co-operative action of the gelsolin segments can obviously account for this, as well as interactions made by those parts of the molecule in which structural di¡erences are observed (see Fig. 3 A,B). Using an alanine scanning mutational analysis, Lappalainen et al. [31] indeed show that in addition to residues in the above-described actin-binding module, residues in the co¢lin NH2 - and COOH-terminus are also involved in the yeast co¢lin^F-actin interaction. In addition, the sequence variation between gelsolin S2 and co¢lin, also in the DK1E module, may result in speci¢c e¡ects, i.e. the stabilising e¡ect of gelsolin S2-binding (Van Troys et al., unpublished results) versus the faster turnover induced by co¢lin-binding [41]. The recent discovery of a duplicated form of co¢lin by the group of Drubin [56] provides a fascinating sequel to this co¢lin story. The yeast protein twin¢lin (with homologues in humans and mice) consists of two repeats, each having approximately 20% identity to yeast co¢lin and most likely a similar fold as the latter. Unlike co¢lins, the actin-binding e¡ect of twin¢lin seems restricted to strong actin monomer sequestration and does not include enhancing the rate of pointed end ¢lament dissociation [56]. In this respect, twin¢lin displays a similar e¡ect as

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

338

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

Fig. 4. (A) Alignment of amino acid sequences of a possible L-thymosin from Hirudo nipponia (Hn TL4), human thymosin L4 (Hs TL4), the four repeats of the related C. elegans protein (Ce repeats 1^4) and the two repeats from the Acanthamoeba castellanii actobindin (Ac repeats 1^2) and from its Entamoeba histolytica homologue (Eh repeats 1^2). Residues conserved across the repeats are shown in blue, this is especially clear in the actin-binding motif (underlined) known to be functional in the actobindin repeats and in TL4. Alignments were made in a similar manner as described for Tables 1^3 and optimised manually. The similarity of the fourth repeat of the C. elegans protein is only obvious in its postulated actin-binding motif. (B) Schematic positioning of the L-thymosin variants with regard to the actobindins and the C. elegans protein, based on the alignment in A.

the actin-binding protein actobindin from A. castellanii. This small 88-residue-long protein consists of an internal tandem repeat (38/39 amino acids long) [127] and each repeat is able to contact one actin monomer at a similar site [24,128]. The bound actin dimer is, however, not favourably orientated to promote nucleation of actin ¢laments [129,130] and therefore, the strong inhibitory e¡ect of actobindin

on actin polymerisation [131] is a direct consequence of its duplicated nature. A similar protein is present in Entamoeba histolytica [132] (accession number AB002757) and it has been suggested that actobindin is a duplicated form of the monomer sequestering protein TL4 [133]. Interestingly, we recently found a Caenorhabditis elegans protein [134] (genome sequencing project, predicted protein encoded by

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

cDNA CEESP89F) that contains three (or perhaps four) such repeats. From the alignment of the two actobindin repeats, it is obvious they start with the actin-binding motif LKHAET. Fig. 4A shows the alignment of the repeats from actobindin, its Entamoeba homologue, TL4, a possible TL4 homologue from Hiduro nipponia (the ¢rst indication of a true Lthymosin in lower invertebrates (accession number D63651)) and the C. elegans protein. Given the repeated structure, aligning TL4 is not straightforward. In separate alignments, it scores best with the COOH-terminal part of the ¢rst repeat and the NH2 -terminal part of the second repeat from the C. elegans and the Entamoeba protein, but with the ¢rst repeat from Acanthamoeba (despite the fact that the Acanthamoeba and Entamoeba homologues are 58% similar). Therefore, we tentatively position TL4 as is shown in Fig. 4B and suggest it contains part of the ¢rst and part of the second repeat. If true, this indicates that TL4 is a remnant of an originally duplicated form (like, for instance, actobindin) which has subsequently lost parts of both repeats. Future studies on the C. elegans protein will have to reveal the nature of the actin interaction of its individual repeats and whether they display a similar synergistic e¡ect as found for actobindin. The co¢lin/twin¢lin and the actobindin/TL4/`C. elegans protein' proteins form yet other examples that duplication and subsequent diversi¢cation (and possibly deletion) have formed a recurrent theme, leading to the extensive variation in the family of actinbinding proteins. Those cases that have been studied, show this resulted in synergistic e¡ects on actin dynamics. Whether this is based on targeting of the repeated domains of these proteins to similar sites on the actin molecule ^ and consequently showing whether these protein sets also ¢t in our second evolutionary scenario ^ will be revealed by future studies on their respective interfaces with actin. A seemingly similar overall tertiary structure to the one described for the gelsolin segments has also been observed in another family of actin-binding proteins, namely the pro¢lin family. These small proteins are widespread and in many organisms multiple isoforms have been identi¢ed. They are extremely versatile both in their number of binding partners (actin [135], polyproline [136], polyphosphoinositides [137] and a multiprotein complex containing two actin-re-

339

lated proteins ¢rst isolated from Acanthamoeba [138]), and in their e¡ect on actin polymerisation [40,139]. Fig. 2B shows the three-dimensional structure of bovine pro¢lin demonstrating that, in analogy with segments of gelsolin and with co¢lins, it consists of a central L-sheet sandwiched on either site by a couple of K-helices, although the diagram in Fig. 3C reveals that this global structural similarity has a di¡erent topology. However, at present, we can only speculate whether this is another example of topological isomers as has been described for other evolutionary related proteins (see [140]). Bovine pro¢lin is shown, but the tertiary structures of pro¢lins from Acanthamoeba and plants have also been elucidated (references in Table 4). In spite of their low sequence identity (30%), their fold is well conserved and variation is limited to the region between strands B and C, in the length of the loop between C and D (plant pro¢lins) and in the orientation of the NH2 terminal helix (birch pro¢lin) [27,61]. In a recent report, Thorn and colleagues [27] present a detailed structure comparison of plant, amoeba and vertebrate pro¢lins. Only four out of the 16 pro¢lin amino acids that, based on the L-actin^bovine pro¢lin complex, interact with actin, are structurally conserved. Although several studies prove that the overall properties of pro¢lins from di¡erent species are the same in vitro and in vivo (references in [141]), the poor conservation of actin interacting residues may re£ect subtle, but functionally relevant, variation in the interaction of di¡erent pro¢lins with actin as already suggested by the observation that, unlike vertebrate pro¢lins, Arabidopsis pro¢lin does not increase the rate of nucleotide exchange [142]. Analysis of co-crystals of actin with gelsolin segment 1 [4] and with bovine pro¢lin [5] revealed that this similar fold allows both gelsolin S1 and pro¢lin to bind in the cleft between actin subdomains 1 and 3, to make contacts with both subdomains and thereby ¢t in the presented scenario. This binding site on actin overlaps with an actin^actin contact and can explain the sequestering activity of pro¢lin [40] as well as the interaction of both gelsolin S1 and pro¢lin with free barbed ends [36,40,143]. However despite these similarities in the target site, functional di¡erences are observed between pro¢lin and isolated S1 with regard to this last activity. S1 stays tightly

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

340

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

Fig. 5. Alignment of the primary structures from the COOH-terminus of human villin [145], dematin [63] and protovillin [64] and from the NH2 -terminal part of human TL4 [30,146]. The amino acids that are part of the consensus sequence (HxxHxxHxxLKK) are shown in blue.

bound to the ¢lament end and prevents further elongation or depolymerisation, whereas pro¢lin functions as a monomer shuttle: pro¢lin^actin complexes bind to the end, whereupon pro¢lin rapidly dissociates, possibly as a consequence of ATP hydrolysis [142]. Consequently, the pro¢lin^actin interaction must be more sensitive to the conformational changes which an actin subunit undergoes upon incorporation and ATP hydrolysis. In agreement with this, pro¢lin has a higher a¤nity for ATP than for ADP^actin, whereas the inverse is true for gelsolin [40,144]. The globally analogous contacts of S1 and pro¢lin on actin domains 1 and 3 are established through di¡erential interactions from which the different functional e¡ects described above may originate. In both proteins, the long K-helix in the COOH part participates in the interaction, but is positioned in a slightly di¡erent manner in the cleft between the two actin subdomains. In addition, strands or strand-connecting strands of the underlying L-sheet of each protein make di¡erent contacts with each of the two subdomains (Fig. 3A,C). Fig. 5 illustrates the striking similarity between the primary structure of the actin-binding regions of Lthymosins and the headpiece of villin, dematin and Dictyostelium protovillin. The consensus consists of a stretch of hydrophobic residues (H) preceding an LKK sequence (HxxHxxHxxxLKK) and is strongly conserved in all L-thymosin and headpiece sequences (see also repeat 1 in Fig. 4A). The importance of the latter three amino acids in the interaction with actin has been demonstrated for both TL4 [133,147] and the villin headpiece [148,149]. In addition, TL4 and a synthetic peptide corresponding to the last 22 residues of the villin headpiece compete for binding to actin [147], suggesting that TL4 and the villin headpiece also form a protein pair that interacts with a

similar site on the actin molecule. Do they share a similar actin-binding module? Structural data suggest this may only be partly the case. The hydrophobic patch in TL4 forms one side of an actin interacting NH2 -terminal amphipatic K-helix [147] and NMR solution structures of TL4 [150,151] suggest that this NH2 -terminal helix stops at the LKK sequence. Moreover, mutational analysis strongly suggest that the LKK sequence in TL4 does not adopt an K-helical structure in order to establish the contact with actin (Rossenu et al., submitted, Siminel et al., unpublished results). In contrast, the three-dimensional structure of the COOH-terminal 22 and 35 amino acids of the villin headpiece [62,152] demonstrate that the entire sequence considered in the alignment in Fig. 5 is mostly K-helical (Fig. 2C). The last 35 residues form a highly stable con¢guration of three short tightly packed K-helices that surround a hydrophobic core [62]. Two of the residues of the hydrophobic patch (in common with TL4, see Fig. 5) are solvent exposed in the headpiece fold as in TL4, but their position and orientation relative to the LKK motif is expected to be di¡erent from the one in TL4 in view of the fact that in the latter the motif is probably not in an K-helical con¢guration. In the absence of structural data on the actin^TL4 and actin^villin headpiece complex, an interpretation of these structural di¡erences remains di¤cult. However, an intriguing viewpoint is that these di¡erences are a re£ection of the conformational switch between ligands that preferentially bind monomeric (TL4) or polymerised actin (villin headpiece). In this respect, it is noteworthy that not only peptides of the villin headpiece, but also a TL4 variant with a mutation in the motif, have been shown to induce actin to polymerise in the absence of salt [133,148] as well as the fact that TL4 can, albeit only at high concen-

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

341

Fig. 6. Schematic presentation of the di¡erent evolutionary strategies in which an ancient structural actin-binding module may have evolved with regard to regulation of actin binding, actin target site and subsequent e¡ect on actin dynamics. The box and the attached symbol represent the global sca¡old and the actual surface contacting actin, respectively. Examples of scenario B and C are described in Section 3.2, whereas those of scenario D are described in Section 3.1. The lower part of the scheme (C and D) shows the possible scenarios in the special case of duplication of the module.

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

342

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

trations and with a low a¤nity, form a complex with F-actin [153]. 4. Conclusions Previous classi¢cations of actin-binding proteins employed functional characteristics [102,154,155] or the primary structure [46] and whereas these were useful and valid approaches, using these criteria a particular protein could often be assigned to two functional classes (e.g. pro¢lin, co¢lin) or relationships between actin-binding proteins remained unnoticed (e.g. gelsolin S2 and co¢lin). The recent boom in tertiary structure determination within the ¢eld of actin-binding proteins allows readdressing classi¢cation. Indeed, di¡erent actin-binding proteins share similar three-dimensional modules with actin-binding capacity. Thus, at present, it is probably more instrumental to build a classi¢cation based upon these structural modules rather than on the entire protein. Subsequently, this structural similarity provides a valuable starting point in the study of structure^ function relationships and for de¢ning docking sites on actin, since, as repeatedly illustrated by the examples in this review, identifying a `common structural domain' does not necessarily imply one can immediately derive or associate a precise function common to that particular class. This is in agreement with a recent global analysis of the relationships between protein fold and function [156]. 4.1. What are the possible evolutionary strategies that have led to the structure^function relationships observed within the current actin system? The multitude of actin-binding proteins in a cell (e.g. in yeast, a relatively simple eukaryotic organism, 38 known actin-binding proteins are present (Yeast genome directory [157])) all bind to a rather limited area of the accessible surface of the actin molecule. Hence from an evolutionary point of view, one can expect that once a given actin-binding module with a particular docking site on actin has arisen, the actin-binding protein may have evolved with a gain in function or an altered activity, though still interacting with a similar binding site on actin. This scenario implies two possible routes (Fig. 6A,B).

In a number of cases, the change will not a¡ect the manner in which the module contacts actin but rather the way it is regulated or a second activity, thereby providing possibilities for ¢ne-tuning the temporal and spatial actin organisation in the cell (Fig. 6). Examples of this pathway ^ that is quite common, but is out of the scope of this review ^ are numerous: the di¡erent a¤nities of pro¢lin isoforms for either phospholipids or polyproline-rich sequences [60], the di¡erences in Ca2‡ -dependence between members of the gelsolin family (e.g gelsolin and adseverin [37,158], gelsolin and villin [159]) or the di¡erent activities linked to actin-binding within the classes of unconventional myosin [160]. However, in other cases (described in Section 3.2) actin-binding modules will contact nearly the same site on the actin molecule by a number of conserved residues, whereas neighbouring residues are changed and cause a di¡erent e¡ect on actin dynamics (Fig. 6B). Perhaps a more interesting evolutionary route, is when, after duplication of a structural module, these remained associated, resulting in synergistic e¡ects (Fig. 6C,D, described in Sections 3.1 and 3.2) A priori, after such a duplication, two scenarios are possible: either both modules are fully functional and target to similar sites on actin or only one module remains functional and the binding site of the other becomes cryptic or occluded. In the ¢rst case, this can easily result in co-operation between the duplicated sites which may be advantageous for the cellular system (e.g. actobindin and possibly twin¢lin) (Fig. 6C). In the second scenario, one of the modules is allowed to undergo mutational drift. This may result in an unstable fold which is obviously selected against; on the other hand, other amino acid exchanges may leave the original three-dimensional sca¡old intact, but change the surface, ultimately leading to a gain of function (Fig. 6D). It is a somewhat puzzling observation that in the actin system selection for an additional and di¡erent actin-binding site has been preferred rather than any other gain of function module. From the three examples discussed above (gelsolin, scruin, calponin homology domains), it is evident this occurred during evolution. For instance, after duplication of segment 1 or 2 it evolved from barbed end capping (S1) or ¢lament side binding (S2) to e¤cient severing (S1+S2). Could it be that the physical linkage of a functional actin-binding

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

module to the mutationally drifting module ^ targeting the latter constantly to the close vicinity of the repetitive actin ¢lament structure ^ has promoted the selection of complementarity to a di¡erent site on actin? Alternatively, part of the problem may be that people have not looked hard enough for other functions in modules that originate from ancient actin-binding units. It may well be that, for instance, other gelsolin segments (S3, S5, S6) serve a di¡erent function than a mere spacer for Ca2‡ regulation by interacting with as yet unknown partners. Understanding the complex relationships of apparently structurally similar units coupled to functional diversity in the actin system forms a major challenge for the future.

[2]

[3] [4]

[5]

[6] [7] [8]

Note added in proof The participation of both CH-domains (CH1 and CH2) ^ with a dominant role for CH1 ^ has recently also been put forward by ¢tting the ¢mbrin crystal structure to helical reconstructions of ¢mbrin decorated actin ¢laments. Hanein, D., Volkmann, N., Goldsmith, S., Michon, A.-M., Lehman, W., Craig, R., DeRosier, D., Almo, S. and Matsudaira, P. An atomic model of ¢mbrin binding to F-actin and its implications for ¢lament crosslinking and regulation. Nat. Struct. Biol. 5 (1998) 787^792.

[9]

[10]

[11] [12]

Acknowledgements

[13]

We heartily thank Prof. S. Almo for sharing unpublished results and Prof. R.C. Robinson for providing the co-ordinates of the full-length gelsolin. We thank J.L. Verschelde for his large contribution in preparing the ¢gures of molecular structures. C.A is a Research Associate of the National Fund for Scienti¢c Research (F.W.O. Vlaanderen). M.V.T. was supported by Fund 01105698 of the B.O.F. and by the Interuniversity Attraction Poles of the Prime Minister (IUAP).

[14] [15]

[16]

[17]

[18]

References [1] J. Vandekerckhove, K. Weber, Chordate muscle actins di¡er

343

distinctly from invertebrate muscle actins. The evolution of the di¡erent vertebrate muscle actins, J. Mol. Biol. 179 (1984) 391^413. W. Kabsch, H.G. Mannherz, D. Suck, E.F. Pai, K.C. Holmes, Atomic structure of the actin:DNaseI complex, Nature 347 (1990) 37^44. K.C. Holmes, D. Pope, W. Gebhard, W. Kabsch, Atomic model of the actin ¢lament, Nature 347 (1990) 44^49. P.J. McLaughlin, J.T. Gooch, H.-G. Mannherz, A.G. Weeds, Structure of gelsolin segment 1^actin-complex and the mechanism of ¢lament severing, Nature 364 (1993) 685^692. C.E. Schutt, J. Myslik, M.D. Rozycki, N.C.W. Goonesekere, U. Lindberg, The structure of crystalline pro¢lin^L-actin, Nature 365 (1993) 810^816. R.F. Doolittle, Reconstructing history with amino acid sequences, Protein Sci. 1 (1992) 191^200. P. Sheterline, J. Clayton, J.C. Sparrow, Actin. Protein Pro¢le, vol. 2, 3rd edn., 1996 Academic Press Limited, London. K.M. Flaherty, D.B. McKay, W. Kabsch, K.C. Holmes, Similarity of the three-dimensional structure of actin and the ATPase fragment of a 70 K heat shock cognate protein, Proc. Natl. Acad. Sci. USA 88 (1991) 5041^5045. P. Bork, C. Sander, A. Valencia, An ATPase domain common to prokaryotic cell cycle proteins, sugar kinases, actin and Hsp70 heat shock proteins, Proc. Natl. Acad. Sci. USA 89 (1992) 7290^7294. J. Reizer, A. Reizer, M.H. Saier Jr., P. Bork, C. Sander, Exopolyphosphate phosphatase and guanosine pentaphosphate phosphatase belong to the sugar kinase/actin/Hsp70 superfamily, Trends Biol. Sci. 18 (1993) 247^248. W. Kabsch, K.C. Holmes, The actin fold, FASEB J. 9 (1995) 167^174. J.H. Hurley, The sugar kinase/heat shock protein 70/actin superfamily: implications of conserved structure for mechanism, Annu. Rev. Biophys. Biomol. Struct. 25 (1996) 137^ 162. W. Kabsch, J. Vandekerckhove, Structure and function of actin, Annu. Rev. Biophys. Biomol. Struct. 21 (1992) 49^76. M.M. Tirion, D. ben Avahram, Normal mode analysis of G-actin, J. Mol. Biol. 230 (1993) 186^195. M. Lorentz, D. Popp, K.C. Holmes, Re¢nement of the F-actin model against X-ray ¢bre di¡raction data by the use of a directed mutation algorithm, J. Mol. Biol. 234 (1993) 826^836. M. Hirono, R. Tanaka, Y. Watanabe, Tetrahymena actin: co-polymerisation with skeletal muscle actin and interactions with muscle actin binding proteins, Proc. Natl. Acad. Sci. USA 86 (1989) 75^79. R. A Mortara, Studies on trypanosomatid actin. Immunochemical and biochemical identi¢cation, J. Protzool. 36 (1989) 8^13. J. Vandekerckhove, K. Weber, At least six di¡erent actins are expressed in a higher mammal: an analysis based on the amino acid sequence of the amino-terminal tryptic peptide, J. Mol. Biol. 126 (1978) 783^802.

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

344

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

[19] J. Vandekerckhove, K. Weber, Actin typing on total cellular extracts, Eur. J. Biol. 113 (1981) 595^603. [20] D. Heintz, H. Kany, H.R. Kalbitzer, Mobility of the N-terminal segment of rabbit skeletal muscle F-actin detected by 1 H and 19 F nuclear magnetic resonance spectroscopy, Biochemistry 35 (1996) 12686^12693. [21] R.K. Cook, W.I. Blake, P.A. Rubenstein, Removal of the amino-terminal acidic residues of yeast actin, J. Biol. Chem. 267 (1992) 9430^9436. [22] K. Sutoh, Identi¢cation myosin binding sites on the actin sequence, Biochemistry 21 (1982) 3654^3661. [23] N. Yonezawa, E. Nishida, M. Ohba, M. Seki, H. Kumagai, H. Sakai, An actin-interacting peptide in the co¢lin sequence, Eur. J. Biochem. 183 (1989) 235^238. [24] K. Vancompernolle, J. Vandekerckhove, M.R. Bubb, E.D. Korn, The interfaces of actin and Acanthamoeba actobindin. Identi¢cation of a new actin binding motif, J. Biol. Chem. 266 (1991) 15427^15431. [25] R.K. Cook, W.I. Blake, P.A. Rubenstein, Enhanced stimulation of myosin subfragment 1 ATPase activity by addition of negatively charged residues to the yeast actin NH2 -terminus, J. Biol. Chem. 268 (1993) 2410^2415. [26] C.J. Miller, W.W. Wong, E. Bobkova, P.A. Rubenstein, E. Reisler, Mutational analysis of the role of the N-terminus of actin in actomyosin interactions. Comparison with other mutant actins and implications for the cross-bridge cycle, Biochemistry 35 (1996) 16557^16565. [27] K.S. Thorn, H.E.M. Christensen, R. Shigeta Jr., D. Huddler Jr., L. Shalaby, U. Lindberg, N.-H. Chua, C.E. Schutt, The crystal structure of a major allergen from plants, Structure 5 (1997) 19^32. [28] U. Gottwald, R. Brokamp, I. Karakesisisoglou, M. Schleicher, A. Noegel, Identi¢cation of a cyclase-associated protein (CAP) homologue in Dictyostelium discoideum and characterisation of its interaction with actin, Mol. Biol. Cell 7 (1996) 261^272. [29] A. Moon, G.D. Drubin, The ADF/co¢lin proteins: stimulusresponsive modulators of actin dynamics, Mol. Biol. Cell 6 (1995) 1423^11431. [30] D. Safer, P.K. Chowrashi, L-Thymosins from marine invertebrates: primary structure and interaction with actin, Cell Motil. Cytoskel. 38 (1997) 163^171. [31] P. Lappalainen, E.V. Fedorov, A.A. Fedorov, S.C. Almo, D.G. Drubin, Essential functions and actin-binding surfaces of yeast co¢lin revealed by systematic mutagenesis, EMBO J. 16 (1997) 5520^5530. [32] M. Van Troys, D. Dewitte, M. Goethals, J.-L. Verschelde, J. Vandekerckhove, C. Ampe, Analogous F-actin binding by co¢lin and gelsolin segment 2 substantiates their structural relationship, J. Biol. Chem. 272 (1997) 32750^32758. [33] A.A. Fedorov, P. Lappalainen, E.V. Fedorov, D.G. Drubin, S.C. Almo, Structure determination of yeast co¢lin, Nat. Struct. Biol. 4 (1997) 366^369. [34] C. Ampe, J. Vandekerckhove, The F-actin-capping proteins of Physarum polycephalum: cap42(a) is very similar, if not, identical to fragmin and is structurally and functionally very

[35]

[36]

[37]

[38] [39]

[40]

[41]

[42] [43]

[44] [45]

[46]

[47] [48] [49]

[50]

[51]

[52]

homologous to gelsolin; cap42(b) is Physarum actin, EMBO. J. 6 (1987) 4149^4157. M. Way, A. Weeds, Nucleotide sequence of pig plasma gelsolin. Comparison of protein sequences with human gelsolin and other actin-severing proteins shows strong sequence homology and evidence for long internal repeats, J. Mol. Biol. 203 (1988) 1127^1133. D.J. Kwiatkowski, P.A. Janmey, H.L. Yin, Identi¢cation of critical functional and regulatory domains in gelsolin, J. Cell Biol. 108 (1989) 1717^1726. J. Robbens, J. Louahed, K. De Pestel, I. Van Colen, C. Ampe, J. Vandekerckhove, J.-C. Renauld, Murine adseverin(D5), a novel form of the gelsolin family, and murine adseverin are induced by interleukin-9 in murine T-helper lymphocytes, Mol. Cell. Biol. 18 (1998) 4589^4596. A. Weeds, S. Maciver, F-actin-capping proteins, Curr. Opin. Cell Biol. 5 (1993) 63^69. M.J. Cope, J. Whisstock, I. Rayment, J. Kendrick-Jones, Conservation within the myosin motor domain: implications for structure^function, Curr. Biol. 4 (1996) 969^987. D. Pantaloni, M.-F. Carlier, How pro¢lin promotes actin ¢lament assembly in the presence of TL4, Cell 75 (1993) 1007^1014. M.-F. Carlier, V. Laurent, J. Santolini, R. Melki, D. Didry, G.-X. Xia, Y. Hong, N.H. Chua, D. Pantaloni, Actin depolymerizing factor (ADF/co¢lin) enhances the rate of ¢lament turnover: implications in actin-based motility, J. Cell Biol. 136 (1997) 1307^1322. M.-F. Carlier, D. Pantaloni, Control of actin dynamics in cell motility, J. Mol. Biol. 269 (1997) 459^467. Biosym Technologies INSIGHT II, version 2.3.0, DISCOVER, version 2.9.5, Biosym Technologies, San Diego, CA, 1994. A. McGough, F-actin binding proteins, Curr. Opin. Struct. Biol. 8 (1998) 166^176. T.D. Pollard, Actin and actin binding proteins, in: T. Kreis, R. Vale (Eds.), Guidebook to the Cytoskeleton and Motor Proteins, Oxford Univ. Press, 1993, pp. 3^11. J. Vandekerckhove, K. Vancompernolle, Structural relationships of actin-binding proteins, Curr. Opin. Cell Biol. 4 (1992) 36^42. R.F. Doolittle, The multiplicity of domains in proteins, Annu. Rev. Biochem. 64 (1995) 287^314. C. Chothia, One thousand families for the molecular biologist, Nature 357 (1992) 543^544. C.A. Orengo, D.T. Jones, J.M. Thornton, Protein superfamilies and domain superfolds, Nature 372 (1994) 631^ 634. L.D. Burtnick, E.K. Koepf, J. Grimes, E.Y. Jones, D.I. Stuart, P.J. McLaughlin, R.C. Robinson, The crystal structure of plasma gelsolin: implications for actin severing, capping and nucleation, Cell 90 (1997) 661^670. M.A. Markus, T. Nakayama, P.T. Matsudaira, G. Wagner, Solution structure of villin 14T, a domain conserved among actin severing proteins, Protein Sci. 3 (1994) 70^81. M.A. Markus, P. Matsudaira, G. Wagner, Re¢ned structure

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

[53]

[54]

[55]

[56]

[57]

[58]

[59]

[60]

[61]

[62]

[63]

[64]

[65]

[66]

[67]

of villin 14T and a detailed comparison with other severing domains, Protein Sci. 6 (1997) 1197^1209. A. Schnuchel, R. Wiltscheck, L. Eichinger, M. Schleicher, T. Holak, Structure of severin domain 2 in solution, J. Mol. Biol. 247 (1995) 21^27. H. Hatanaka, K. Ogura, K. Moriyama, S. Ichikawa, I. Yahara, F. Inagaka, Tertiary structure of destrin and structural similarity between two actin-regulating protein families, Cell 85 (1996) 1047^1055. S.A. Leonard, A.G. Gittis, E.C. Petrella, T.D. Pollard, E.E. Lattman, Crystal structure of the actin-binding protein actophorin from Acanthamoeba, Nat. Struct. Biol. 4 (1997) 369^ 373. B.L. Goode, D.G. Drubin, P. Lappalainen, Regulation of the cortical actin cytoskeleton in budding yeast twin¢lin, a ubiquitous actin monomer sequestering protein, J. Cell Biol. 142 (1998) 723^733. V.K. Vinson, S.J. Archer, E.E. Latttman, T.D. Pollard, D.A. Torchia, Three-dimensional solution structure of Acanthamoeba pro¢lin I, J. Cell Biol. 122 (1993) 1227^1283. A.A. Fedorov, K.A. Magnus, M.H. Graupe, E.E. Lattman, T.D. Pollard, S.C. Almo, X-ray structures of isoforms of the actin binding protein pro¢lin that di¡er in their a¤nity for phosphatidylinositol phosphates, Proc. Natl. Acad. Sci. USA 91 (1994) 8636^8640. W.J. Metzler, B.T. Farmer II, K.L. Constantine, M.S. Friedrichs, T. Lavoie, L. Mueller, Re¢ned solution structure of human pro¢lin I, Protein Sci. 4 (1995) 450^459. A. Lambrechts, J.-L. Verschelde, V. Jonckheere, M. Goethals, J. Vandekerckhove, C. Ampe, The mammalian pro¢lin isoforms display complementary a¤nities for PIP2 and proline-rich sequences, EMBO J. 16 (1997) 484^494. A.A. Fedorov, T. Ball, N.M. Mahoney, R. Valenta, S.C. Almo, The molecular basis for allergen cross-reactivity : crystal structure and IgE-epitope mapping of birch pollen pro¢lin, Structure 5 (1997) 33^45. C.J. McKnight, P. Matsudaira, P.S. Kim, NMR-structure of the 35-residue villin headpiece subdomain, Nat. Struct. Biol. 4 (1997) 180^184. A.P. Rana, P. Ru¡, G.J. Maalouf, D.W. Speicher, A.H. Chishti, Cloning of human erythroid dematin reveals another member of the villin family, Proc. Natl. Acad. Sci. USA 90 (1993) 6651^6655. A. Hofmann, A.A. Noegel, I. Bomblies, F. Lottspeich, M. Schleicher, The 100 kDa F-actin capping protein of Dictyostelium amoeba is a villin prototype (`protovillin'), FEBS Lett. 328 (1993) 71^76. M. Mahajan-Miklos, L. Cooley, The villin-like protein encoded by the Drosophila `quail' gene is required for actin bundle assembly during oogenesis, Cell 78 (1994) 291^ 301. K.N. Pestonjamasp, R.K. Pope, J.D. Wulfkuhle, E.J. Luna, Supervillin (p205); a novel membrane associated, F-actin binding protein in the villin/gelsolin superfamily, J. Cell Biol. 139 (1997) 1255^1269. K.D. Carugo, S. Ban¬uelos, M. Saraste, Crystal structure of a

[68]

[69]

[70]

[71]

[72] [73]

[74]

[75]

[76]

[77]

[78]

[79]

[80]

[81] [82]

[83]

345

calponin homology domain, Nat. Struct. Biol. 4 (1997) 175^ 179. A. McGough, M. Way, A. Weeds, Determination of the Kactinin binding site on actin ¢laments by cryo-electronmicroscopy and image analysis, J. Cell Biol. 126 (1994) 433^443. S.C. Goldsmith, N. Pokala, W. Shen, A.A. Fedorov, P. Matsudaira, S.C. Almo, The structure of an actin crosslinking domain of human ¢mbrin, Nat. Struct. Biol. 4 (1997) 708^ 712. D. Hanein, P. Matsudaira, D.J. DeRosier, Evidence for a conformational change in actin induced by ¢mbrin (N375) binding, J. Cell Biol. 139 (1997) 387^396. J.L. Hodgkinson, M. El-Mezgueldi, R. Craig, P. Vibert, S.B. Marston, W. Lehman, 3-D image reconstruction of reconstituted smooth muscle thin ¢laments containing calponin: visualisation of interactions between F-actin and calponin, J. Mol. Biol. 273 (1997) 150^159. J. Castrena, M. Saraste, Does Vav bind to F-actin through a CH-domain?, FEBS Lett. 134 (1995) 149^151. C. Owen, D. DeRosier, A 13 angstro«m map of the actinscruin ¢lament from the limulus acrosomal process, J. Cell Biol. 123 (1993) 337^344. M.F. Schmid, M.A. Jacob, J. Janaka, P. Matsudaira, W. Chiu, Three-dimensional structure of a single ¢lament in the Limulus acrosomal bundle : scruin binds to homologous helix^loop^helix motifs in actin, J. Cell Biol. 124 (1994) 341^ 350. M. Way, M. Sanders, C. Garcia, J. Sakai, P. Matsudaira, Sequence and domain organisation of scruin, an actin crosslinking protein in the acrosomal process of Limulus sperm, J. Cell Biol. 128 (1995) 51^60. P. Bork, R.F. Doolittle, Drosophila kelch motif is derived from a common enzyme fold, J. Mol. Biol. 236 (1994) 1277^1282. L. Eichinger, L. Bomblies, J. Vandekerckhove, M. Schleicher, J. Gettemans, A novel type of protein kinase phosphorylates actin in the actin fragmin complex, EMBO J. 15 (1996) 5547^5556. I. Rayment, H.M. Holden, M. Whittaker, C.B. Yohn, M. Lorentz, K.C. Holmes, R.A. Milligan, Structure of the actin^myosin complex and its implications for muscle contraction, Science 261 (1993) 58^65. F.J. Kull, E.P. Sablin, R.J. Lau, R.J. Fletterick, R.D. Vale, Crystal structure of the kinesin motor domain reveals a structural similarity to myosin, Nature 380 (1996) 550^ 555. E.P. Sablin, F.J. Kull, R. Cooke, R.D. Vale, R.J. Fletterick, Crystal structure of the motor domain of the kinesin-related motor ncd, Nature 380 (1996) 555^559. L.A. Amos, R.A. Cross, Structure and dynamics of molecular motors, Curr. Opin. Struct. Biol. 7 (1997) 239^246. J. Habazettl, M. Schleicher, J. Otlewski, T.A. Holak, The structure of hisactophilin is similar to the interleukin-1L and ¢broblast growth factor, J. Mol. Biol. 228 (1992) 156^ 169. D. Suck, C. Oefner, W. Kabsch, Three-dimensional structure

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

346

[84]

[85]

[86]

[87]

[88]

[89]

[90]

[91]

[92]

[93]

[94]

[95]

[96]

[97]

[98]

[99]

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348 î resolution, EMBO J. of bovine pancreatic DNase I at 2.5 A 3 (1984) 2423^2430. N. Ito, S.E.V. Philips, K.D.S. Yadav, P.F. Knowles, Crystal structure of a free radical enzyme, galactose oxidase, J. Mol. Biol. 238 (1994) 794^814. P.J. Kraulis, Molscript: a program to produce both detailed and schematic plots of protein structures, J. Appl. Crystallogr. 24 (1991) 946^950. S. Sun, M. Footer, P. Matsudaira, Modi¢cation of Cys-837 identi¢es an actin binding site in the L-propeller protein scruin, Mol. Biol. Cell 8 (1997) 421^430. B. Pope, M. Way, A.G. Weeds, Two of the three actin binding domains of gelsolin bind to the same subdomain of actin, FEBS Lett. 280 (1991) 70^74. B. Pope, S. Maciver, A.G. Weeds, Localisation of Ca2‡ -sensitive actin monomer binding sites in gelsolin segment 4 and identi¢cation of calcium binding sites, Biochemistry 34 (1995) 1583^1588. M.G. Marcu, L. Zhang, A. Elzagallaai, J.M. Trifard, Localisation by segmental analysis and functional characterisation of a third actin binding site in domain 5 of scinderin, J. Biol. Chem. 273 (1998) 3661^3668. T.P. Flores, D.M. Moss, J.M. Thornton, An algorithm for automatically generating protein topology cartoons, Protein Eng. 7 (1994) 31^37. D.R. Westhead, D.C. Hatton, J.M. Thornton, An atlas of protein topology cartoons available on the world-wide web, Trends Biochem. Sci. 23 (1998) 35^36. H.-Q. Sun, D.C. Wooten, P.A. Janmey, H. Yin, The actin side binding domain of gelsolin also caps actin ¢laments. Implications for actin ¢lament severing, J. Biol. Chem. 269 (1994) 9473^9479. M. Van Troys, D. Dewitte, M. Goethals, J. Vandekerckhove, C. Ampe, Evidence for an actin binding helix in gelsolin segment 2: have homologous sequences in segment 1 and 2 of gelsolin evolved to divergent actin binding functions?, FEBS Lett. 397 (1996) 191^196. A. McGough, M. Way, Molecular model of an actin ¢lament capped by a severing protein, J. Struct Biol. 115 (1995) 144^150. A. McGough, W. Chiu, M. Way, Determination of the gelsolin binding site on the F-actin: implications for severing and capping, Biophys. J. 74 (1998) 764^772. M.C. Sanders, M. Way, J. Sakai, P. Matsudaira, Characterisation of the actin crosslinking properties of the scruin calmodulin complex from the acrosomal process of Limulus sperm, J. Biol. Chem. 271 (1996) 2651^2657. M.F. Schmid, J. Agris, J. Jakana, P. Matsudaira, W. Chiu, Crystallographic analysis of acrosomal bundle of Limulus sperm, J. Mol. Biol. 221 (1991) 711^725. J. Gettemans, Y. De Ville, J. Vandekerckhove, E. Waelkens, Physarum actin is phosphorylated as the actin^fragmin complex at residues Thr203 and Thr202 by a speci¢c 80 kDa kinase, EMBO J. 11 (1992) 3185^3191. F. Xue, L. Cooley, Kelch encodes a component of inter-

[100]

[101]

[102] [103] [104] [105]

[106]

[107]

[108]

[109]

[110]

[111]

[112]

[113]

[114]

cellular bridges in Drosophila egg chambers, Cell 72 (1993) 681^693. L.G. Tilney, M.S. Tilney, G.M. Guild, Formation of actin ¢lament bundles in the ring canals of developing Drosophila follicles, J. Cell Biol. 133 (1996) 61^74. D.N. Robinson, L. Cooley, Drosophila Kelch is an oligomeric ring canal actin organiser, J. Cell Biol. 138 (1997) 799^810. J. Vandekerckhove, Actin binding proteins, Curr. Opin. Cell Biol. 2 (1990) 41^50. P. Matsudaira, Actin crosslinking proteins at the leading edge, Sem. Cell Biol. 5 (1994) 165^174. J.H. Hartwig, The spectrin superfamily, Protein Pro¢le 1 (1994) 706^778. Y. Yan, E. Winograd, A. Viel, T. Cronin, S.J. Harrison, D. Branton, Crystal structure of the repetitive segments of spectrin, Science 262 (1993) 2027^2030. P. Fucini, C. Renner, H. Herberhold, A. Noegel, T.A. Holak, The repeating segments of the F-actin-crosslinking gelation factor (ABP-120) have an immunoglobulin-like fold, Nat. Struct. Biol. 4 (1997) 223^230. J. Faix, M. Steinmetz, H. Boves, R.A. Kammerer, F. Lottspeich, U. Mintert, J. Murphy, A. Stock, U. Aebi, G. Gerisch, Cortexillins, major determinants of cell shape and size, are actin-bundling proteins with a parallel coiled-coil tail, Cell 86 (1996) 631^642. W. Witke, A. Hofmann, B. Ko«ppel, M. Schleicher, A.A. Noegel, The Ca2‡ -binding domains in non-muscle type Kactinin: biochemical and genetical analysis, J. Cell Biol. 121 (1993) 599^606. M.V. deArruda, S. Watson, C.S. Lin, J. Leavitt, P. Matsudaira, Fimbrin is a homologue of the cytoplasmic phosphoprotein plastin and has domains homologous with calmodulin and actin gelation factors, J. Cell Biol. 111 (1990) 1069^1079. K.A. Taylor, D.W. Taylor, Projection image of smooth muscle K-actinin from two-dimensional crystals formed on positively charged lipid layers, J. Mol. Biol. 230 (1992) 196^ 205. K. Vancompernolle, M. Gimona, M. Herzog, J. Van Damme, J. Vandekerckhove, J.V. Small, Isolation and sequence of a tropomyosin-binding fragment of turkey gizzard calponin, FEBS Lett. 274 (1990) 146^150. M. Gimona, J.V. Small, Calponin, in: M. Ba¨ra¨ny (Ed.), Biochemistry in Smooth Muscle Contraction, Academic Press, San Diego, 1996, pp. 91^103. M. Mimuro, A. Asano, Further characterisation of a conserved actin binding 27 kDa fragment of actinogelin and Kactinins and mapping of their binding sites on the actin molecule by chemical crosslinking, J. Biol. Chem. 262 (1987) 4717^4723. J.E. Honts, T.S. Sandrock, S.M. Brower, J.L. O'Dell, A.E. Adams, Actin mutations that show suppression with ¢mbrin mutations identify a likely ¢mbrin-binding site on actin, J. Cell Biol. 126 (1994) 413^422.

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348 [115] M. Gimona, R. Mital, The single CH domain of calponin is neither su¤cient nor necessary for F-actin binding, J. Cell Sci. 111 (1998) 1813^1821. [116] M. Mezgueldi, A. Fattoum, J. Derancourt, R. Kassab, Mapping of the functional domains in the aminoterminal region of calponin, J. Biol. Chem. 267 (1992) 15943^15951. [117] M. El-Mezgueldi, P. Strasser, A. Fattoum, M. Gimona, Expressing functional domains of mouse calponin: involvement of the region around alanine 145 in the actomyosin inhibitory activity of calponin, Biochemistry 35 (1996) 3654^3661. [118] T. Mino, U. Yirasa, F. Nakamura, M. Naka, T. Tanaka, Two distinct actin binding domains in smooth muscle calponin, Eur. J. Biochem. 251 (1998) 262^268. [119] A.-M. Bashour, A.T. Fullerton, M.J. Hart, G.S. Bloom, IQGAP, a Rac and Cdc42-binding protein directly binds and crosslinks micro¢laments, J. Cell Biol. 137 (1997) 1555^1566. [120] F. Romero, S. Fisher, Structure and function of Vav, Cell Signal. 8 (1996) 545^553. [121] M. Way, B. Pope, A.G. Weeds, Evidence for functional homology in the F-actin binding domains of gelsolin and K-actinin: implications for the requirements for severing and capping, J. Cell Biol. 119 (1992) 835^842. [122] P. Lappalainen, D.G. Drubin, Co¢lin promotes rapid actin ¢lament turnover in vivo, Nature 388 (1997) 78^82. [123] B. Rost, Protein structures sustain evolutionary drift, Curr. Biol. Fold Design 2 (1997) 519^524. [124] N. Yonezawa, E. Nishida, K. Iida, H. Kumagai, I. Yahara, H. Sakai, Inhibition of actin polymerisation by a synthetic dodecapeptide patterned around the actin-binding site of co¢lin, J. Biol. Chem. 266 (1991) 10485^10489. [125] K. Moriyama, N. Yonezawa, H. Sakai, I. Yahara, E. Nishida, Mutational analysis of an actin binding site of co¢lin and characterisation of chimeric proteins between co¢lin and destrin, J. Biol. Chem. 267 (1992) 7240^7244. [126] A. McGough, B. Pope, W. Chui, A. Weeds, Co¢lin changes the twist of F-actin: implications for actin ¢lament dynamics and cellular function, J. Cell Biol. 138 (1997) 771^781. [127] J. Vandekerckhove, J. Van Damme, K. Vancompernolle, M.R. Bubb, P.K. Lambooy, E.D. Korn, The covalent structure of Acanthamoeba actobindin, J. Biol. Chem. 265 (1990) 12801^12805. [128] M.R. Bubb, M.S. Lewis, E.D. Korn, The interaction of monomeric actin with two binding sites on Acanthamoeba actobindin, J. Biol. Chem. 266 (1991) 3820^3826. [129] M.R. Bubb, J.R. Knutson, D.K. Porter, E.D. Korn, Actobindin induces the accumulation of actin dimers that neither nucleate polymerisation nor self-associate, J. Biol. Chem. 269 (1994) 25592^25597. [130] M.R. Bubb, E.D. Korn, Kinetic model for the inhibition of actin polymerisation by actobindin, Biochemistry 34 (1995) 3921^3926. [131] P.K. Lambooy, E.D. Korn, Inhibition of an early stage of actin polymerisation by actobindin, J. Biol. Chem. 263 (1988) 12836^12843.

347

[132] T. Tanaka, M. Tanaka, Y. Mitui, Analysis of expressed sequence tags of the parasitic protozoa Entamoeba histolytica, Biochem. Biophys. Res. Commun. 236 (1997) 611^615. [133] K. Vancompernolle, M. Goethals, C. Huet, D. Louvard, J. Vandekerckhove, G- to F-actin modulation by a single amino acid substitution in the actin binding site of actobindin and thymosin L4, EMBO J. 11 (1992) 4739^4746. [134] R. Wilson, R. Ainscough, K. Anderson, C. Baynes, M. Berks, J. Bon¢eld, J. Burton, M. Connell, T. Copsey, J. Cooper, A. Coulson, M. Craxton, S. Dear, Z. Du, R. Durbin, A. Favello, L. Fulton, A. Gardner, P. Green, T. Hawkins, L. Hillier, M. Jier, L. Johnston, M. Jones, J. Kershaw, J. Kirsten, N. Laister, P. Latreille, J. Lightning, C. Lloyd, A. McMurray, B. Mortimore, M. O'Callaghan, J. Parsons, C. Percy, L. Rifken, A. Roopra, D. Saunders, R. Shownkeen, N. Smaldon, A. Smith, E. Sonnhamer, R. Staden, J. Sulston, J. Thierry-Mieg, K. Thomas, M. Vaudin, K. Vaughan, R. Waterston, A. Watson, L. Weinstock, J. Wilkinson-Sproat, P. Wohldman, 2.2 Mb of contiguous nucleotide sequence from chromosome III of C. elegans, Nature 368 (1994) 32^38. [135] L. Carlsson, L.-E. Nystro«m, I. Sundkvist, F. Markey, U. Lindberg, Actin polymerisability is in£uenced by pro¢lin, a low molecular weight protein in non-muscle cells, J. Mol. Biol. 115 (1977) 465^483. [136] M. Tanaka, H. Shibata, Poly(L-proline)-binding proteins from chick embryos are a pro¢lin and a pro¢lactin, Eur. J. Biochem. 151 (1985) 291^297. [137] I. Lassing, U. Lindberg, Speci¢c interactions between phosphatidylinositol-4,5-bisphosphate and pro¢lactin, Nature 314 (1985) 472^474. [138] L.M. Machesky, S.J. Atkinson, C. Ampe, J. Vandekerckhove, T. Pollard, Puri¢cation of a cortical complex containing two unconventional actins from Acanthamoeba by af¢nity chromatography on pro¢lin agarose, J. Cell Biol. 127 (1994) 107^115. [139] P.J. Goldschmidt-Clermont, M.I. Furman, D. Wachsstock, D. Safer, V.T. Nachmias, T.D. Pollard, The control of actin nucleotide exchange by thymosin L4 and pro¢lin. A potential regulatory mechanism for actin polymerisation in cells, Mol. Biol. Cell 3 (1992) 1015^1024. [140] A.G. Murzin, How far divergent evolution goes in proteins, Curr. Opin. Struct. Biol. 8 (1998) 380^387. [141] C. Staiger, B.C. Gibbon, D.R. Kovar, L.E. Zonia, Pro¢lin and actin-depolymerizing factor: modulators of actin organisation in plants, Trends Plant Sci. 2 (1997) 275^281. [142] I. Perelroizen, D. Didry, H. Christensen, N.H. Chua, M.-F. Carlier, Role of nucleotide exchange and hydrolysis in the function of pro¢lin in actin assembly, J. Biol. Chem. 271 (1996) 12302^12309. [143] M. Pring, A. Weber, M.R. Bubb, Pro¢lin^actin complexes directly elongate actin ¢laments at the barbed end, Biochemistry 31 (1992) 1827^1836. [144] L.E. Laham, J.A. Lamb, P.G. Allen, P.A. Janmey, Selective binding of gelsolin to actin monomers containing ADP, J. Biol. Chem. 268 (1993) 14202^14207.

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart

348

M. Van Troys et al. / Biochimica et Biophysica Acta 1448 (1999) 323^348

[145] M. Arpin, E. Pringault, J. Finidori, A. Garcia, J.-M. Jeltsch, J. Vandekerkhove, D. Louvard, Sequence of human villin: a large duplicated domain homologous with other actin severing proteins and a unique small carboxy-terminal domain related to villin speci¢city, J. Cell Biol. 107 (1988) 1759^1766. [146] D. Safer, M. Elzinga, V.T. Nachmias, Thymosin L4 and Fx, an actin sequestering peptide, are indistinguishable, J. Biol. Chem. 266 (1991) 4029^4032. [147] M. Van Troys, D. Dewitte, M. Goethals, M.-F. Carlier, J. Vandekerckhove, C. Ampe, The actin binding site of thymosin L4 mapped by mutational analysis, EMBO J. 15 (1996) 201^210. [148] E. Friederich, K. Vancompernolle, C. Huet, M. Goethals, J. Finidori, J. Vandekerckhove, D. Louvard, An actin binding site containing a conserved motif of charged amino acid residues is essential for the morphogenic e¡ect of villin, Cell 70 (1992) 81^92. [149] D.S. Doering, P. Matsudaira, Cysteine scanning mutagenesis at 40 of 76 positions in villin headpiece maps the Factin binding site and structural features in the domain, Biochemistry 35 (1996) 12677^12685. [150] J. Zarbock, H. Oschkindt, E. Hannappel, H. Kalbacher, W. Voelter, T.A. Holak, Solution conformation of thymosin L4: a nuclear magnetic resonance and simulated annealing study, Biochemistry 29 (1990) 7814^7821. [151] M. Czisch, M. Schleicher, S. Ho«rger, W. Voelter, T.A. Holak, Conformation of thymosin L4 in water determined by NMR spectroscopy, Eur. J. Biochem. 218 (1993) 335^344. [152] C. Siminel, T. Rose, M. Goethals, J. Vandekerckhove, E. Friederich, D. Louvard, M. Delepierre, Conformational behaviour of a synthetic peptide of the C-terminus of villin

[153]

[154]

[155]

[156]

[157]

[158]

[159]

[160]

that interacts with actin: an NMR, CD and simulated annealing study, Int. J. Pept. Protein Res. 45 (1995) 574^586. M.-F. Carlier, D. Didry, I. Erk, J. Lepault, M. Van Troys, J. Vandekerckhove, I. Perelroizen, H.L. Yin, D. Pantaloni, Thymosin L4 is not a simple G-actin sequestering protein and interacts with F-actin at high concentration, J. Biol. Chem. 271 (1996) 9231^9239. T.P. Stossel, C. Chaponnier, R.M. Ezzell, J.H. Hartwig, P.A. Janmey, D.J. Kwiatkowski, S.S. Lind, D.B. Smith, F.S. Southwick, H.L. Yin, Non-muscle actin binding proteins, Annu. Rev. Cell Biol. 1 (1985) 353^402. T.D. Pollard, J.A. Cooper, Actin and actin binding proteins. A critical evaluation of mechanisms and functions, Annu. Rev. Biochem. 55 (1986) 987^1035. A.C.R. Martin, C.A. Orengo, E.G. Hutchinson, S. Jones, M. Karmirantzou, R.A. Laskowski, J.B.O. Mitchell, C. Taroni, J.M. Thornton, Protein folds and function, Structure 6 (1998) 875^884. H.W. Mewes, K. Albermann, M. Bahr, D. Frishman, A. Gleissner, J. Hani, K. Heumann, K. Kleine, S.G. Oliver, F. Pfei¡er, A. Zollner, Overview of the yeast genome, Nature 387 (6632 Suppl.) (1997) 7^65. A. Rodriguez Del Castillo, S. Lemaire, L. Tchakarov, M. Jeyapragasan, J.-P. Doucet, M.L. Vitale, J.-M. Trifaro, Chroma¤n cell scinderin, a novel calcium dependent actin ¢lament severing protein, EMBO J. 9 (1990) 43^52. J.R. Glenney Jr., K. Weber, Calcium control of micro¢laments: uncoupling of the F-actin-severing and -bundling activity of villin by limited proteolysis in vitro, Proc. Natl. Acad. Sci. USA 78 (1981) 2810^2814. J.P. Baker, M.A. Titus, Myosins: matching functions with motors, Curr. Opin. Cell Biol. 10 (1998) 80^86.

BBAMCR 14420 6-1-99 Cyaan Magenta

Geel

Zwart