REVIEW ARTICLE Thermal unfolding and aggregation of actin Stabilization and destabilization of actin filaments Dmitrii I. Levitsky 1,2 , Anastasiya V. Pivovarova 1,3 , Valeria V. Mikhailova 1 and Olga P. Nikolaeva 2 1 A. N. Bach Institute of Biochemistry, Russian Academy of Sciences, Moscow, Russia 2 A. N. Belozersky Institute of Physico-Chemical Biology, Moscow State University, Russia 3 School of Bioengineering and Bioinformatics, Moscow State University, Russia Actin is one of the most abundant and highly con- served proteins found in all eukaryotic cells. It is involved in many different cellular processes that are essential for growth, differentiation and motility. Moreover, bacterial homologs of eukaryotic actin, ParM and MreB, have recently been identified. It is now clear that prokaryotic cells also possess actin and that a dynamic, actin-like cytoskeleton is involved in a variety of essential cellular processes in bacteria [1]. The 43 kDa actin monomer (globular actin or G-actin) spontaneously assembles in vitro to form long polar filaments (filamentous actin, or F-actin) upon the addition of neutral salts (usually 50–100 mm KCl, 2–4 mm MgCl 2 , or both). Actin filaments have a crucial role in biological motility as the main partners of the myosin-based motor systems and as the major constituent of the cytoskeleton. The polymerization of G-actin into F-actin is accompanied by the hydrolysis Keywords actin; actin filaments; cofilin; differential scanning calorimetry; heat-induced aggregation; inorganic phosphate analogs; phalloidin; small heat shock proteins; thermal stability; thermal unfolding Correspondence D. I. Levitsky, A.N. Bach Institute of Biochemistry, Russian Academy of Sciences, Leninsky Prosp. 33, 119071 Moscow, Russia Fax: +7 495 954 2732 Tel: +7 495 952 1384 E-mail: levitsky@inbi.ras.ru (Received 14 April 2008, revised 31 May 2008, accepted 24 June 2008) doi:10.1111/j.1742-4658.2008.06569.x Actin is one of the most abundant proteins in nature. It is found in all eukaryotes and plays a fundamental role in many diverse and dynamic cellular processes. Also, actin is one of the most ubiquitous proteins because actin-like proteins have recently been identified in bacteria. Actin filament (F-actin) is a highly dynamic structure that can exist in different conforma- tional states, and transitions between these states may be important in cyto- skeletal dynamics and cell motility. These transitions can be modulated by various factors causing the stabilization or destabilization of actin filaments. In this review, we look at actin stabilization and destabilization as expressed by changes in the thermal stability of actin; specifically, we summarize and analyze the existing data on the thermal unfolding of actin as measured by differential scanning calorimetry. We also analyze in vitro data on the heat- induced aggregation of actin, the process that normally accompanies actin thermal denaturation. In this respect, we focus on the effects of small heat shock proteins, which can prevent the aggregation of thermally denatured actin with no effect on actin thermal unfolding. As a result, we have pro- posed a mechanism describing the thermal denaturation and aggregation of F-actin. This mechanism explains some of the special features of the thermal unfolding of actin filaments, including the effects of their stabilization and destabilization; it can also explain how small heat shock proteins protect the actin cytoskeleton from damage caused by the accumulation of large insoluble aggregates under heat shock conditions. Abbreviations AlF 4 ) , anions of aluminum fluoride; BeF x , anions of beryllium fluoride; DSC, differential scanning calorimetry; sHSP, small heat shock proteins; T m , midpoint of thermally induced unfolding transition. 4280 FEBS Journal 275 (2008) 4280–4295 ª 2008 The Authors Journal compilation ª 2008 FEBS of bound ATP followed by a slower release of P i ;asa result, each F-actin protomer contains either tightly bound ADP or ADP-P i . In vivo, actin polymerization is a highly regulated process controlled not only by nucleotide binding and hydrolysis, but also by the action of a number of actin-binding proteins that can nucleate, cleave, cross-link, bundle, stabilize or desta- bilize the filaments [2,3]. Monomeric G-actin is a globular protein consisting of a single polypeptide chain of 375 residues. It contains one tightly bound nucleotide, ATP or ADP, and a single, high-affinity divalent cation, Ca 2+ or Mg 2+ .In order to be crystallized, actin must be rendered non-polymerizable and, thus, almost all actin crystal structures available to date either represent complexes of G-actin with actin-binding proteins (e.g. DNase I [4], gelsolin [5] or profilin [6]) and organic toxins [7], or were obtained from G-actin that had been changed in differ- ent ways (e.g. by chemical modifications [8–10], proteol- ysis [11] or mutation [12]). The first atomic resolution (2.8 A ˚ ) structure of actin co-crystallized with DNase I was published in 1990 by Kabsch et al. [4]. This publica- tion, as well as other 3D structures of actin, revealed that it consists of two easily distinguishable domains separated by a deep cleft, each domain being subdivided into two subdomains (Fig. 1A). The nucleotide is bound in the cleft between the two domains. Subdomains 1 and 2 comprise the so-called ‘small’ domain and are linked to subdomains 3 and 4 (the so-called ‘large’ domain) by a ‘hinge’ or connecting piece between subdomains 1 and 3. In fact, the ‘small’ domain is not significantly smaller than the ‘large’ domain [13]. Numerous solution studies suggest that the nucleotide-binding cleft between the two domains can exist in two main states, closed and open, and nucleotide-induced conformational changes in G-actin are associated with a transition between these two states of the cleft. However, among the numerous crystal structures that have been published, significant opening of the nucleotide-binding cleft has been observed only in profilin-bound actin crystals [14]. It is possible that crystal packing interactions favor a closed state for G-actin even though the state of the cleft in solution may be shifted by the bound nucleotide [11,12,15]. Actin normally exists in the monomeric G-form only in solutions with a very low ionic strength, and it poly- merizes rapidly upon the addition of 100 mm KCl with the formation of long polar filaments of F-actin, which are double-stranded spiral polymers of actin molecules (Fig. 1B). Polymerization of actin fully prevents its crystallization which needs high protein and salt concentrations. As a result, the atomic resolution structure for F-actin remains unknown, and only models are available. The most important model, generated by Holmes et al. [16], used a rotational and translational search to place the G-actin crystal D D D D T D D D D D D D DD D D D D D D D D D D D D D D D D T T T D D D D DD D D D D D D D T D DT D D D D D D T T T T T T D D D D T T T T T T D ATP (T) P i D D (+) (–) ADP (D) A B Fig. 1. (A) Three-dimensional atomic structure of monomeric G-actin shown in approximately the same orientation as originally illustrated by Kabsch et al. [4] (Protein Data Bank ID code 1ATN). This actin structure was obtained from co-crystals of actin–DNase I from which the DNase I component has been removed [4]. The four subdomains are indicated by the numbers encircled. D-loop, DNase- I-binding loop. A molecule of nucleotide bound in the nucleotide- binding cleft is indicated as ATP, and the metal cation (Ca 2+ or Mg 2+ ) is indicated as a sphere. (B) Schematic representation of actin polymerization and treadmilling of actin filaments. Actin mole- cules are shown by circles. Monomeric G-actin contains bound ATP (T inside the circles). Actin polymerization includes three steps, namely monomer activation, nucleation and filament elongation, the latter being accompanied by ATP hydrolysis. Actin filaments initially grow with terminal subunits containing ATP; at later stages, after ATP hydrolysis, subunits containing ADP-P i transiently accumulate (not shown here); at steady states, the filament is made of subunits containing tightly bound ADP (denoted as D inside the circles) except for terminal subunits at the barbed end of the filament (denoted as ‘+’) that contain ATP or ADP-P i . Polar actin filaments depolymerize from their pointed ends (denoted as ‘)’). After disso- ciation from the filament and replacement of bound ADP by ATP, actin monomers can again bind to the barbed ends of the filaments. D. I. Levitsky et al. Actin unfolding and aggregation FEBS Journal 275 (2008) 4280–4295 ª 2008 The Authors Journal compilation ª 2008 FEBS 4281 structure into a helical filament so as to best match the observed X-ray actin fiber diffraction pattern from an oriented F-actin gel. The original ‘Holmes model’ [16] and its variations [17] contain the explicit assumption that no large-scale conformational change is needed between a G-actin monomer and an F-actin protomer. The only change in actin structure that was introduced involved the movement of a hydrophobic loop (resi- dues 264–273 in subdomain 4) from the body of the subunit to form a contact with subunits on the oppo- site strand of a double-stranded spiral polymer [15]. According to the Holmes model of F-actin [16], sub- domains 3 and 4 of each actin protomer are located close to the filament axis, whereas subdomains 1 and 2 are at high radius near the surface of the filament. Because of the helical symmetry of F-actin, each pro- tomer in the filament may be in contact with four adjacent protomers [13]. Residues involved in the actin–actin contacts are located in all four subdomains of the protomer. For example, there are contacts between several sites in subdomain 3 and sites on subdomains 4 and 2, as well as contact between sub- domains 1 and 4 [13], and the DNase I-binding loop (D-loop) at the top of subdomain 2 is proposed to interact with the C-terminal region in subdomain 1 of the adjacent protomer [18,19] (Fig. 1A). A growing body of data suggests that the actin fila- ment does not exist in a single state, but can be quite dynamic [15]. One of the most remarkable properties of F-actin is that although subunits have a very fixed axial rise in the filament (of  27 A ˚ ), the rotation between adjacent subunits can be quite variable [20]. An actin-binding protein, cofilin, has been shown to change the average twist of subunits in F-actin by  5 o per subunit (from  167 to  162 o ) [21]. How- ever, detailed analysis of pure actin filaments has sug- gested that cofilin stabilizes an already existing twisted state of F-actin that can be formed spontaneously in the absence of other proteins [22]. Another dynamic mode within F-actin involves the ability of subunits to undergo a substantial tilt (up to  30 o ) [15]. This tilt was first observed within actin filaments decorated by cofilin [22], but was subsequently observed within F-actin in the absence of other proteins [23]. Thus, according to recent studies, F-actin can exist in different structural states, and transitions between these states may play an important role in cytoskeletal dynamics and in the contractile cycle of actomyosin [15]. Many factors (such as the binding of nucleotides, drugs and actin-binding proteins) may modulate such transitions, i.e. isomerizations of F-actin from one structural state to another, thus stabilizing or destabi- lizing the actin filaments. In this review, we look at actin stabilization or destabilization as expressed by changes in its thermal stability; specifically, we describe and analyze existing data on the thermal unfolding of actin as measured using differential scanning calorimetry (DSC), which is the most effective and commonly used method to study the thermal unfolding of proteins [24,25]. In the fol- lowing sections, we summarize and discuss the effects of stabilization or destabilization for both G-actin and F-actin that are expressed by an increase or decrease in actin thermal stability. We also summarize the in vitro data on the heat-induced aggregation of actin, a process that normally accompanies actin thermal unfolding. In this respect, we focus mainly on the effects of small heat shock proteins (sHSPs), which can effectively prevent the aggregation of thermally dena- tured actin with no effect on its thermal unfolding. Thermal unfolding of monomeric G-actin In 1984 Tatunashvili and Privalov [26] used DSC to investigate the thermal denaturation of monomeric G-actin and suggested the presence of at least two inter- acting domains in the molecule. The existence of two domains in the G-actin molecule was also proposed by Bertazzon et al. [27] after computer deconvolution of the G-actin heat sorption curve into two individual ther- mal transitions. It should be noted that the thermal denaturation of actin is irreversible, and the use of such approaches to analyze irreversible thermal transitions is rather controversial [28]. However, the existence of a domain structure in the G-actin molecule, initially pro- posed from the DSC data [26,27], was confirmed by the 3D atomic structure of the G-actin published in 1990 [4], and showed the presence of two easily distinguish- able domains separated by a deep cleft (Fig. 1A). The D-loop at the top of subdomain 2 (Fig. 1A) seems to play a very important role in the thermal stability of G-actin. Binding of DNase I to G-actin strongly increased the thermal stability of both proteins. Separate proteins had thermal transitions with single maxima of  61 °C for G-actin and  56 °C for DNase I. In the complex, both proteins were mutually stabilized and were denatured as a unit, resulting in a new sharp thermal transition with maximum at 70 °C (data not shown). A very similar effect was observed in DSC studies on the binding of cofilin to G-actin [29,30]. In this case, both interacting proteins also formed a complex in which they stabilized each other and denatured together resulting in a new, highly cooperative thermal transition with maximum at 66.4 °C [29] or  68 °C Actin unfolding and aggregation D. I. Levitsky et al. 4282 FEBS Journal 275 (2008) 4280–4295 ª 2008 The Authors Journal compilation ª 2008 FEBS [30]. Thus, as in the case with DNase I, the binding of cofilin to G-actin significantly increased the thermal stability of both proteins. According to recent views, cofilin binds preferentially to G-actin either in the cleft between subdomains 1 and 3 [31] or in the cleft between subdomains 1 and 2 [32], or it may bind in both these sites [31]. The D-loop in subdomain 2 is also proposed to be involved in a weak binding of cofilin to G-actin [31,33]. Cofilin binding induces closure of the nucleotide-binding cleft of G-actin [32,34], and this closure, together with changes in the D-loop, may be one of the main reasons for the increased thermal stability of G-actin. In favor of this assumption are the DSC data on the thermal unfolding of G-actin complexed with thymosin b 4 [35]. It has been proposed that this small protein, a member of a family of actin-sequestering proteins, binds to subdomains 1 and 3 on actin, as well as to the D-loop in subdomain 2, and this binding causes changes in the spatial orienta- tion of G-actin subdomains, closing the nucleotide-bind- ing cleft [35,36]. It has been shown using DSC that the binding of thymosin b 4 results in the thermal stabiliza- tion of G-actin, shifting its thermal transition by 1.8 °C towards a higher temperature and making the transition more cooperative [35]. However, the opposite effect was observed by DSC with G-actin specifically cleaved by a bacterial ECP32 protease which has been shown to cleave actin at a single site between Gly42 and Val43 within the D-loop [37]. It has been shown that cleavage of G-actin with ECP32 dramatically decreases its thermal stability, lowering the thermal transition (T m ) by 7–8 °C (A. V. Pivovarova, S. Yu. Khaitlina & D. I. Levitsky, unpublished results). These DSC results are in good agreement with data showing a more open conformation for the nucleotide-binding cleft between two domains in ECP32-cleaved actin, as evidenced by the increased nucleotide exchange rate and its higher susceptibility to limited proteolysis [38,39]. It thus seems possible that the thermal stability of G-actin is mainly determined by the conformational state of the nucleotide-binding cleft between the two domains in the actin molecule. The thermal stability of G-actin increases when the cleft is closed (e.g. due to the binding of various proteins to the D-loop), whereas opening of the cleft leads to significant destabilization of G-actin. Thermal unfolding of F-actin filaments Polymerization of G-actin to F-actin The use of DSC allows very clear probing of the changes caused by actin polymerization, i.e. the trans- formation of monomeric G-actin into F-actin filaments (Fig. 1B). These changes, observed by many authors [27,30,40–42], are expressed as a significant increase in the denaturation temperature and a sharp change in the shape of the peak (the peak becomes much narrower, indicating a significant increase in the coo- perativity of thermal denaturation) (Fig. 2). It is evi- dent that the changes in thermal unfolding that accompany actin polymerization are due to numerous contacts established between adjacent actin protomers in the filament. Effects of nucleotides on the thermal unfolding of F-actin Although bound nucleotide is not required for actin polymerization [43], it is very important for the stabil- ization of actin. It is well known that actin lacking bound nucleotide is very unstable and denatures easily [44] unless protected by high sucrose concentrations, which permit actin to retain its stability and the ability to form filaments of F-actin [43,45]. Under normal conditions, ADP is tightly bound in F-actin subunits and is unable to exchange with free nucleotides. Surprisingly, our DSC experiments have shown that 40 50 60 70 80 90 100 0 50 100 150 ΔC p (kJ·mol –1 ·K –1 ) Temperature (°C) G-actin F-actin F-actin + phalloidin F-actin + AlF 4 – F-actin + phalloidin + AlF 4 – Fig. 2. Temperature dependences of the excess heat capacity (DC p ) of G-actin, F-actin in the presence of 1 mM ADP and F-actin stabilized by phalloidin or AlF 4 ) , or simultaneously by both these stabilizers. The actin concentration was 1 mgÆmL )1 . Other condi- tions: for G-actin, G-buffer (2 m M Tris ⁄ HCl, pH 8.0, 0.2 mM ATP, 0.2 m M CaCl 2 , 0.5 mM b-mercaptoethanol and 1 mM NaN 3 ); for F-actin, 30 m M Hepes, pH 7.3, containing 100 mM KCl, 1 mM MgCl 2 and 1 mM ADP. Concentrations of stabilizers: 33 lM phalloi- din and 0.5 m M AlF 4 ) (5 mM NaF and 0.5 mM AlCl 3 ). Heating rate 1KÆmin )1 . Adapted from Levitsky et al. [41] and Levitsky [54] with some changes and additions. D. I. Levitsky et al. Actin unfolding and aggregation FEBS Journal 275 (2008) 4280–4295 ª 2008 The Authors Journal compilation ª 2008 FEBS 4283 the thermal stability of F-actin depends strongly on the concentration of ADP added. The transition tem- perature (T m ) of F-actin increased by 5–6 °C with increasing ADP concentrations up to 1 mm, and reached a plateau at higher ADP concentrations, a half-maximum increase in T m being observed in the presence of 0.1 mm ADP [46]. The stabilizing effect of ADP was highly specific, being observed only with ADP, and not with other nucleoside diphosphates (IDP, UDP, GDP, CDP). A similar effect of the F-actin thermal stabilization was also seen in the presence of ATP; however, this effect was much less specific, and was observed, although to a lesser extent, for other nucleoside triphosphates (ITP, UTP and GTP). Another difference between the effects of ATP and ADP was that an increase in ATP concentration from 1 to 5 mm led to further significant increase in the thermal stability of F-actin (T m increased by >3°C), whereas a similar increase in ADP concentra- tion had no influence on the thermal unfolding of F-actin [46]. These findings suggest that the stabilizing effect of ATP [46,47] differs significantly in its mecha- nism from the effect caused by ADP [46]. It seems very likely that the stabilizing effect of ATP and other nucleoside triphosphates is caused not only by their binding to high-affinity sites, but also by their interaction with some additional (‘second’) nucleotide- binding site on F-actin. The presence of a second, low-specificity and low-affinity nucleotide-interacting site on actin has been postulated by some studies [47– 50]. Nucleotide binding in this site occurred at milli- molar concentrations, ITP and CTP being even more effective than ATP [50]. It is important to note that the most striking changes in the thermal stability of F-actin were observed at low concentrations of added nucleotide (0.1–0.2 mm), and under these conditions the effects of ADP and ATP were very similar [46]. It is difficult to explain these effects by some interaction of ADP or ATP with the second nucleotide-binding site on F-actin, which needs much higher concentrations of nucleotide. It seems most likely that these effects, which are observed at low ADP or ATP concentra- tions, are caused by their binding with highly specific, high-affinity sites in the nucleotide-binding cleft of actin subunits, whereas the additional stabilization of actin filaments, which occurs at high ATP concentra- tions, is caused by ATP binding to low-affinity sites that are able to bind ATP and other nucleoside triphosphates, but not ADP. It seems to us that the most likely explanation for the stabilizing effect of nucleotides on the thermal unfolding of F-actin is as follows. We have proposed that irreversible thermal unfolding of F-actin is pre- ceded by a reversible stage involving nucleotide (ADP) dissociation from specific nucleotide-binding sites in actin subunits [46]. Reversible dissociation of ADP from actin subunits probably occurs only upon heat- ing, just before the irreversible denaturation of the protein. In the absence of free nucleotides, actin-bound ADP dissociates easily and, as a result, actin denatures easily and rapidly. However, the presence of free nucle- otide (ADP or ATP) in solution prevents the dissocia- tion of actin-bound ADP, and this may explain the nucleotide-induced increase in the thermal stability of F-actin. Stabilization of F-actin by phalloidin It is well known that F-actin interacts specifically with a cyclic heptapeptide, phalloidin – one of the principal toxins of the mushroom Amanita phalloides. Phalloidin binds to F-actin with very high affinity and causes sub- stantial stabilization of the actin filaments, preventing the depolymerization of F-actin and protecting it from proteolytic cleavage. Therefore, actin filaments stabi- lized by phalloidin are often used in experiments, e.g. in vitro motility assays. Le Bihan and Gicquaud [51] showed that the binding of phalloidin to F-actin signif- icantly increases the temperature of F-actin thermal denaturation, raising the temperature at which thermal transition occurs by 14 °C. This effect of phalloidin (Fig. 2), also observed by us [41,52] and other authors [53], is very useful for DSC studies on F-actin com- plexes with other proteins, and we often use phalloi- din-stabilized F-actin to obtain a better separation of the thermal transitions of F-actin and actin-bound proteins (e.g. myosin head, tropomyosin) on DSC curves [54]. Phalloidin binds to F-actin at the interface of three adjacent actin protomers [17,55] and stabilizes lateral interactions between the two filament strands. These physical contacts appear to be the most important factors for the thermal stabilization of F-actin by phalloidin. It has been shown using electron micros- copy that phalloidin can stabilize F-actin with a high cooperativity, at a 1 : 20 molar ratio with actin [56]. The cooperativity of the phalloidin-induced stabil- ization of F-actin has also been observed in DSC experiments [57]. Stabilization of F-actin by P i analogs During the ATP hydrolysis that normally accompanies actin polymerization, intermediate states are formed in actin subunits, which are actin complexes with ATP or Actin unfolding and aggregation D. I. Levitsky et al. 4284 FEBS Journal 275 (2008) 4280–4295 ª 2008 The Authors Journal compilation ª 2008 FEBS ADP-P i . These intermediate states can be studied using stable complexes of F-actin with beryllium fluoride (BeF x ) or aluminum fluoride (AlF 4 ) ) anions, which have been found to be good structural analogs of P i . Com- beau and Carlier [58] found that AlF 4 ) and BeF x (BeF x stands for the BeF 3 ) and BeF 2 (OH) ) complexes) bind strongly to F-actin with an affinity three orders of mag- nitude higher than P i and compete with P i for binding to the nucleotide-binding cleft of ADP–F-actin protom- ers in the place of the c-phosphate of ATP. Both these P i analogs strongly stabilize F-actin by decreasing the rate of protomer dissociation from filaments by 150-fold [58]. In the case of BeF x , this stabilization was shown to be associated with structural changes in subdomain 2, as indicated by the strong inhibition of its proteolytic cleavage [59] and by electron microscopy [60]. Stabilization of actin filaments by BeF x or AlF 4 ) can be clearly seen in DSC studies on the thermal unfolding of F-actin (Fig. 2). It has been shown that both these P i analogs increase the thermal stability of F-actin drastically, increasing the thermal transition temperature by more than 16 °C [41,61]; the effects of BeF x and AlF 4 ) are very similar [61]. Additive effect of F-actin stabilization by phalloidin and P i analogs It is interesting to note that phalloidin and AlF 4 ) (Fig. 2) (or BeF x ) [41] have similar effects on the thermal denaturation of F-actin, and these effects are expressed as a significant increase of the transition temperature. However, when we simultaneously added both phalloi- din and BeF x (or AlF 4 ) ) to F-actin we observed an addi- tional stabilization of F-actin expressed as an increase in the thermal transition temperature of more than 25 °C [41] (Fig. 2). This means that phalloidin and BeF x (or AlF 4 ) ) stabilize F-actin in different ways independent of each other and most likely affect different sites on the actin molecule. This is consistent with the literature data that phalloidin and BeF x bind to different sites on actin. Indeed, bound phalloidin is located at the contact region of three actin subunits thus stabilizing their interactions [55], whereas BeF x (or AlF 4 ) ) binds to the nucleotide- binding cleft, affects the structure of subdomain 2 and its interaction with the adjacent longitudinal protomer in ADP–F-actin [59,60], and probably favors a closed state for the cleft. Stabilization and destabilization of F-actin by cofilin Very little is known how the actin-binding proteins affect the thermal unfolding of F-actin. One of the first studies in this direction was our attempt to apply the DSC method to investigate the thermal unfolding of F-actin in complexes with cofilin, a small actin-binding protein belonging to the actin-depolymerizing factor (ADF) ⁄ cofilin family of proteins. ADF ⁄ cofilin proteins have attracted much attention because of their important role in regulating actin dynamics in cells [2,3,62,63]. These proteins can depo- lymerize [64] and sever [65,66] actin filaments by weak- ening longitudinal [22,67] and lateral [68,69] interprotomer contacts in F-actin. It has been shown that the binding of cofilin induces a transition in sub- domain 2, which is accompanied by disordering of the D-loop [70], and these cofilin-induced conformational changes in F-actin expose subdomain 2 to proteolysis [71]. Also, cofilin binding was shown to change the twist of F-actin by decreasing the rotation of one protomer with respect to its neighbors [21]. All these conformational changes in the actin filament are predicted to lead to filament destabilization. It was shown for the first time in our DSC experi- ments [29] and then confirmed by Bobkov et al. [30] that cofilin has a dual effect on the thermal unfolding of F-actin, depending on the molar ratio of cofilin to actin. At saturating concentrations, cofilin strongly increases the thermal stability of F-actin increasing the temperature at which thermal transition occurs by 7 °C and increasing the cooperativity of the transition (Fig. 3) [29]. The stabilizing effect of cofilin on F-actin was very similar to that observed with G-actin in the presence of cofilin [29,30]. This suggests that cofilin binding to F-actin subunits induces conformational changes similar to those seen in monomeric G-actin (presumably the closure of the nucleotide-binding cleft) [32], and these changes result in a significant increase in the thermal stability of F-actin. According to mod- els of cofilin–F-actin complexes derived from electron microscopy studies [21–23,70], cofilin binds to subdo- main 2 of a lower protomer and subdomain 1 of an upper protomer in F-actin and overlaps the inter- protomer interface between these subdomains. This interprotomer interaction may also contribute to the cofilin-induced thermal stabilization of F-actin. Interestingly, the stabilization of F-actin by cofilin at saturating concentrations fully abolished the effects of the other F-actin stabilizers, phalloidin or AlF 4 ) [29]. In the case of AlF 4 ) , this is consistent with a recent observation that cofilin can dissociate BeF x (another P i analog, which is very similar to AlF 4 ) in its effects on F-actin) [29,58] from the actin filament, whereas BeF x does not bind to F-actin saturated with cofilin, presumably because of the cofilin-induced changes in the nucleotide-binding cleft of F-actin D. I. Levitsky et al. Actin unfolding and aggregation FEBS Journal 275 (2008) 4280–4295 ª 2008 The Authors Journal compilation ª 2008 FEBS 4285 subunits [72]. Phalloidin, unlike BeF x and AlF 4 ) , can bind to F-actin that is fully saturated with cofilin and compete with cofilin for F-actin [72]. However, cofilin- induced structural changes in the actin filament (such as the changes in twist [21] and the weakening of lateral contacts in the filament [68]) may alter the contacts between actin protomers in the filament and prevent the simultaneous binding of phalloidin with three adjacent protomers, thereby abolishing its stabi- lizing effect on F-actin. Contrary to the stabilizing effect of cofilin on F-actin thermal unfolding at saturating concentrations of cofi- lin, a strong decrease in the thermal stability of F-actin was observed at sub-saturating concentrations of cofilin [29,30]. Under these conditions, the temperature at which the thermal transition of F-actin occurred was lowered by 1–6 °C, depending on the cofilin ⁄ actin monomer molar ratios. The most pronounced effect was observed at cofilin to actin molar ratios between 1 : 6 and 1 : 1.5. Under these conditions, two peaks were simultaneously observed: a cofilin-stabilized F-actin peak at 65–67 °C and a cofilin-destabilized peak at 56–58 °C (Fig. 3) [29]. On decreasing the cofilin ⁄ actin monomer molar ratio, the cofilin-stabilized peak decrea- sed, whereas the cofilin-destabilized peak increased. At low cofilin ⁄ actin monomer molar ratios (< 1 : 12) only the destabilized peak was observed (Fig. 3) and its maximum shifted to higher temperatures as the cofilin ⁄ actin molar ratio decreased [29]. The destabilizing effect of cofilin was highly cooperative as it was observed even at cofilin ⁄ actin molar ratios as low as 1 cofilin per 100–200 actin protomers [29,30]. It has been suggested from these DSC results that cofilin, when bound to F-actin, stabilizes only those actin subunits to which it binds directly, whereas it destabilizes with a very high cooperativity neighboring regions of the actin filament that are free of cofilin [29]. This is consistent with electron microscopy observations showing that actin subunits within cofilin-free regions of cofilin-decorated actin filaments differ in their orientation from those in undecorated F-actin [22]. The DSC data also suggest that cofilin-induced changes in the conformation of F-actin can be propagated over a long distance along the filament from subunits stabilized by cofilin to sub- units free of cofilin. Obviously, these changes become weaker at low cofilin concentrations, i.e. when the average separation between cofilin molecules bound to F-actin increases. Cofilin-induced destabilization of F-actin was fully prevented by the addition of phalloidin or AlF 4 ) [29], thus indicating that actin subunits destabilized by cofilin retained their ability to bind other stabilizers of F-actin. Interestingly, the cofilin-induced destabilization of F-actin was observed by DSC even with F-actin, in which Gln41 in the D-loop on subdomain 2 was cross- linked to Cys374 near the C-terminus on subdomain 1 of the adjacent protomer within the same strand of the long-pitch helix [30]. This may suggest that cofilin, when it binds to F-actin at low molar ratios, weakens the lateral contacts between actin subunits rather than the longitudinal contacts. The destabilizing effect of cofilin demonstrated by DSC studies may play an important role in actin dynamics in living cells. In particular, it may be impor- tant to provide a possible molecular mechanism for the actin-severing and depolymerizing activities of cofilin. Destabilization of F-actin by formins Members of the formin family of proteins are known to be key regulators of actin polymerization that nucle- ate actin filaments and play essential roles in the regu- lation of the actin cytoskeleton. It has been shown using DSC that the binding of formins to F-actin decreases the F-actin thermal stability slightly, lower- ing the T m by 1.5 °C [73]. This effect was observed at a low formin ⁄ actin molar ratio (1 : 20). The authors assigned this decrease in the thermal stability of F-actin to an increase in the flexibility of the filament. 40 50 60 70 0 10 20 30 40 50 F-actin (25 µ M) F-actin + cofilin (1 µ M) F-actin + cofilin (8 µ M) F-actin + cofilin (32 µ M) Tem p erature (°C) Apparent heat capacity ( µ W) Fig. 3. DSC curves of F-actin alone and in complexes with cofilin obtained at various cofilin ⁄ actin monomer molar ratios. The F-actin concentration (25 l M) was held constant, and the cofilin concentra- tions were as indicated for each curve. Other conditions: 30 m M Hepes, pH 7.3, 2 m M MgCl 2 and 0.2 mM ADP. Heating rate 1KÆmin )1 . Adapted from Dedova et al. [29] . Actin unfolding and aggregation D. I. Levitsky et al. 4286 FEBS Journal 275 (2008) 4280–4295 ª 2008 The Authors Journal compilation ª 2008 FEBS It has been proposed that formins can regulate actin filament flexibility via long-range allosteric interactions in the filament. Other interactions affecting F-actin thermal stability Actin filaments can interact with many other proteins. First, interaction of F-actin with myosin heads and muscle regulatory proteins (tropomyosin and troponin in striated muscle or caldesmon and calponin in smooth muscle) plays a key role in the molecular mechanism and regulation of muscle contraction. We used the DSC method to investigate how these proteins affect the thermal stability of F-actin. Although the binding of myosin heads to F-actin significantly increased the thermal stability of myosin, it had no appreciable influence on the thermal unfold- ing of F-actin either in the absence [74] or presence [41,75] of phalloidin, presumably because the thermal denaturation of actin-bound myosin heads occurred at a much lower temperature than that of F-actin. Also, we did not observe any changes in the thermal unfolding of F-actin in complex with smooth muscle tropomyosin [52], caldesmon and calponin, as well as in complex with skeletal a-tropomyosin [76,77]. In the case of tropomyosin, this is easily explained by the fact that it dissociates from F-actin and denatures before the thermal denaturation of F-actin [52,76,77]. However, we observed a pronounced change in the thermal denaturation of F-actin upon addition of troponin I, one of the components of the troponin complex. These changes were expressed as a significant decrease in the enthalpy and cooperativity of the melting of F-actin [76]. The data indicated that there is a direct interaction between troponin I and F-actin. The effect was much weaker if troponin I was added to F-actin stabilized by AlF 4 ) [76]. Apparently, stabilization of F-actin by AlF 4 ) prevents the effects of troponin I on the actin filament structure. A very interesting effect was observed by Gicquaud [78] who used the DSC method for studies on the direct interaction of F-actin with membrane lipids. It has been shown that due to the interaction with liposomes, actin undergoes to a major conformational change resulting in the complete disappearance of its thermal transition on the DSC profile. Proposed mechanism for the thermal denaturation of F-actin It is obvious that the thermal denaturation of such a complicated system as the actin filament cannot be explained by simple models that describe the thermal unfolding of proteins. We attempted to describe a mechanism for F-actin thermal denaturation using special DSC approaches. It is known that, upon irreversible thermal denatur- ation of many oligomeric proteins and enzymes, the maximum thermal transition temperature increases remarkably with the increase in protein concentration. Recent views, such as the dependence of T m on the protein concentration, suggest the presence of a revers- ible dissociation stage for the subunits of the oligomer prior to their irreversible denaturation [79,80]. We applied this approach to F-actin and have shown that the T m value depends strongly on the protein concen- tration. The T m of the F-actin thermal transition increased by more than 3 °C when the concentration of F-actin was increased from 0.5 to 2.5 mgÆmL )1 [46]. A similar dependence of the T m value on the protein concentration was demonstrated for F-actin stabilized by phalloidin: in this case, the T m increased from 81 to 84 °C as the actin concentration increased from 0.5 to 2.0 mgÆmL )1 . However, such dependence was much less pronounced in the presence of AlF 4 ) ; in this case, the T m value increased by only 1.1 °C. As expected, for monomeric G-actin, the T m value for its thermal transition was independent of the protein concentra- tion [46]. These results are consistent with a dissocia- tive mechanism proposed for the irreversible thermal denaturation of other oligomeric proteins [79,80]. Combining these DSC results with the above- described effects of ADP on the thermal unfolding of F-actin, we proposed the following dissociative mecha- nism for the thermal denaturation of F-actin [46]. We propose that at least two reversible stages precede the irreversible thermal unfolding of F-actin. One is the dissociation of nucleotide (ADP) from the nucleotide- binding sites of actin subunits, and the other is the fragmentation of actin filaments or the dissociation of relatively short oligomers from the filament. The proposed mechanism for the thermal denaturation of actin filaments is described in Fig. 4. During heating, destabilization of actin filaments occurs, leading to increased mobility of the filament and weakening of the bonds between subunits. As a result, short oligomers dissociate from one end of the polar actin filament [presumably from the pointed (‘)’) end where the dissociation rate is much higher than on growing barbed (‘+’) end of the fila- ment]. These oligomers either lose bound ADP and then immediately denature and aggregate, or the ADP-containing oligomers may bind again to the actin filament, but they most likely bind to another filament and to the other end (the ‘+’-end). This D. I. Levitsky et al. Actin unfolding and aggregation FEBS Journal 275 (2008) 4280–4295 ª 2008 The Authors Journal compilation ª 2008 FEBS 4287 mechanism (Fig. 4) explains why the thermal unfold- ing of F-actin depends on both ADP and protein concentration. The presence of excess free ADP in solution impedes the thermally induced reversible dis- sociation of tightly bound ADP from actin subunits, whereas the increase in the protein concentration increases the number of actin filaments and, corre- spondingly, the number of ‘+’-ends to which short actin oligomers can bind. It is important to note that two F-actin stabilizers, phalloidin and AlF 4 ) , differ from one another in the extent to which they influence the dependence of T m on the protein concentration [46]. Both stabilizers increase the thermal stability of F-actin substantially (Fig. 2), although independently of each other. Phalloi- din increases the thermal stability of F-actin by strengthening the bonds between adjacent subunits in the actin filament, whereas AlF 4 ) demonstrates a simi- lar effect by trapping ADP in the nucleotide-binding site of actin. Thus, the complex ADP–AlF 4 ) –actin, which mimics the ADP–P i –actin intermediate state of actin polymerization, should prevent the dissociation of ADP from actin subunits and, by contrast, stimu- late the binding of actin monomers or short oligomers to the barbed (‘+’) end of the filament. According to the mechanism proposed above, this means that AlF 4 ) may influence both reversible stages preceding the irre- versible thermal denaturation of F-actin, thus making the dependence of T m on the F-actin concentration less pronounced and more similar to those characteristic of monomeric proteins. We are aware that the proposed mechanism for the thermal denaturation of F-actin [46] (Fig. 4) is based mainly on indirect data (i.e. the dependence of T m on the actin and ADP concentrations). However, Gicquaud and Heppel directly observed two reversible steps preceding the irreversible thermal unfolding of F-actin [81]. These steps were revealed by measuring the temperature dependences of the fluorescence of F-actin labeled with pyrene at Cys374 and e-ATP bound to F-actin at the nucleotide-binding site [81]. Although the authors did not interpret these results, it is possible that the changes in fluorescence observed before the irreversible thermal denaturation of F-actin may correspond to dissociation of nucleotide from the actin subunits and the dissociation of subunits from the filament. Let us briefly consider the main statements of the proposed mechanism (Fig. 4). One is that the actin filament denatures, not as a whole, but as separate short oligomers which dissociate from the filament during heating. In favor of this assumption are the results of recent studies on the interaction of F-actin with sHSPs [82,83]. It has been found that these pro- teins effectively prevent the aggregation of thermally denatured actin by forming small soluble complexes (Fig. 4), the size of these complexes being much less than that of intact F-actin [82,83] (see below). The other important statement of the proposed dissociative mechanism of the F-actin thermal dena- turation is that the ADP-containing short oligomers dissociated from the ‘)’-end of the actin filament may bind again to the ‘+’-ends of the filaments (Fig. 4). In favor of this assumption is a well-established pheno- menon of end-to-end annealing of actin filaments, i.e. the formation of long polar actin filaments from very short filaments due to an interaction between the ‘+’- and ‘)’-ends of the filaments [84–86]. The annealing process proceeds spontaneously, and, unlike actin polymerization, the presence of ATP-containing G-actin monomers and ATP hydrolysis is not required [84]. Although the annealing is a rather slow process, its rate increases linearly with actin concentration [87], D D D D D D D D D DD D D D D D D D D D D D D D D D DD D D D D D D D D D D D D D D D D D DD D D D D D D D D D D D D D D D D D D D DD D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D DD D D D (–))+( (+) (–) (+) (–) D D D F-actin Denaturation Aggregation sHSP Fig. 4. The proposed mechanism for the thermal denaturation and aggregation of F-actin filaments, as well as the protective effect of sHSPs on the aggregation of thermally denatured actin. ADP mole- cules, both free in solution and bound to actin monomers, are marked as D. The hatched short oligomers of actin are those that dissociate upon heating from the pointed (‘)’) end of actin filament and can bind to the barbed (‘+’) end of the same or other filament. See the text for more details. Actin unfolding and aggregation D. I. Levitsky et al. 4288 FEBS Journal 275 (2008) 4280–4295 ª 2008 The Authors Journal compilation ª 2008 FEBS in good agreement with the above proposed dissocia- tive mechanism for F-actin thermal denaturation. Fur- thermore, it has been shown that the annealing rate near a hydrophobic surface in the presence of crowd- ing agents is substantially higher (by a factor 20) than in solution, and this rate is also much higher than the rate of actin polymerization [87]. These results support our viewpoint that short oligomers which have dissoci- ated from the actin filament in the initial stages of F-actin thermal denaturation may again bind to the filaments (Fig. 4). Aggregation of thermally denatured actin The irreversible thermal denaturation of actin is nor- mally accompanied by aggregation of the protein. However, the aggregation process is quite different for G-actin and F-actin. ‘Inactivated’ G-actin It has been found that thermal denaturation of mono- meric G-actin (after 30 min incubation at 60 or 70 °C) leads to the formation of so-called ‘inactivated actin’, which is represented by stable homogeneous aggregates consisting of a limited number of unfolded protein molecules [88–91]. Inactivated actin can be obtained by both heat treatment and at moderate concentrations of urea (4 m) or guanidinium chloride (1.5 m) [90,91]. According to recent studies, inactivated actin is the off-pathway misfolded state stabilized by aggregation of partially folded actin molecules [92]. It has been shown that inactivated actin represents a specific stable aggregate (named ‘monodisperse associate’ by the authors), which is characterized by a sedimentation constant of  20S (compared with  3.3S for intact G-actin) [88,90]. The hydrodynamic radius of ther- mally denatured G-actin, as measured by dynamic light scattering, was  12 nm (V. V. Mikhailova, unpublished results). A Stokes radius of  8 nm (com- pared with 2.8 nm for intact G-actin) and an apparent molecular mass of  700 kDa have been estimated for inactivated actin using size-exclusion chromatography [90,91]. These data suggest that the inactivated actin associate contains up to 15–17 monomers of denatured actin [91]. It is important to note that inactivated actin can exist in the form of stable monodisperse associates only under very specific conditions, at very low ionic strength. Addition of salts induces the aggregation of inactivated actin, which is accompanied by a signifi- cant increase in light scattering [93]. Heat-induced aggregation of F-actin and effects of sHSPs F-actin aggregates easily when denatured, and its ther- mally induced aggregation is accompanied by a signifi- cant increase in light scattering. Very good correlation was found between thermal denaturation and the aggregation of F-actin [46,82] (Fig. 5). Upon thermal stabilization of F-actin (e.g. in the presence of ADP [46] or phalloidin), both the thermal transition measured by DSC and the aggregation curve measured by light scattering shifted to a higher temperature. F-actin denatures and aggregates at a rather high temperature, > 50 °C (Fig. 5), much higher than the 40 50 60 70 0 100 200 300 400 0 5 10 15 20 25 F-actin F-actin + Hsp27-3D Aggregation (Light scattering) Light scattering (rel. u.) Tem p erature (°C) F-actin F-actin + Hsp27-3D Apparent heat capacity ( µ W) Thermal unfolding (DSC) A B Fig. 5. Thermal unfolding measured by DSC (A) and heat-induced aggregation measured as an increase in light scattering at 350 nm (B) of F-actin (1.0 mgÆmL )1 ) in the absence and presence of small heat shock protein Hsp27 with mutations mimicking its phosphory- lation, denoted as Hsp27-3D (1.0 mgÆmL )1 ). The DSC and light-scat- tering measurements were performed under the same conditions (30 m M Hepes, pH 7.3, 100 mM KCl and 1 mM MgCl 2 ) and at the same heating rate of 1 KÆmin )1 . Adapted from Pivovarova et al. [82,83]. D. I. Levitsky et al. Actin unfolding and aggregation FEBS Journal 275 (2008) 4280–4295 ª 2008 The Authors Journal compilation ª 2008 FEBS 4289 [...]... types of stress, for example heat shock, can induce actin unfolding, leading to the disruption of actin filaments and the aggregation of fully or partially denatured actin The accumulation of aggregated proteins is dangerous for the cell, and this is especially important in the case of abundant proteins, such as actin There are different mechanisms for preventing formation of insoluble aggregates, and. .. occurring in the actin filament (e.g in the case of highly cooperative destabilization of F -actin by cofilin) We have also summarized and discussed the existing data on the aggregation of actin induced by its thermal denaturation It has been shown for the first time using DSC that sHSPs have no influence on the thermal unfolding of F -actin, although they effectively prevent its subsequent aggregation by... on the thermal unfolding of actin, paying special attention to the effects of stabilization and destabilization of F -actin by different factors, expressed as the changes in the thermal stability of F -actin In general, all the DSC data summarized here are in good agreement with the data obtained by other methods Furthermore, in some cases, the DSC results can provide insight into the mechanism of structural... (2004) The effect of phalloidin and jasplakinolide on the flexibility and thermal stability of actin filaments FEBS Lett 565, 163–166 Levitsky DI (2004) Structural and functional studies of muscle proteins by using differential scanning calorimetry In The Nature of Biological Systems as Revealed by Thermal Methods (Lorinczy D, ed.), pp 127–158 ¨ Kluwer, Dordrecht Actin unfolding and aggregation 55 Oda... stabilization or destabilization of actin filaments by different factors can be explained by transitions of F -actin from one structural state to another, and these transitions can be clearly demonstrated in DSC studies as the changes in the F -actin thermal unfolding Thus, DSC in combination with other methods provides a promising approach for studying highly cooperative structural changes in actin filaments. . .Actin unfolding and aggregation D I Levitsky et al temperature in the cell The question arises as to whether the thermal denaturation and aggregation of actin filaments can occur at lower temperatures To answer this, we performed special experiments and found that F -actin denatures and aggregates, although rather slowly, upon prolonged incubation at heatshock temperature (43 °C) in the absence of. .. specific features of F -actin thermal unfolding and, furthermore, how sHSPs protect the cytoskeleton against damage caused by accumulation of large, insoluble aggregates under heat shock conditions Actin unfolding and aggregation Acknowledgements This study was supported by the Russian Foundation for Basic Research (grant 06-04-48343 to DI Levitsky) and the program ‘Molecular and Cell Biology’ of the Russian... complexes with denatured actin These results provide new insight into the mechanism by which sHSPs prevent the aggregation of F -actin induced by its thermal denaturation Finally, combining the DSC results with data obtained from other methods, we have proposed a model that can describe the mechanism of the thermal unfolding of F -actin and its subsequent aggregation, including the effects of sHSPs (Fig 4) This... have no influence on the thermal unfolding of F -actin as measured by DSC, but they effectively prevent the aggregation of thermally denatured actin [82,83] Furthermore, we used co-sedimentation experiments to analyze the interaction between denatured actin and the S15D ⁄ S78D ⁄ S82D mutant construct of Hsp27 (denoted as Hsp27-3D), which has been proposed to mimic the properties of phosphorylated Hsp27... investigation of G -actin denaturation Biofizika 29, 583–585 27 Bertazzon A, Tian GH, Lamblin A & Tsong TY (1990) Enthalpic and entropic contributions to actin stability: calorimetry, circular dichroism, and fluorescence study and effects of calcium Biochemistry 29, 291–298 28 Le Bihan T & Gicquaud C (1993) Kinetic study of the thermal denaturation of G -actin using differential scanning calorimetry and intrinsic . whereas opening of the cleft leads to significant destabilization of G -actin. Thermal unfolding of F -actin filaments Polymerization of G -actin to F -actin The use of DSC allows very clear probing of the changes. ARTICLE Thermal unfolding and aggregation of actin Stabilization and destabilization of actin filaments Dmitrii I. Levitsky 1,2 , Anastasiya V. Pivovarova 1,3 , Valeria V. Mikhailova 1 and Olga. the thermal unfolding of actin, paying special attention to the effects of stabilization and destabilization of F -actin by different factors, expressed as the changes in the thermal stability of F -actin.