Molecular and Structural Basis for Redox Regulation of Β-Actin

doi:10.1016/j.jmb.2007.04.056 J. Mol. Biol. (2007) 370, 331–348

Molecular and Structural Basis for Redox Regulation of β-Actin

Ingrid Lassing1†, Florian Schmitzberger2†, Mikael Björnstedt3 Arne Holmgren2, Pär Nordlund2, Clarence E. Schutt4 and Uno Lindberg1⁎

1Department of Microbiology, An essential consequence of growth factor-mediated signal transduction is the Tumor Biology, and Cell generation of intracellular H2O2. It operates as a second messenger in the Biology, Karolinska Institutet, control of actin microfilament dynamics, causing rapid and dramatic changes SE-171 77 Stockholm, Sweden in the morphology and motile activity of stimulated cells. Little is understood about the molecular mechanisms causing these changes in the actin system. 2Department of Medical Here, it is shown that H O acts directly upon several levels of this system, and Biochemistry and Biophysics, 2 2 some of the mechanistic effects are detailed. We describe the impact of Karolinska Institutet, SE-171 77 oxidation on the polymerizability of non-muscle β/γ-actin and compare with Stockholm, Sweden that of muscle α-actin. Oxidation of β/γ-actin can cause a complete loss of 3Division of Pathology F46, polymerizability, crucially, reversible by the thioredoxin system. Further, Department of Laboratory oxidation of the actin impedes its interaction with profilin and causes Medicine, Karolinska University depolymerization of filamentous actin. The effects of oxidation are critically Hospital, Huddinge, SE-141 86 dependent on the nucleotide state and the concentration of Ca2+.Wehave Stockholm, Sweden determined the crystal structure of oxidized β-actin to a resolution of 2.6 Å. 4 The arrangement in the crystal implies an antiparallel homodimer connected Department of Chemistry, by an intermolecular disulfide bond involving cysteine 374. Our data indicate Princeton University, that this dimer forms under non-polymerizing and oxidizing conditions. We Princeton, NJ 08544-1009, USA identify oxidation of cysteine 272 in the crystallized actin dimer, likely to a cysteine sulfinic acid. In β/γ-actin, this is the cysteine residue most reactive towards H2O2 in solution, and we suggest plausible structural determinants for its reactivity. No other oxidative modification was obvious in the structure, highlighting the specificity of the oxidation by H2O2. Possible consequences of the observed effects in a cellular context and their potential relevance are discussed. © 2007 Elsevier Ltd. All rights reserved. β Keywords: ADP- -actin; antiparallel dimer; cysteine sulfinic acid; H2O2; *Corresponding author thioredoxin

sincereactiveoxygenspecies(ROS)quenching Introduction 1 abolishes it. Intracellular H2O2 is transiently gen- In many eukaryotic cells redox effects appear to erated upon activation of receptors for peptide growth factors and cytokines as well as integrin- have central roles in signal transduction, and H2O2 is essential for growth factor-induced signaling, mediated cell adhesion. It acts as a rapidly produced and effective second messenger, whose spatiotem- poral presence in cells often correlates with changes in the microfilament system.2,3 † I.L. and F.S. contributed equally to this work. Growth factor-stimulated H2O2 production is also Abbreviations used: DTNB, closely interrelated with increases in the intracellular 5,5′-dithiobis(2-nitrobenzoic acid); DTT, 1,4-dithiothreitol; concentration of Ca2+,4 and appears to correspond to Gdn-HCl, guanidine hydrocloride; ROS, reactive oxygen the kinetics of actin polymerization-dependent out- species; Trx, thioredoxin; TR, thioredoxin reductase. growth of lamellipodia and filopodia.5,6 E-mail address of the corresponding author: Actins are highly conserved eukaryotic proteins [email protected] existing in different cellular isoforms. Non-muscle

β-actin and γ-actin, amongst the most abundant of assembly/disassembly and related changes of proteins in many cell types, are pivotal in maintain- actin during redox conditions as well as its regula- ing and mediating the infrastructure of the cellular tion by associated proteins. matrix. They form an essential part of chemo- Most of the research on actin has been dedicated mechanical transduction systems in non-muscle to characterizing its attributes under reducing cells, where they, together with actin-binding pro- conditions. Equally important is an understanding teins, generate force for various activities involved of its properties in an oxidizing environment, in translocation and cell shape change. representative of the circumstances during signal The constant and rapid reorganization of the actin transduction by H2O2. Experiments to study an microfilament system during cell motility and effect of oxidation on actin in vitro have been done migration, mitogenesis, and phagocytosis depends with muscle α-actin,13 whereas the non-muscle β/γ- on nucleation, elongation, and depolymerization of actins would seem to be the more relevant isoform actin filaments as elementary processes.6,7 The with respect to cell shape and motility. Indeed, one dynamic reorganization of cellular actin is highly of the few significant differences between α-actin regulated, and ROS appear to be one vital regula- and β/γ-actin is the number of cysteine residues, the tory element.2 Experiments have indicated that actin thiol groups of which are the protein moieties could constitute a direct target for oxidative mod- generally most reactive towards H O . Non-muscle – 2 2 ification in vivo.8 10 Moreover, actin was shown to be β and γ-actins have six cysteine residues (Figure 1), oxidatively modified in pathophysiological states and α-actin has five. suggestive of oxidation as a cause of mechanical Here, we have examined the biochemical effects of 11,12 dysfunction. Thus, an explanation of the mole- oxidation by H2O2 on the capacity of the actin iso- cular basis of redox-regulated microfilament pro- formstoformfilamentsandcharacterizedthe cesses requires an understanding of the mechanism structural consequences of oxidation in β-actin.

α Figure 1. Illustration of the position of cysteine residues and the profilin and DNase I interaction sites in actin. The C atom cartoon represents the β-actin structure.17 The color scheme for this and the following model illustrations, unless stated otherwise, is as follows. Ball-and-stick representation for the atoms: carbon, grey; nitrogen, blue; oxygen, red; phosphorus, orange; sulfur, yellow. Amino acid residues involved in the interaction with DNase I (derived from the α- actin–DNase I complex104) are colored magenta and those in contact with profilin are colored light-blue. Redox Regulation of β-Actin 333

The results show that oxidation by H2O2 has Table 2. Effect of calcium on oxidation of cysteine profound and differential effects on both mono- residues in ADP-β/γ-actin meric and filamentous non-muscle actin, and can lead to the formation of disulfide-linked antiparallel Oxidized cysteine 2+ residues homodimers. [Ca ]free Oxidation conditions (μM) Number Identity

5mMH2O2, 10 min, 25 °C 1 2 Cys272/Cys374 Results 5mMH2O2, 15 min, 10 0.6 Cys272 at 25 °C 5mMH2O2, 15 min, 37 °C 10 2.3 Cys272/Cys374 Actin thiol groups and their reactivity towards 5mMH2O2, 15 min, 37 °C 100 1 Cys272 ′ 20 mM H2O2, 40 min, 100 2 Cys272/Cys374 5,5 -dithiobis(2-nitrobenzoic acid) (DTNB) 37 °C and H2O2 Number of oxidized SH-groups after incubation with H2O2 in the presence of different concentrations of Ca2+. The number of accessible SH-groups in α-actin and β/γ-actin in the presence of ATP or ADP (see Materials and Methods), is given in Tables 1 and 2. the lag phase or the steady-state level of polymer- At high concentrations (0.1 mM) of Ca2+, mono- α β γ ization significantly (Figure 2(a)). In contrast, oxida- meric -actin and / -actin in the presence of both tion of ADP-β/γ-actin at 1 μMCa2+ (two cysteine ATP and ADP had one and two accessible cysteine residues oxidized) led to a complete loss of poly- residues, respectively. At submicromolar concentra- β γ 2+ merizability (Figure 2(b)). Incubation of ADP- / - tions of Ca , an additional DTNB-reactive SH- actin with H O with 0.1 mM Ca2+ (one cysteine group appeared in α-actin-ADP, as shown earlier.14 2 2 β γ oxidized; Table 2) resulted in an actin which Similarly, the ADP-form of / -actins exposed an polymerized more slowly but reached a steady- extra DTNB-reactive cysteine residue at 1 μMCa2+. β γ state level comparable to that of the non-oxidized In the ATP-form of / -actin, the additional SH- control (data not shown). Similarly, oxidation of α- group was only partially accessible and, in the 2+ 2+ actin in the ADP form at 0.1 mM Ca , for a presence of 0.1 mM Ca , its appearance required a prolonged period, led to decreased polymerizability higher concentration of H2O2 or prolonged incuba- (data not shown). tion time. Thus, one of the three cysteine residues could not be oxidized by H2O2, even though it was Effect of the thioredoxin system on oxidized accessible to DTNB. Notably, for the extra cysteine ADP-β/γ-actin to become unavailable, it was sufficient to increase the concentration of Ca2+ to 10 μM, under otherwise The reversibility of oxidation by H2O2 and its similar conditions (Table 2). effect on the polymerizability of ADP-β/γ-actin at 1 μMCa2+ are illustrated in Figure 2. Oxidized Effect of oxidation on actin polymerizability protein was incubated with the thioredoxin system, β γ which reduces disulfide bonds and cysteine sulfenic Oxidation of ATP- / -actin with 5 mM H2O2 at acids, products of oxidation of cysteine residues 1 μMofCa2+ (one cysteine oxidized) did not affect with H2O2. Indeed, thioredoxin rapidly reduced the oxidized cysteine residues in ADP-β/γ-actin. All Table 1. Number of DTNB reactive SH-groups/actin components of the thioredoxin (Trx) system molecule before and after oxidation with H2O2 (NADPH, thioredoxin reductase (TR), and Trx; see SH groups in Materials and Methods) were required for the Total SH non-denatured actin reduction of the cysteine residues and subsequent 2+ restoration of the capacity of the protein to form A. In 0.1 mM Ca β γ ATP-α-actin 5.0 (n=1) 1.0 (n=1) polymers (Figure 2(d)). Incubation of ADP- / - ADP-α-actin 5.0 (n=7) 1.0 (n=7) actin under more oxidative conditions (higher Oxidized concentrations of H2O2, longer incubation time) ATP- α-actin 4.0 (n=1) 0.5 (n=1) α led to a product that could not be reduced by the ADP- -actin 4.1 (n=4) 0.5 (n=3) Trx-system, likely due to formation of cysteine ATP-β/γ-actin 5.9 (n=3) 2.0 (n=3) sulfinic or sulfonic acids. ADP-β/γ-actin 6.0 (n=8) 2.0 (n=4) Oxidized Molecular status of oxidized ADP-β/γ-actin ATP-β/γ-actin 4.3 (n=2) 0.9 (n=2) ADP-β/γ-actin 4.0 (n=7) 1.0 (n=3) The molecular mass distribution of the non- B. In 1 μMCa2+ (non ox.) polymerizable ADP-β/γ-actin, formed at 0.1 mM ADP-β/γ-actin 6.0 (n=1) 3.0 (n=1) Ca2+ incubated with 20 mM H O for 40 min, is β γ 2 2 ATP- / -actin 6.0 (n=1) 2.3 (n=1) illustrated in Figure 3(a). The protein was essentially The total number of SH groups reactive towards DTNB was recovered as a major peak at the position expected assayed in the presence of GdnHCl, and SH groups available in for monomeric actin, with smaller amounts of non-denatured monomers were analyzed in G-buffer. Oxidized protein appearing next to it in the higher molecular samples were exposed to 20 mM H O for 40 min at 37 °C. 2 2 mass region. The predominant monomeric nature of 334 Redox Regulation of β-Actin

Figure 2. Effect of oxidation on actin polymerizability. (a) Polymerization of ATP-β/γ-actin (0.4 mg/ml) before (○) ● β γ and after oxidation ( ) with 5 mM H2O2 (see Materials and Methods). (b) Polymerization of ADP- / -actin before and after oxidation with H2O2. Actin was incubated for 10 min with 5 mM H2O2 (oxidant not removed); symbols have the same meaning as in (a). (c) Reduction of oxidized cysteine residues with the Trx system. Upon removal of H2O2, oxidized ADP-β/γ-actin in G buffer with ADP and 50 μMCa2+, was incubated with the Trx system for increasing lengths of time and the number of SH groups was determined with DTNB in guanidine-HCl (see Materials and Methods). (d) Restoration β γ of polymerizability of ADP- / -actin inactivated by H2O2. Oxidized actin was incubated with the Trx system for 60 min at 37 °C, after which polymerization was induced by adding polymerizing salts at time zero and the viscosity was measured at 25 °C. the oxidized ADP-β/γ-actin was confirmed by SDS- ATP-induced polymerization of oxidized PAGE in the absence of reducing agents (see Figure ADP-β/γ-actin 3(a) and (b)). The major peak had ∼4 cysteine residues available for reaction with DTNB (i.e. two Intriguingly, addition of 0.5 mM ATP to oxidized cysteine residues oxidized). In contrast, the actin in ADP-β/γ-actin led to recovery of polymerizability, the higher molecular mass fraction contained ∼2.0 although polymer formation occurred after a con- cysteine residues available for DTNB in guanidine siderable lag phase (Figure 4). Exposure of ADP-β/ hydrochloride (Gdn-HCl), suggesting that a total of γ-actin to more oxidative conditions (higher con- ∼ 4 cysteine residues had been oxidized. centrations of H2O2, longer times) resulted in a Similarly, gel-filtration of the oxidized, non- product that could not be recovered in a polymeri- polymerizable ADP-β/γ-actin in the presence of zable form by the Trx system, again presumably polymerizing salts resulted in one major peak with indicating that cysteine sulfinic or sulfonic acids had monomeric actin, and a second peak of higher formed. However, even after such treatment, ATP molecular mass. Approximately 90% of the applied restored the polymerizability of the protein. protein was recovered, and the major peak contained more than ∼60% of the applied protein, Interaction of oxidized ADP β/γ-actin with demonstrating that the monomeric form of oxi- profilin and DNase I dized actin remained in the non-polymerizable form. The actin sample from the major peak had In order to obtain information about potential two oxidized cysteine residues, in agreement with changes in the structure of actin upon oxidation, its the results obtained under non-polymerizing con- interactions with profilin and DNase I were studied. ditions. Actin from this fraction could be concen- The results shown in Figure 5 illustrate the dif- trated to more than 15 mg/ml in G-buffer (see ference in affinity of non-oxidized and oxidized Materials and Methods) without the solution ADP-β/γ-actin for profilin. Here, the ADP-β/γ- becoming viscous, confirming that the self-associat- actin samples were incubated with a twofold molar ing capacity of the actin monomers had been excess of profilin and then applied to poly-l-proline eliminated by oxidation. Sepharose resin, which binds profilin as well as Redox Regulation of β-Actin 335

actin molecule. However, oxidation of the most reactive cysteine did not influence the binding of DNase I.

Effect of H2O2 on filamentous actin

The effect of H2O2 on actin, polymerized to steady-state, was followed by monitoring the viscosity. The oxidant caused an immediate drop in viscosity of the solution (Figure 6(a)). Centrifuga- tion of H2O2-treated samples at 30 psi (1 psi ≈6.9 kPa) for 30 min showed that almost all of the actin was in the supernatant (Figure 6(a)), suggesting that exposure to H2O2 had resulted in fragmen- tation or depolymerization of the filaments. When pyrenyl-containing filaments were exposed to 5 mM H2O2, no significant change in the fluorescence could be detected, demonstrating that the filaments were not disassembled into monomers, but rather had become fragmented (see Figure 6(b)). Increasing the concentration of H2O2 to 20 mM, however, decreased the fluorescence to background levels within 2 h (Figure 6(c)), indicative of complete depolymerization. Oxidation of filaments in the presence of a low concentration of Ca2+ (1 μM) did not seem to cause significant depolymerization, regardless of whether they had been formed in a low concentration of Ca2+, or in 0.1 mM Ca2+, with subsequent addition of EGTA to lower the concentration of Ca2+ (Figure 2+ Figure 3. Size distribution of oxidized ADP-β/γ-actin. 6(c)). However, at concentrations of Ca above (a) Analysis of oxidized ADP- β/γ-actin by chromato- 10 μM, significant depolymerization did occur graphy on S-300 in 5 mM KH2PO4/K2HPO4 (pH 7.6), (data not shown). μ 0.5 mM ADP, 0.1 mM CaCl2,10 M EDTA. Inset: protein from fractions 39 (lane a) and 49 (lane b) after non- Effect of diamide on SH groups and reducing SDS-PAGE (10% polyacrylamide gel) and polymerizability of β/γ-actin stained with Coomassie brilliant blue (trace amounts of actin dimers, trimers etc. are seen above the oxidized actin monomers; no protein band was seen below the major Incubation of eukaryotic cells with diamide, a β relatively thiol-specific oxidizing agent, causes actin band). (b) Chromatography of oxidized -actin 16 under polymerizing conditions. The shaded area indicates reorganization of the microfilament system. fractions used for crystallization. Inset: lane a fr. 39, lane b Here, β/γ-actin incubated with equimolar amounts from shaded area. of diamide, which eliminated the DTNB-reactivity of one SH-group, still polymerized (Figure 7). However, incubation of the actin with a higher profilin:actin complexes, but not actin alone. The concentration of diamide, oxidizing two to four major fraction of the non-oxidized actin bound to cysteine residues, significantly decreased the poly- the poly-l-proline Sepharose, indicating stable com- mer-forming capability of the actin, in agreement plex formation, whereas oxidized actin was recov- with the results obtained with H2O2. In these cases, ered mostly in the flow-through. These results addition of ATP to the inactivated actin rapidly demonstrate clearly that oxidation of ADP-β/γ- restored its polymerizability, except in the case actin weakens its interaction with profilin. Notably, where five out of six SH-groups had become ADP-β/γ-actin with only one cysteine oxidized was oxidized. retained by poly-l-proline Sepharose; i.e. its interaction with profilin was not affected significantly (data Crystal structure of β-actin after incubation with not shown). H2O2 The effect of oxidation of ADP-β/γ-actin on the binding to DNase I was analyzed with the DNase I An ADP-β-actin sample, which had been incu- 15 inhibition assay. Oxidation of two cysteine resi- bated with 20 mM H2O2 and had retained its dues in ADP-β/γ-actin caused a two- to threefold polymerization potential, was crystallized in the increase in the Kd for the interaction with DNase I, presence of ADP. The structure solution (Table 3)is suggesting that oxidation at the C terminus of the consistent with four molecules, here referred to as A, actin affected the interdomain relationship and B, C, and D, respectively, in the asymmetric unit. conformation of the DNAse I interaction site of the Interpretable electron density is present for the full- 336 Redox Regulation of β-Actin

D are related by a nearly 2-fold, non-crystallographic axis located between their C termini (angle of rotation 178°) and the interface between them buries a calculated solvent-accessible surface area of ∼780 Å2 per protomer (Figure 8(a)). Protomers A and C are arranged in an essentially identical way with a crystallographic 2-fold axis relating them to their counterpart molecules. The interaction interface is composed of residues from subdomains 1 and 3,19 and displays a mixed electrostatic character with ∼60% non-polar atoms. In protomers B and D, the interface comprises residues 166–173 located in a loop in subdomain 3, which together with residues 286 and 289 (subdomain 3) line up against residues 350–355 located Figure 4. ATP-induced polymerization of oxidized in an α-helix in subdomain 1 from the adjacent ADP-β/γ-actin. The first part of the graph (0–60 min) molecule. Apparent hydrophilic interactions include represents the polymerization of non-oxidized actin in a hydrogen bond formed across the gap of the two ADP (□), oxidized actin in ADP (△ and ▴), and oxidized molecules between the hydroxyl group of Tyr143 ADP-actin, to which ATP was added before polymerizing ● (subdomain 1) and the carboxyl group of Glu167 salts ( ). The second part shows that addition of 0.5 mM (subdomain 3). The interface further consists of ATP restores the polymerizability of oxidized ADP-β/γ- – ▴ residues 370 374 (subdomain 1), which are situated actin, even after incubation for 60 min ( ). Addition of α 0.5 mM DTT to oxidized ADP-actin after 60 min of incu- alongside the equivalent, mainly -helical, range of bation is shown (△). Polymerization of actin (0.4 mg/ml) the adjacent molecule. was monitored by the pyrenyl-assay (see Materials and Methods). Structural effects of oxidation by H2O2

Inspection of difference Fourier electron density maps indicated a covalent modification of cysteine length protein, except for residues 1–5(1–3in 272 (Figure 9(a)) in all four molecules, essentially molecules B and C), 40–49 (DNase I binding loop), consistent with an oxidation to cysteine sulfenic or 372–375 in molecules A and C, as well as residue 375 sulfinic acid. Modeling of either form agrees with in molecule B. At the nucleotide-binding site, the data and the corresponding real space density fit contiguous electron density for the presumed posi- and cross-validated R-factor display no significant tion of three phosphate groups was visible after differences. Also, a cysteine sulfonic acid residue fits σA – simulated annealing refinement in the initial - the electron density of the 2Fo Fc type and no – – weighted omit map of the 2Fo Fc type at a level of negative peaks in electron density maps of the Fo Fc greater than the 3.0σ providing a convincing type at below the −2.0σ contour level at the site of argument for the presence of an ATP, rather than the modeled oxygen atoms became apparent. How- an ADP, molecule. The presence of ATP in the actin ever, modeling of a sulfinic acid residue was molecule was unexpected. However, although β- actin was purified in the presence of ADP, a subfraction of β-actin with bound trace amounts of ATP could have been copurified and this sample crystallized. Alternatively, actin may have crystallized with ADP, in a closed state, and ADP then exchanged to ATP because of the greater affinity of actin for ATP. However, interpretable electron density for the histidine 73 methyl group, present in the profilin:β-actin structure was not observed.17 Superposition on previously solved actin structures demonstrates that, overall, the structure closely resembles those, except for the relative position of subdomain 2 in the open profilin:β- actin structure.18 The actin is consequently in a conformation typically referred to as the closed Figure 5. Interaction between profilin and actin, state, and the four molecules in the asymmetric unit before and after oxidation of actin. Non-oxidized, and are conformationally congruent. However, this is oxidized, actin was mixed with a twofold excess of profilin and then fractionated on a poly-l-proline Sepharose the first time that this particular arrangement of the column. Fractions from the flow-through (lanes 2 and 4) actin molecules has been observed. In the asym- show actin not interacting with profilin, and lanes 1 and 3 metric unit, three protomers are oriented almost show the protein recovered from the matrix (profilin and identically, whereas another molecule is located in profilin:actin) stained with Coomassie brilliant blue after an antiparallel orientation to them. Protomers B and SDS-PAGE (15% polyacrylamide gel). Redox Regulation of β-Actin 337 eventually preferred, because of indications for such a form in positive difference Fourier peaks (Figure 9(a)). Nonetheless, it is not unlikely that the observed electron density represents a superposition of several oxidation states in the crystal, consistent with mass spectrometry analysis of the protein sample (data not shown). No significant conformational change in the protein is observable in the vicinity of cysteine 272, apart from the side- chain of glutamate 276. It is displaced from its corresponding position in the structure of profilin:β- actin,17 which seems attributable to electrostatic repulsion with the oxygen atoms of the oxidized Cys272 in our β-actin structure (Figure 9(a)). Upon reprocessing of the diffraction data using Figure 7. Effect of diamide on the polymerization of images, where radiation damage was less severe, β γ ADP- / -actin. Diamide was added to the protein contiguous electron density between the presumed (0.5 mg/ml, 2% (w/v) pyrenyl-actin) in G-ADP buffer at μ 10 MCaCl2 to give final concentrations of diamide of 10 μM(●), 30 μM(▴), 0.1 mM (Δ), 1 mM (■), 10 mM (□), and the control without diamide (○) (total volume 200 μl). Polymerization was induced by addition of 2 mM MgCl2 and 100 mM KCl. After 80 min, 2 μl of 0.5 M ATP was added to all samples and the reaction was followed for another 30 min. In a parallel experiment, the number of oxidized cysteine residues was determined by the DTNB method.

position of the sulfur of Cys374 in molecules B and D could be interpreted as a disulfide bond (Figure 9(b)). It displays an architecture consistent with that of a right-handed conformation with most of the parameters refining close to the expected values.20 However, the dihedral angle around the S–S bond refined to 132°, deviating from the energetically favorable 100°,21 and the torsion angle around the Cα–Cβ bond in protomer B to 155°, deviating from the mean value of −60°. This may indicate an energetically less favorable conformation of the

Figure 6. Effect of H2O2 on filamentous actin. (a) Poly- merization of ADP-β/γ-actin (0.4 mg/ml) in G-ADP buffer (see Materials and Methods) at 0.1 mM Ca2+, induced by the addition of polymerizing salts at time zero. After 32 min, 5 mM H2O2 (final conc.) was added and polymer formation was monitored by viscometry at 25 °C. Inset: Status of the actin after the drop in viscosity: 0.25 ml samples of the oxidized actin were centrifuged for 30 min in an airfuge at 30 psi at room temperature. Pelleted material and precipitated protein from the supernatant was resuspended in 0.2 ml of G-ADP buffer boiled with sample buffer, and analyzed by SDS-PAGE (10% polyacrylamide gel). The supernatant and pellet obtained with non-oxidized F-actin were used as a control. (b) Effect of adding 0.1 mM EGTA before polymerizing salt (final 2+ μ Ca free is 1 M), and 5 mM H2O2 (final conc.) at 20 min, followed by the addition of 0.1 mM CaCl2 at 40 min. (c) The β γ effect of H2O2 on ADP- / -actin polymers monitored by the pyrenyl assay (see Materials and Methods). For this, ADP-β/γ-actin (0.1 mM Ca2+) was polymerized. In one sample, the Ca2+ concentration was lowered to 1 μMbythe addition of 0.1 mM EGTA 45 min after initiation of poly- ○ merization ( ). After 55 min, 5 mM H2O2 (final conc.) was added to both samples, and after 100 min 20 mM H2O2 (final conc.) was added. 338 Redox Regulation of β-Actin

Table 3. Crystallographic data statistics sibly an intermolecular disulfide bond with their counterpart molecule, related by the crystallo- A. Data collection and integration graphic 2-fold axis could be present. Wavelength (Å) 1.0000 Noticeably, the hydroxyl groups of four tyrosine Space group C2221 Unit cell dimensions residues, Tyr133 and Tyr169 from both molecules, are a (Å) 119.17 positioned around the disulfide bond (Figure 9(b)). b (Å) 222.59 The oxygen atoms of the hydroxyl groups of Tyr169 c (Å) 133.71 ∼ ∼ Resolution range (Å) 35-2.60 (2.68-2.60) are at a distance of 4 Å and 4.5 Å, respectively, No. unique reflections (multiplicity) 52 031 (3.6) from the closest sulfur atom (with the valency state of a Rmeas 0.15 (0.79) sulfur in the disulfide bond corresponding to a van Mean bIN/σbIN 9.34 (1.72) der Waals radius of ∼1.8 Å), whereas those of Tyr133 Completeness (%) 94.7 (76.4) ∼ ∼ 2 are at a distance of 5 Å and 6.5 Å, respectively. In Wilson B-factor (Å )39 the crystal structure of oxidized β-actin these are the B. Refinement only hydrophilic protein moieties, apart from back- b Rcryst 0.210 (0.279) bone carbonyl groups and amides, in the immediate c Rfree 0.288 (0.371) vicinity of the disulfide bond. No. reflections: working/test set 48 371/2548 Average refined B-factor (Å2)d 42 No other oxidative modification was obvious, No. non-H atoms suggesting that the applied concentration of H2O2 Protein 11 028 was not excessive. It is notable that Cys217 and Water 494 Cys257, which are in close proximity in the Ligand 161 structure, did not seem to be affected. Likely, this C. Model analysis is a reflection of the specificity of the oxidation. Estimated coordinate error (Å)e 0.37 r.m.s. deviation Bond lengths (Å)f 0.010 Bond angles (deg.)f 1.389 Discussion Ramachandran plotg No. residues in most favored region 1375 No. residues in generously allowed region 40 Cys272–the most reactive cysteine No. residues in disallowed region 0 Values in parentheses are for the highest resolution shell. Oxidative modifications of cysteines in response rffiffiffiffiffiffiffiffiffiffiffiffiffiffi X Xnh to transmembrane signaling has emerged as a highly nh jhI iI ; j n 1 h h i relevant regulatory and reversible mechanism of a h h i meas ¼ 23,24 R PPnh ; central function in a wide variety of systems. Ih;i Here, we show that exposure of non-muscle ADP- 2+ h i β/γ-actin to H O at high concentrations of Ca Xnh 2 2 〈 〉 1 with Ih = Ih,i and nh multiplicity. (0.1 mM) in G-buffer results in the oxidation of one nh Xi cysteine, and that a second cysteine is modified by jjFobsj jFcalcjj prolonged incubation with H2O2 at higher tempera- b ¼ h X ture. Importantly, it is only after the second cysteine Rcryst ;whereFobs and Fcalc are the observed jFobsj had been oxidized that an interference with the h polymerizability of the protein and its binding to and calculated structure-factors respectively. profilin was observed. Since chemical modification c 95 Rfree: cross-validation of Rcryst. d of Cys374 by various means affects polymerizability Total refined B-factor including the TLS component. 25 e and profilin binding by the actin, these results Estimated coordinate error of the structure based on Rfree as calculated by Refmac.94 imply that another cysteine in non-muscle actin f Calculated with Refmac. reacts more readily than Cys374, and that this is g Calculated with MolProbity.99 Cys272. Consistently, our crystal structure revealed that Cys272 is oxidized and displays a higher oxidation state than Cys374, showing it to be the disulfide bond and/or the onset of radiation most reactive cysteine in β-actin. Significantly, in α- damage.22 The latter is supported by calculations actin and non-vertebrate actin isoforms the amino of electron density maps using the data ranging acid in position 272 is alanine and no cysteine is from 90° to 180° (see Materials and Methods), where present in the structurally equivalent location. no interpretable contiguous electron density for the position of the disulfide bond was observable, likely Reactivity of Cys272: structural considerations symptomatic of the radiation sensitivity of it, its breakage, and subsequent mobility of the C termini. The intrinsic reactivity of thiol groups is modu- The C-terminal five residues of molecules B and D lated by their chemical microenvironment and by are viewed down the rotational axis, oriented the extent to which they are accessible to reactive towards the protein part of the D protomer (Figure groups. In β-actin, the oxidized Cys272 is located at 8(a)). A similar arrangement may exist in molecules the surface of the protein, is solvent-exposed (Figure A and C. Although the electron density for residues 9(a)), and therefore easily accessible to H2O2 in C-terminal of His371 was not interpretable, plau- monomeric or dimeric actin. The only notable pro- Redox Regulation of β-Actin 339

Figure 8. Illustration of the antiparallel β-actin homodimer. (a) Secondary structure representation of the β-actin dimer in an orientation with the approximately 2-fold non-crystallographic axis approximately perpendicular to the plane of the paper. The disulfide bond (green) and the ATP molecule are shown in ball-and stick-representation. Color scheme: red, α-helices; blue, β-strands; grey, loops. (b) Stereo-representation of the structural variability of the β-actin homodimer (black, without ligands) and the superimposed α-actin dimer49 (magenta, without ligands) via the protomer to the right. The disulfide bond is shown in ball-and-stick representation for both dimers. tein moieties in the microenvironment of Cys272 are Oxidation of Cys374 and the exposure of Cys17 the side-chains of Glu276 and Asn280. A cysteine α thiol would be expected to have a pKa near 9 and It is known from studies on -actin that the largely protonated at physiological pH, and thus highly conserved Cys374 is available for reaction have a low reactivity. While the microenvironment with various types of thiol reagents, and that of Cys272 does not appear to lower the pKa value of manipulation of the C terminus affects the ability Cys272, the removal of the thiol proton in the of the protein to form polymers.25,28,29 However, reaction might be facilitated by the carboxyl group the mechanism by which the C terminus is in- of Glu276. An activation mechanism involving an volved in monomer–monomer interactions during aspartate/glutamate and a catalytic cysteine has filament assembly is not known. Yet, there is been proposed for proteins of the thioredoxin-like evidence for a reciprocal allosteric communication family, where the carboxyl moieties are thought to between the C terminus and both the nucleotide- assist in deprotonation of the thiol group of the binding site and the DNase I interaction site within – reactive cysteine.26,27 In β-actin, Glu276 might play the actin molecule.25,30 32 In this respect, it is worth a role in the reaction with H2O2 in mediating proton pointing out that the C terminus with the carboxyl transfer to generate a leaving water molecule. group of Phe375, in the conformation observed in Moreover, in solution the effect of metal ions on molecule D in our crystal structure, forms a hydro- the reactivity of Cys272, though not directly infer- gen bonding interaction with the backbone nitrogen able from our data, cannot be excluded. atom of Ile136. This residue, in turn, is situated next 340 Redox Regulation of β-Actin

Figure 9. Illustration of the effects of oxidation on Cys272 and Cys374 in the β-actin structure. (a) Stereo- – σ − 3 representation of the omit electron density with mFo DFc coefficients contoured at 3.0 (0.13 e /A , blue) before modeling the side-chain oxygen atoms of sulfinic acid and the final refined cysteine sulfinic acid model. A potential hydrogen bonding interaction of one of the sulfinic acid oxygen atoms of oxidized Cys272 with Asn280 is indicated by a broken light – σ − 3 blue line. (b) Stereo-representation of the omit electron density with mFo DFc coefficients contoured at 3.7 (0.15 e /A , blue) before modeling the sulfur atoms and the final refined disulfide bond coordinates. Amino acid residues at the C termini and in the vicinity of the disulfide bond are shown. to Gln137, proposed to form the hinge point for Here, it is shown that non-muscle ADP-β/γ-actin domain motions in the actin molecule,19 and also has a DTNB-reactive cysteine, in addition to potentially involved in ATP hydrolysis. Of note, Cys272 and Cys374, accessible at low levels (1 μM) the carboxyl group of Phe375 in the oxidized of Ca2+ but shielded at increased concentrations of 2+ N μ structure, positioned this deep in the hydrophobic Ca ( 10 M) and not oxidized by H2O2. Amino pocket, has not been observed in other actin crystal acid residue 10 in non-muscle actins is valine. structures, where instead the corresponding space However, Cys17 occupies a similar position in the is occupied by water molecules or co-crystallized interior of subdomain 1. The appearance of an addi- ligands. This interaction may constitute a contact tional DTNB-reactive cysteine at low concentrations point between the C terminus and the nucleotide- of Ca2+ in the absence of ATP suggests strongly that binding site, and in this way relay changes from the Cys17 is exposed via a conformational change C terminus to the DNase I interaction site via the similar to that occuring in α-actin under these nucleotide-binding site, explaining how different conditions. The high degree of flexibility in this chemical modifications of Cys374 alter the proper- region of the molecule is of particular interest, since ties of actin. alterations in the structure are expected to affect the Faulstich and co-workers showed that at low configuration and/or position of the DNase I concentrations of Ca2+, modification of Cys374 in binding loop crucial for the formation of actin ADP-α-actin results in a slow, reversible conforma- filaments. The fact that chemical modification of tional transformation, exposing a second cysteine, Cys374 influences Cys17 provides further evidence identified as Cys10.14 Additionally, the presence for a communication between the C terminus and – of ATP or metal ions in concentrations sufficient the DNase I binding loop.25,30 32 Significantly, our for binding to moderate-affinity sites shielded present results imply that these conformational Cys10. changes are under Ca2+ control. Redox Regulation of β-Actin 341

Oxidation of Cys374: Ca2+ sensitivity and loss of state and susceptible to breakage by radiation- polymerizability induced radicals.22,37 The protein context of the disulfide bond in the The reversibility of the inactivation of the poly- oxidized β-actin structure is largely hydrophobic, mer-forming capacity of ADP-β/γ-actin caused by with the exception of the hydroxyl groups of Tyr133 2+ H2O2 at low concentrations of Ca implies that the and Tyr169. It has been pointed out that hydroxyl structural changes associated with this effect may be groups or aromatic rings through interactions with physiologically relevant. Notably, Ca2+ at concen- the disulfide bond confer stability and restrict trations above 10 μM shields Cys374 from the accessibility to reducing agents.20,38 oxidative effects of H2O2, including exposure of Cys374 is partially solvent-accessible in mono- Cys17. Stimulation of cells with various kinds of meric and filamentous actin.36,39 It is oriented agonists, such as growth factors and cytokines, is towards the protein interior in most monomeric – almost invariably linked to increases in the intracel- actin structures having intact C termini.17,40 42 In lular concentration of Ca2+, from submicromolar this orientation, the aromatic ring of Tyr133 may be values in resting cells to more than 10 μM.33 It is involved in π-bonding to the sulfur of Cys374, within this concentration range that calcium ions which are at a distance of 3.4–4.0 Å in these struc- –π modify the sensitivity of actin to H2O2. tures. Sulfur aromatic interactions have been Interestingly, mere addition of ATP, even to the suggested to modulate the reactivity of cysteine over-oxidized form, restored the capacity of actin to residues.43 Alternatively, Arg116 is a potential polymerize, showing that ATP is effective at restor- residue involved in facilitating the reactivity of ing interdomain connectivity and the normal posi- Cys374, as it is situated in its proximity in many tioning of subdomain 2 involved in filament actin structures. formation. Antiparallel homodimer formation Cys272, Cys374 and the actin filament The interaction between actin molecules B and D Clearly, Cys374 of actin is of importance from a in the crystal is consistent with an antiparallel β- regulatory point of view. Changes in its avail- actin homodimer connected via an intermolecular ability upon polymer formation and effects of disulfide bond (Figures 8 and 9). In solution, α-actin chemical modification of this residue on polymer- antiparallel dimers, inferred to be cross-linked via – izability and stability of the actin polymer25,28,29 Cys374, have been observed.44 48 However, from indicate a high degree of flexibility of the C comparison of the crystal structures (Figure 8(b)) it terminus of the protein, and that the penultimate would appear that the discrepancies in interatomic Cys374 has an important role in establishment of distances between the reported dimers could be due intermonomer contacts. During polymerization, to the inherent structural plasticity rather than Cys374 on one monomer is near residue 41 in fundamentally different species. subdomain 2 and residues 262–274,31,34,35 compris- Oxidized β-actin in our preparations is predomi- ing the “hydrophobic plug”, allegedly involved in nantly in the monomeric form in solution. Even so, stabilizing the interstrand relationship in the actin the presence of an antiparallel dimer in the crystal filament.36 Thus, intermonomer interactions would implies that a subpopulation existed in the prepara- be expected to be destabilized by oxidation of tions of oxidized actin (see Figure 3;higher Cys374, a reaction that, as shown here for β-actin, molecular mass species). Consequently, the dimer results in depolymerization. is arranging under non-polymerizing conditions. H2O2 influences the organization of actin micro- Previously, it was shown that an antiparallel dimer filaments and motility of many eukaryotic cells, can assemble during different polymerizing con- which are known to contain significant amounts of ditions.45,49 However, the driving force for its glutathionylated actin. It is conceivable that oxida- origination seems to be the tendency of actin to tion of both Cys272 and Cys374 with subsequent form a disulfide bond via Cys374 due to oxidation of glutathionylation could be important in the disas- its sulfur atom. This is consistent with reports that sembly of filaments as the environment becomes slow thiol oxidation of monomeric α-actin by more oxidative during a shift in the reduced/oxi- atmospheric oxygen gives rise to disulfide-bonded dized glutathione (GSH/GSSH) equilibrium. Fur- antiparallel dimers.48 In the presence of crosslinking thermore, oxidative modification of Cys272 might reagents, oxidized Cys374 may react readily with alter the interaction of filamentous actin with actin- those, which then may restrict certain functional binding proteins. conformational changes.44 In light of our observations, it seems conceivable that the observed Disulfide bond: architecture and environment disulfide bond in the α-actin antiparallel dimer did form due to oxidation of a fraction of non-cross- In the crystal structure of the oxidized β-actin, linked α-actin, present together with cross-linked Cys374 in protomers B and D are engaged in an protein.49 The modeled intermolecular disulfide intermolecular disulfide bond, which is partly bond in α-actin homodimer crystal structures solvent-accessible in the crystal lattice. The disulfide appeared to be susceptible to breakage by synchro- bond appears to be in an energetically unfavorable ton radiation or refined to an unfavorable torsion 342 Redox Regulation of β-Actin angle, corroborating our argument for the general the actin filaments further away from an advancing lability of the intermolecular disulfide bond in actin cell edge are long and apparently unbranched,6 – dimers of this kind.49 51 there must be a mechanism for remodeling the filaments within a short time-span.7,55 If, as sug- Implications of an antiparallel dimer formed gested here, the Arp2/3 were bound to the junction under oxidizing conditions formed by an antiparallel dimer, dissociation of the branched filaments could be achieved by reduction Transient interactions are increasingly being of the disulfide bond in a more reducing environ- recognized as crucial in modulating protein–protein ment, further away from the cell edge. Such a recognition and function in regulatory and signal mechanism could explain the fact that in vitro, transduction processes.52 The cellular significance of branches form primarily on newly formed filaments, an antiparallel dimer is unclear. The apparent rather than on older ones.7 In the absence of actin- lability of the dimer assembly and the disulfide binding proteins, mechanical strain on the unstable bond implies that the association represents a non- disulfide bond, produced by the branched filaments, permanent interaction. Thus, the antiparallel dimer may suffice to break it, consistent with the observa- may play a role as a rapidly activated regulatory tion of debranching of filaments with integrated component, reactive and reversible towards changes antiparallel dimers.54 in the oxidative environment. A dimer interaction could be generally unstable but, upon formation of a H2O2 and the impact on the actin microfilament disulfide bond, become more lasting under oxidiz- system ing conditions, such as signal transduction by H2O2. The in vitro formation of disulfide-bonded anti- ROS appear to be required effectors in the control parallel actin dimers might be important for under- of actin-based activities in various cell types.2,3 standing cellular redox responses to growth factor However, a central question is whether ROS act or integrin stimulation, including branching of directly on actin in vivo. Recent redox proteomics filaments within lamellipodia of motile cells.7 The studies detected actins amongst the most prominent advancement of a cell edge depends on polymeriza- proteins oxidized in response to exposure of cells to – tion of actin, formation of dynamic filament ensem- oxidants.8 10 Given our observations about the bles, and disassembly of the actin filaments. During profound mechanistic impact of H2O2 on actin in the process of lamellipodia formation, the multi- vitro, we reason that in vivo a direct redox control of protein complex Arp2/3 appears to be engaged in actin could be one of the most important processes nucleation of branches, though the exact mechanism regulating the dynamics of the microfilament sys- by which Arp2/3 causes filaments to branch is tem. Indeed, observation of a direct interaction in unknown.53 In electron microscopy images, long, vivo between ROS and actin was reported recently.8 unbranched filaments are the dominating structures Clearly, for H2O2 to play a role in signaling to the in lamellipodia, and bundles of long seemingly microfilament system, its site of production must be unbranched filaments constitute the core of filopo- localized near the system, because efficient reduc- dia, suggesting that more than one mechanism tion systems, such as glutathione peroxidase, per- 56 exists for the formation of actin filaments in ad- oxiredoxins, and catalase rapidly eliminate H2O2. vancing cell edges. An initial oxidative burst would first have to Integration of an entire antiparallel β-actin dimer, overcome such oxidative defenses and necessitate 57 in the conformation observed in the crystal, into a amplification. However, a H2O2 gradient may also single helical actin filament is not compatible with have differential oxidative effects of functional the polarity of the filament. However, it has been relevance, leading to dimer formation, glutathiony- demonstrated that antiparallel α-actin dimers lation, and depolymerization in the actin system, become integrated into growing actin filaments in depending on the location of the actin molecules, the vitro and participate directly in actin filament source of the oxidant, the availability of surrounding growth.48,54 Small protrusions at intervals along reducing systems, and the presence of glutathione. the filaments were observed by electron microscopy, The major sources of ROS within both phagocyto- suggesting that one of the protomers of the dimer tic and non-phagocytotic cells are NAD(P)H oxi- had been incorporated into the filaments. Evidence dase-like protein complexes and lipoxygenase. Both for branched filaments was also detected, reminis- function downstream of cytokine receptors,58,59 and cent of the branching filaments seen in the leading a direct interaction of their components with actin edge of highly motile cells. One could imagine that was confirmed.60,61 These enzyme complexes are incorporated dimer sites constitute platforms for the membrane-associated and activated by association binding of Arp2/3 to commence the formation of with the GTPase Rac,62 over-expression of which daughter filaments. It is possible that the inherent leads to increased levels of ROS in endothelial cells.63 flexibility around the disulfide bond of a filament- Consistently, ROS generation was inhibited by over- integrated antiparallel dimer would necessitate expression of a dominant negative form of Rac1.64 In actin-binding proteins to stabilize a defined struc- turn, in all cell types examined so far, Rac serves as a tural state. Thus, Arp2/3 could enforce a particular pivotal regulator of actin assembly and seems to conformation of the dimer to facilitate directed determine where and when actin polymerizes.65 growth of the filament at a particular angle. Since Association with guanine nucleotide exchange Redox Regulation of β-Actin 343 factors appears to target Rac1 to the membrane filament disassembly in vivo.Thein vitro results where, in migrating cells, actin reorganization also shown here suggest that the possibility of oxidation appears to be maximal, and causes its concomitant, of filaments as a step in the depolymerization spatially restricted activation.66 Amplification of an should be given serious consideration. H2O2 signal in vivo seems to occur through the In conclusion, the evidence suggests strongly that integration of the Rac and phosphatidylinositol lipid many aspects of the actin response to various cellular kinase pathways.57 Increased phosphorylation, stimuli occur under oxidizing conditions. This view through oxidative inhibition of phosphatases, is a is further strengthened by observations that a per- hallmark of H2O2 signaling, and H2O2 is known to turbed oxidative balance with increased ROS levels downregulate PTEN by oxidation, leading to seems to be symptomatic of a number of disease increased formation of lipid kinase products (phos- types as well as cellular senescence.77,78 Over- phatidylinositol 4,5 bisphosphate, PIP2 and the 3,4,5- oxidation may have severe consequences, as oxida- trisphosphate, PIP3).67 These phosphoinositides, in tive damage is surmised to occur in neurodegenera- turn, appear to target guanine nucleotide exchange tive diseases.79 Actin oxidation was elevated in brain factors to the membrane, thus activating Rac1.68 extracts of patients with Alzheimer's disease,80 actin Consequently, the impact of Rac on the actin- glutathionylation was increased in cells of patients containing network in vivo, in addition to its apparent with Friedreich's ataxia,11 and disulfide-linked actin protein interaction-mediated role,69 could be ratio- oligomers in post-ischemic reperfusion in rat hearts nalized to a significant extent by its induction of were reported.12 In autism, a serious neurobiological the generation of ROS. This may explain Rac activ- condition, neuronal migration defects and synaptic ities previously assumed to be unrelated, such as deficiencies have been suggested as the causes of activation of lipoxygenase and NAD(P)H oxidase- hippocampal and amygdalar dysfunctions. PTEN, like protein complexes and actin reorganization. The discussed above in connection with it being the link potential role of ROS as initiators and central between oxidation by H2O2 and Rac, is implicated in molecules in the process may not have been appre- the etiology of autism and would appear to exert its ciated sufficiently, since research into the signaling influence on brain development via the actin micro- cascades has focused primarily on identifying pro- filament system.81 tein–protein interactions, mostly under reducing The relevance of ROS action on the actin system in conditions. vivo remains to be determined. Our results provide a The formation of PIP2 caused by binding of basis for an understanding of the effect of H2O2 on agonists to cell-surface receptors in platelets peaks the microfilament. Undoubtedly, future studies will in 5–10 s after the addition of agonists.70,71 Hydro- have to take into account also the effect of ROS on lysis of PIP2 releases inositol trisphosphate, which in actin in conjunction with actin-binding proteins. turn releases Ca2+ from cytosolic stores with about the same kinetics.71 Lamellipodia 2–4 μm long grow out in about 30–60 s after addition of agonists as Materials and Methods seen with platelets and serum-starved cells.72,73 Accumulation of PIP2 in the plasma membrane in Chemicals local areas, lipid rafts, could bind profilin:actin, with the consequent release of actin monomers in the 74,75 ATP and ADP were from Roche; Hypatite C was from neighborhood of the PIP2-producing centers. At Clarkson Chromatography Products, South Williamsport, the same time, the receptor-associated mechanisms PA; DTT was from Saveen Werner AB; guanidine-HCl was could deliver increasing amounts of H2O2 into the from Fluka; and N-(1-pyrene) iodoacetamide was from advancing cell edges, although maximal concentra- Molecular Probes. All other chemicals were from Sigma. tion of the oxidant is not reached until 5–10 min after The purity of ADP was confirmed by chromatography on addition of a growth factor, and later in spreading Mono Q resin (GE Healthcare). Thioredoxin and thior- 1,76 82 cells. The peak of H2O2 is reached long after edoxin reductase were produced in Escherichia coli. 2+ 33 Vivaspin protein concentrators were from Vivascience. initial waves of PIP2-formation, Ca release, and polymerization of actin have passed, and at a time when depolymerization has become a significant Preparation of actin and profilin contributor to steady-state turnover of filaments in advancing cell edges. Calf thymus profilin:actin was prepared and profilin and It is plausible that in the early phase of oxidation, actin were isolated as described.83 Filamentous actin was the release of monomeric actin from profilin:actin, disassembled by extensive dialysis against 5 mM Tris-HCl μ interacting with PIP2 in the periphery of the cell, (pH 7.6), 0.2 mM ATP, 0.1 mM CaCl2,10 M EDTA, 0.5 mM becomes involved in actin dimer formation, and that DTT at 4 °C, and finally gel-filtered using Sephacryl S-300 Arp2/3-dependent polymerization of actin involves (S-300, HR, GE Healthcare, Sweden) equilibrated with G-buffer (5 mM KH2PO4/K2HPO4 (pH 7.6), 0.5 mM ATP, dimer incorporation into filaments, augmenting μ 0.1 mM CaCl2,10 M EDTA). To lower the concentration of branch formation. It is not inconceivable that in the free Ca2+ to micromolar levels, 0.1 mM EGTAwas added to later phase, when the H2O2 level is at a maximum, the buffer. For preparation of ADP-actin the S-300 column glutathionylation and depolymerization of fila- was equilibrated with G-buffer containing 0.5 mM ADP. ments potentiate the turnover of filaments. Rela- Before experiments, DTT was removed by filtration on tively little is known about the mechanism of actin PD-10 columns (GE Healthcare, Sweden) or by S-300 344 Redox Regulation of β-Actin

μ chromatography equilibrated with G-buffer without DTT. Tris-HCl (pH 7.6), 0.5 mM ADP, 0.1 mM CaCl2,10 M In some experiments, actin was kept in G-buffer containing EGTA, 100 mM KCl. Actin from fractions 45–50 (Figure 3(b)) μ α 50 MCaCl2. -Actin from rabbit muscle was prepared as was concentrated to 5 mg/ml. Crystals were grown at 4 °C described,84 and subjected to S-300 chromatography as in hanging drops (Linbro plates), mixing equal volumes of above. Concentrations of actin were determined spectro- protein (5 mg/ml) and crystallization buffer (0.1 M Hepes 85 photometrically. SDS-PAGE was performed with or with- (pH 7.5), 1.6 M (NH4)2SO4, 0.1 M NaCl). Crystals with out reducing agents.86 maximal dimension of 140 μm appeared after about three months. They were transferred to a cryo-solution (0.1 M Hepes (pH 7.5), 1.8 M (NH ) SO , 0.1 M NaCl, 0.1 M KCl, Determination of SH groups 4 2 4 1mMCaCl2, 10 mM ADP) with 5% (v/v) glycerol, and then to a cryo-solution containing 25% (v/v) glycerol, with Protein thiol groups were determined using DTNB and – − − incubation periods of 1 5minforboth. applying a molar extinction coefficient of 13,600 M 1cm 1 87 at 412 nm (A412). The total number of reduced thiol groups was verified by mixing one volume of actin (7– Diffraction data processing, structure solution, and 20 μM) with five volumes of DTNB in guanidine refinement hydrochloride (Gdn-HCl). DTNB was dissolved in 99.5% (v/v) ethanol (10 mM) and mixed in a 1:9 (v/v) ratio with Crystals, mounted in cryoloops, were flash-frozen in ∼ 6 M Gdn-HCl in 0.2 M Tris (pH 8.0). The background liquid N2. X-ray diffraction data, extending to 2.45 Å signal was established by mixing one volume of G-buffer resolution, were measured from a single crystal at −173 °C, with five volumes of DTNB in Gdn-HCl. Accessible thiol in 180 frames of 1° increment, at the X06SA PXI beamline groups in the folded protein in G-buffer were determined (Swiss Light Source) with a Mar225 CCD detector. by mixing one volume of actin sample (7–20 μM) with six Reflections were indexed and integrated with XDS and 89 volumes of 0.5 mg/ml DTNB in G-buffer, while monitor- scaled in XSCALE. A decrease in the intensity of the ing A412 for 8 min. higher-resolution reflections and a marked deterioration of the data-merging statistics with increasing data collection time were indicative of significant radiation damage. Data Oxidation and reduction of actin conversion into structure factor amplitudes and subsequent format adaptions were performed with programs of Oxidation of actin for preparative work was performed the CCP4 suite (version 6.0.0-2).90 The crystal structure was 91 by incubating the protein with H2O2, after which the determined by molecular replacement using Phaser with oxidant was removed by passing over a PD10 or an S-300 the protein part of α-actin (PDB-code 1J6Z) as a search column equilibrated with G-buffer (4 °C). The polymeriz- model.92 Data processing is consistent with space group ability of oxidized ADP-actin and ATP-actin was analyzed C222 with four molecules in the asymmetric unit. – 1 by incubating 0.4 0.5 mg/ml protein in 10 mM KH2PO4/ Crystallographic rigid body refinement and torsion-angle μ K2HPO4 (pH 7.6), 0.1 mM CaCl2, 0.1 mM EGTA, 10 M molecular dynamics simulated annealing refinement with EDTA, supplemented with 0.2 mM ADP or ATP, respec- a slow-cooling protocol starting at 4726 °C (5000 K) with 93 tively, with 5 mM H2O2 for 45 min at 25 °C. Then, CNS, using diffraction data to 2.5 Å resolution were polymerizing salts were added to a concentration of 2 mM followed by restrained, isotropic B-factor refinement with 94 MgCl2 and 100 mM KCl, and the change in viscosity at Refmac. A 5% test set of randomly selected reflections 25 °C was monitored. Reduction was performed by was used for cross-validation throughout refinement,95 incubating the actin with the Trx-system (2 μM thioredoxin, which was iterated by manual rebuilding cycles in σA- 50 nM thioredoxin reductase, 0.2 mM NADPH) at 37 °C for weighted electron density maps with COOT.96 Correction the lengths of time indicated. A sample containing the Trx- of the amino acid sequence register for β-actin, addition of system components in G-buffer was used to determine ligands, non-crystallographic symmetry averaging with their contribution of DTNB-reacting thiol groups. differential weighting for variable parts, and refinement of protein domain-specific TLS parameters improved the 97 2+ Actin polymerization data fit. Refinement parameters for ATP,sulfate, and Ca were retrieved from the CCP4 library,90 and those for cysteine sulfinic acid, added in progressed refinement Actin polymerization was monitored by high-shear cycles, were generated with PRODRG.98 Refinement was viscometry at 25 °C using a Cannon-Manning viscometer, continued until convergence of the cross-validated R- with a buffer flow-time of 60 s and a sample volume of factor. It was then realized that the electron density at the 0.7 ml. Filament formation was also recorded using the β γ 88 C termini of two molecules was potentially indicative of a pyrene assay with 2% pyrene-labeled / -ADP-actin. disulfide bond. The data were reprocessed including only Pyrene-labeled actin was isolated by gel-filtration on an S- the first 90° of data (to 2.6 Å resolution), where radiation 300 chromatography column equilibrated with G-buffer damage was less severe. Using the same reflection indices containing ADP. Polymerization was induced by addition as before for cross-validation, simulated annealing refine- of polymerizing salts as indicated in the Figure legends. ment was performed (as described above), with the previously refined model and residues C-terminal of – Crystallization of oxidized ADP-actin His371 removed (Table 3). Analogously, data from 90 180° were processed separately and the model refined with (data not shown). Isolation of profilin:β-actin was performed as described above. Fractions eluted from the column were immediately supplemented with ATP and the protein was precipitated Model analysis with 80% (w/v) (NH4)2SO4 and stored at 4 °C. The ADP- form of β-actin was prepared as described above. The The protein model was evaluated with MolProbity,99 100 101 protein was then incubated with 20 mM H2O2 at 37 °C for SFCHECK, and WHATCHECK. Structural super- 40minandfinallypassedoveranS-300columnin5mM position was done with LSQKAB.90 Surfaces were Redox Regulation of β-Actin 345 analyzed with the Protein-Protein interaction server,102 8. Fiaschi, T., Cozzi, G., Raugei, G., Formigli, L., PISA,103 and programs of the CCP4 suite. Molecular Ramponi, G. & Chiarugi, P. (2006). Redox regulation representations were prepared with PyMol‡. of beta-actin during integrin-mediated cell adhesion. J. Biol. Chem. 281, 22983–22991. 9. Fratelli, M., Demol, H., Puype, M., Casagrande, S., Protein Data Bank accession number Eberini, I., Salmona, M. et al. (2002). Identification by redox proteomics of glutathionylated proteins in Atomic coordinates and structure factors for the protein oxidatively stressed human T lymphocytes. Proc. Natl model have been deposited in the RCSB Protein Data Bank Acad. Sci. USA, 99, 3505–3510. with accession code 2OAN. 10. Lind, C., Gerdes, R., Hamnell, Y., Schuppe-Koistinen, I., von Lowenhielm, H. B., Holmgren, A. & Cot- greave, I. A. (2002). Identification of S-glutathionylated cellular proteins during oxidative stress and constitutive metabolism by affinity purification and Acknowledgements proteomic analysis. Arch. Biochem. Biophys. 406, 229–240. We are grateful to Ingegärd Andersson for tech- 11. Pastore, A., Tozzi, G., Gaeta, L. M., Bertini, E., Serafini, V., Di Cesare, S. et al. (2003). Actin nical assistance, and Terese Bergfors for introducing glutathionylation increases in fibroblasts of patients I.L. to protein crystallization. Herwig Schüler for with Friedreich's ataxia: a potential role in the help in determining the Kd for the interaction of pathogenesis of the disease. J. Biol. Chem. 278, oxidized actin and DNase I. We thank Andreas Kohl 42588–42595. and Damian Niegowski for help with the diffraction 12. Canton, M., Neverova, I., Menabo, R., Van Eyk, J. & data collection. This work was performed at the Di Lisa, F. (2004). Evidence of myofibrillar protein Swiss Light Source, Paul Scherrer Institut, Villigen, oxidation induced by postischemic reperfusion in isolated rat hearts. Am. J. Physiol. Heart Circ. Physiol. Switzerland. We are grateful to the machine and 286 – beamline groups, whose outstanding efforts have , H870 H877. made these experiments possible. This work was 13. Dalle-Donne, I., Rossi, R., Milzani, A., Di Simplicio, P. & Colombo, R. (2001). The actin cytoskeleton supported by grants from the Swedish Cancer response to oxidants: from small heat shock protein Society to I.L., F.S., A.H., P.N., and U.L, and by a phosphorylation to changes in the redox state of actin grant from Magnus Bergvall foundation to I.L. itself. Free Radic. Biol. Med. 31, 1624–3162. Funding from the Swedish research council to P.N. 14. Drewes, G. & Faulstich, H. (1991). A reversible and U.L., and from the Nancy Lurie Marks Family conformational transition in muscle actin is caused Foundation to C.E.S. is acknowledged. by nucleotide exchange and uncovers cysteine in position 10. J. Biol. Chem. 266, 5508–5513. 15. Blikstad, I., Markey, F., Carlsson, L., Persson, T. & References Lindberg, U. (1978). Selective assay of monomeric and filamentous actin in cell extracts, using inhibition 15 – 1. Sundaresan, M., Yu, Z. X., Ferrans, V. J., Irani, K. & of deoxyribonuclease I. Cell, , 935 943. Finkel, T. (1995). Requirement for generation of H2O2 16. Blikstad, I. & Carlsson, L. (1982). On the dynamics of the microfilament system in HeLa cells. J. Cell Biol. 93, for platelet-derived growth factor signal transduc- – tion. Science, 270, 296–299. 122 128. 2. Moldovan, L., Mythreye, K., Goldschmidt-Clermont, 17. Schutt, C. E., Myslik, J. C., Rozycki, M. D., P. J. & Satterwhite, L. L. (2006). Reactive oxygen Goonesekere, N. C. & Lindberg, U. (1993). The structure of crystalline profilin-beta-actin. Nature, 365, species in vascular endothelial cell motility. Roles of – NAD(P)H oxidase and Rac1. Cardiovasc. Res. 71, 810 816. 236–246. 18. Chik, J. K., Lindberg, U. & Schutt, C. E. (1996). The structure of an open state of beta-actin at 2.65 Å 3. Chiarugi, P. & Fiaschi, T. (2007). Redox signalling in 263 – anchorage-dependent cell growth. Cell Signal, 19, resolution. J. Mol. Biol. , 607 623. – 19. Page, R., Lindberg, U. & Schutt, C. E. (1998). Domain 672 682. 280 – 4. Hong, J. H., Moon, S. J., Byun, H. M., Kim, M. S., Jo, motions in actin. J. Mol. Biol. , 463 474. H., Bae, Y. S. et al. (2006). Critical role of phospho- 20. Petersen, M. T., Jonson, P. H. & Petersen, S. B. (1999). lipase Cgamma1 in the generation of H2O2-evoked Amino acid neighbours and detailed conformational analysis of cysteines in proteins. Protein Eng. 12, [Ca2+]i oscillations in cultured rat cortical astrocytes. – J. Biol. Chem. 281, 13057–13067. 535 548. 5. Watanabe, N. & Mitchison, T. J. (2002). Single- 21. Perahia, D. & Pullman, B. (1971). The conformational energy map for the disulphide bridge in proteins. molecule speckle analysis of actin filament turnover 43 – in lamellipodia. Science, 295, 1083–1086. Biochem. Biophys. Res. Commun. ,65 68. 22. Schmidt, B., Ho, L. & Hogg, P. J. (2006). Allosteric 6. Small, J. V., Stradal, T., Vignal, E. & Rottner, K. (2002). 45 – The lamellipodium: where motility begins. Trends disulfide bonds. Biochemistry, , 7429 7433. Cell Biol. 12,112–120. 23. Forman, H. J., Fukuto, J. M. & Torres, M. (2004). 7. Pollard, T. D. & Borisy, G. G. (2003). Cellular motility Redox signaling: thiol chemistry defines which reactive oxygen and nitrogen species can act as driven by assembly and disassembly of actin 287 filaments. Cell, 112, 453–465. second messengers. Am. J. Physiol. Cell Physiol. , C246–C256. 24. Kwon, J., Lee, S. R., Yang, K. S., Ahn, Y., Kim, Y. J., Stadtman, E. R. & Rhee, S. G. (2004). Reversible ‡ http://www.pymol.org oxidation and inactivation of the tumor suppressor 346 Redox Regulation of β-Actin

PTEN in cells stimulated with peptide growth lytic mechanism. Proc. Natl Acad. Sci. USA, 100, factors. Proc. Natl Acad. Sci. USA, 101, 16419–16424. 5760–5765. 25. Aspenstrom, P., Schutt, C. E., Lindberg, U. & 42. Klenchin, V. A., Khaitlina, S. Y. & Rayment, I. (2006). Karlsson, R. (1993). Mutations in beta-actin: influence Crystal structure of polymerization-competent actin. on polymer formation and on interactions with J. Mol. Biol. 362, 140–150. myosin and profilin. FEBS Letters, 329, 163–170. 43. Bhattacharyya, R., Pal, D. & Chakrabarti, P. (2004). 26. Chivers, P. T. & Raines, R. T. (1997). General acid/ Disulfide bonds, their stereospecific environment base catalysis in the active site of Escherichia coli and conservation in protein structures. Protein Eng. thioredoxin. Biochemistry, 36, 15810–15816. Des. Sel. 17, 795–808. 27. LeMaster, D. M., Springer, P. A. & Unkefer, C. J. (1997). 44. Hesterkamp, T., Weeds, A. G. & Mannherz, H. G. The role of the buried aspartate of Escherichia coli (1993). The actin monomers in the ternary gelsolin: 2 thioredoxin in the activation of the mixed disulfide actin complex are in an antiparallel orientation. Eur. J. intermediate. J. Biol. Chem. 272, 29998–30001. Biochem. 218, 507–513. 28. Stournaras, C., Drewes, G., Blackholm, H., Merkler, I. 45. Millonig, R., Salvo, H. & Aebi, U. (1988). Probing & Faulstich, H. (1990). Glutathionyl(cysteine-374) actin polymerization by intermolecular cross-linking. actin forms filaments of low mechanical stability. J. Cell Biol. 106, 785–796. Biochim. Biophys. Acta, 1037,86–91. 46. Mockrin, S. C. & Korn, E. D. (1981). Isolation and 29. Pelikan Conchaudron, A., Didry, D., Le, K. H., characterization of covalently cross-linked actin Larquet, E., Boisset, N., Pantaloni, D. & Carlier, M. dimer. J. Biol. Chem. 256, 8228–8233. F. (2006). Analysis of tetramethylrhodamine-labeled 47. Schoenenberger, C. A., Bischler, N., Fahrenkrog, B. & actin polymerization and interaction with actin Aebi, U. (2002). Actins propensity for dynamic regulatory proteins. J. Biol. Chem. 281, 24036–24047. filament patterning. FEBS Letters, 529,27–33. 30. Mossakowska, M., Moraczewska, J., Khaitlina, S. & 48. Tang, J. X., Janmey, P. A., Stossel, T. P. & Ito, T. (1999). Strzelecka-Golaszewska, H. (1993). Proteolytic Thiol oxidation of actin produces dimers that removal of three C-terminal residues of actin alters enhance the elasticity of the F-actin network. Biophys. the monomer-monomer interactions. Biochem. J. 289, J. 76, 2208–2215. 897–902. 49. Bubb, M. R., Govindasamy, L., Yarmola, E. G., 31. Crosbie, R. H., Miller, C., Cheung, P., Goodnight, T., Vorobiev, S. M., Almo, S. C., Somasundaram, T. et al. Muhlrad, A. & Reisler, E. (1994). Structural con- (2002). Polylysine induces an antiparallel actin dimer nectivity in actin: effect of C-terminal modifications that nucleates filament assembly: crystal structure at on the properties of actin. Biophys. J. 67, 1957–1964. 3.5-Å resolution. J. Biol. Chem. 277,20999–21006. 32. Kim, E., Wriggers, W., Phillips, M., Kokabi, K., 50. Margarit, S. M., Davidson, W., Frego, L. & Stebbins, Rubenstein, P. A. & Reisler, E. (2000). Cross-linking C. (2006). A steric antagonism of actin polymeriza- constraints on F-actin structure. J. Mol. Biol. 299, tion by a Salmonella virulence protein. Structure, 14, 421–429. 1219–1229. 33. Singh, D. K., Kumar, D., Siddiqui, Z., Basu, S. K., 51. Reutzel, R., Yoshioka, C., Govindasamy, L., Yarmola, Kumar, V. & Rao, K. V. (2005). The strength of E. G., Agbandje-McKenna, M., Bubb, M. R. & receptor signaling is centrally controlled through a McKenna, R. (2004). Actin crystal dynamics: struc- cooperative loop between Ca2+ and an oxidant tural implications for F-actin nucleation, polymeriza- signal. Cell, 121, 281–293. tion, and branching mediated by the anti-parallel 34. Kim, J. R., Kwon, K. S., Yoon, H. W., Lee, S. R. & Rhee, dimer. J. Struct. Biol. 146, 291–301. S. G. (2002). Oxidation of proteinaceous cysteine 52. Blundell, T. L. & Fernandez-Recio, J. (2006). Cell residues by dopamine-derived H2O2 in PC12 cells. biology: brief encounters bolster contacts. Nature, Arch. Biochem. Biophys. 397, 414–423. 444, 279–280. 35. O'Donoghue, S. I., Miki, M. & dos Remedios, C. G. 53. Goley, E. D. & Welch, M. D. (2006). The ARP2/3 (1992). Removing the two C-terminal residues of complex: an actin nucleator comes of age. Nature Rev. actin affects the filament structure. Arch. Biochem. Mol. Cell Biol. 7, 713–726. Biophys. 293,110–116. 54. Steinmetz, M. O., Goldie, K. N. & Aebi, U. (1997). A 36. Holmes, K. C., Popp, D., Gebhard, W. & Kabsch, W. correlative analysis of actin filament assembly, (1990). Atomic model of the actin filament. Nature, structure, and dynamics. J. Cell Biol. 138, 559–574. 347,44–49. 55. Le Clainche, C., Pantaloni, D. & Carlier, M. F. (2003). 37. Weik, M., Ravelli, R. B., Kryger, G., McSweeney, S., ATP hydrolysis on actin-related protein 2/3 complex Raves, M. L., Harel, M. et al. (2000). Specific chemical causes debranching of dendritic actin arrays. Proc. and structural damage to proteins produced by Natl Acad. Sci. USA, 100, 6337–6342. synchrotron radiation. Proc. Natl Acad. Sci. USA, 97, 56. Choi, M. H., Lee, I. K., Kim, G. W., Kim, B. U., Han, Y. 623–628. H., Yu, D. Y. et al. (2005). Regulation of PDGF 38. Bagley, S. C. & Altman, R. B. (1995). Characterizing signalling and vascular remodelling by peroxire- the microenvironment surrounding protein sites. doxin II. Nature, 435, 347–353. Protein Sci. 4, 622–635. 57. Rhee, S. G., Bae, Y. S., Lee, S. R. & Kwon, J. (2000). 39. Tirion, M. M., ben-Avraham, D., Lorenz, M. & Hydrogen peroxide: a key messenger that modulates Holmes, K. C. (1995). Normal modes as refinement protein phosphorylation through cysteine oxidation. parameters for the F-actin model. Biophys. J. 68,5–12. Sci. STKE, 2000, PE1. 40. Rould, M. A., Wan, Q., Joel, P. B., Lowey, S. & Trybus, 58. Hu, T., Ramachandrarao, S. P., Siva, S., Valancius, K. M. (2006). Crystal structures of expressed non- C., Zhu, Y., Mahadev, K. et al. (2005). Reactive polymerizable monomeric actin in the ADP and ATP oxygen species production via NADPH oxidase states. J. Biol. Chem. 281, 31909–31919. mediates TGF-beta-induced cytoskeletal alterations 41. Vorobiev, S., Strokopytov, B., Drubin, D. G., Frieden, in endothelial cells. Am. J. Physiol. Renal Physiol. 289, C., Ono, S., Condeelis, J. et al. (2003). The structure of F816–F825. nonvertebrate actin: implications for the ATP hydro- 59. Radmark, O. & Samuelsson, B. (2005). Regulation of Redox Regulation of β-Actin 347

5-lipoxygenase enzyme activity. Biochem. Biophys. oxidative inhibition of a FAK tyrosine phosphatase is Res. Commun. 338, 102–110. required for cell adhesion. J. Cell Biol. 161, 933–944. 60. Li, J. M. & Shah, A. M. (2002). Intracellular localiza- 77. Finkel, T. & Holbrook, N. J. (2000). Oxidants, tion and preassembly of the NADPH oxidase oxidative stress and the biology of ageing. Nature, complex in cultured endothelial cells. J. Biol. Chem. 408, 239–247. 277, 19952–19960. 78. Schulz, J. B., Lindenau, J., Seyfried, J. & Dichgans, J. 61. Provost, P., Doucet, J., Hammarberg, T., Gerisch, G., (2000). Glutathione, oxidative stress and neurode- Samuelsson, B. & Radmark, O. (2001). 5-Lipoxygen- generation. Eur. J. Biochem. 267, 4904–4911. ase interacts with coactosin-like protein. J. Biol. Chem. 79. Finkel, T. (2005). Opinion: Radical medicine: treating 276, 16520–16527. ageing to cure disease. Nature Rev. Mol. Cell Biol. 6, 62. Abo, A., Pick, E., Hall, A., Totty, N., Teahan, C. G. & 971–976. Segal, A. W. (1991). Activation of the NADPH 80. Aksenov, M. Y., Aksenova, M. V., Butterfield, D. A., oxidase involves the small GTP-binding protein Geddes, J. W. & Markesbery, W. R. (2001). Protein p21rac1. Nature, 353, 668–670. oxidationinthebraininAlzheimer'sdisease. 63. Moldovan, L., Irani, K., Moldovan, N. I., Finkel, T. & Neuroscience, 103, 373–383. Goldschmidt-Clermont, P. J. (1999). The actin cytos- 81. Kwon, C. H., Luikart, B. W., Powell, C. M., Zhou, J., keleton reorganization induced by Rac1 requires the Matheny, S. A., Zhang, W. et al. (2006). Pten regulates production of superoxide. Antioxid Redox Signal, 1, neuronal arborization and social interaction in mice. 29–43. Neuron, 50, 377–388. 64. Irani, K., Xia, Y., Zweier, J. L., Sollott, S. J., Der, C. J., 82. Holmgren, A. & Bjornstedt, M. (1995). Thioredoxin Fearon,E.R.et al. (1997). Mitogenic signaling and thioredoxin reductase. Methods Enzymol. 252, mediated by oxidants in Ras-transformed fibroblasts. 199–208. Science, 275, 1649–1652. 83. Schüler, H., Karlsson, R. & Lindberg, U. (2006). Puri- 65. Meili, R. & Firtel, R. A. (2003). Two poles and a fication of nonmuscle actin. In Cell Biology: A Laboratory compass. Cell, 114, 153–156. Handbook (Celis, J., ed.) vol 2, pp. 165–171. Elsevier 66. Kraynov, V. S., Chamberlain, C., Bokoch, G. M., Science, New York. Schwartz, M. A., Slabaugh, S. & Hahn, K. M. (2000). 84. Pardee, J. D. & Spudich, J. A. (1982). Purification of Localized Rac activation dynamics visualized in muscle actin. Methods Enzymol. 85, 164–181. living cells. Science, 290, 333–337. 85. Houk, T. W., Jr & Ue, K. (1974). The measurement of 67. Rhee, S. G., Kang, S. W., Jeong, W., Chang, T. S., Yang, actin concentration in solution: a comparison of K. S. & Woo, H. A. (2005). Intracellular messenger methods. Anal. Biochem. 62,66–74. function of hydrogen peroxide and its regulation by 86. Laemmli, U. K. (1970). Cleavage of structural peroxiredoxins. Curr. Opin. Cell Biol. 17, 183–189. proteins during the assembly of the head of 68. Bae, Y. S., Sung, J. Y., Kim, O. S., Kim, Y. J., Hur, K. C., bacteriophage T4. Nature, 227, 680–685. Kazlauskas, A. & Rhee, S. G. (2000). Platelet-derived 87. Ellman, G. L. (1959). Tissue sulfhydryl groups. Arch. growth factor-induced H(2)O(2) production requires Biochem. Biophys. 82,70–77. the activation of phosphatidylinositol 3-kinase. J. 88. Kouyama, T. & Mihashi, K. (1981). Fluorimetry study Biol. Chem. 275, 10527–10531. of N-(1-pyrenyl)iodoacetamide-labelled F-actin. 69. Hall, A. (1998). Rho GTPases and the actin cytoske- Local structural change of actin protomer both on leton. Science, 279, 509–514. polymerization and on binding of heavy meromyo- 70. Lassing, I. & Lindberg, U. (1990). Polyphosphoinosi- sin. Eur. J. Biochem. 114,33–38. tide synthesis in platelets stimulated with low 89. Kabsch, W. (1993). Automatic processing of rotation concentrations of thrombin is enhanced before the diffraction data from crystals and initially unknown activation of phospholipase C. FEBS Letters, 262, symmetry and cell constants. J. Appl. Crystallog. 26, 231–233. 795–800. 71. Chong, L. D., Traynor-Kaplan, A., Bokoch, G. M. & 90. Collaborative Computational Project, Number 4. Schwartz, M. A. (1994). The small GTP-binding pro- (1994). The CCP4 suite: programs for protein crystal- tein Rho regulates a phosphatidylinositol 4-phosphate lography. Acta Crystallog. sect. D, 50, 760–763. 5-kinase in mammalian cells. Cell, 79, 507–513. 91. McCoy, A. J., Grosse-Kunstleve, R. W., Storoni, L. C. 72. Karlsson, R., Lassing, I., Hoglund, A. S. & Lindberg, & Read, R. J. (2005). Likelihood-enhanced fast trans- U. (1984). The organization of microfilaments in lation functions. Acta Crystallog. sect. D, 61, 458–464. spreading platelets: a comparison with fibroblasts 92. Otterbein, L. R., Graceffa, P. & Dominguez, R. (2001). and glial cells. J. Cell Physiol. 121,96–113. The crystal structure of uncomplexed actin in the 73. Mellstroom, K., Hoglund, A. S., Nister, M., Heldin, ADP state. Science, 293, 708–711. C. H., Westermark, B. & Lindberg, U. (1983). The 93. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, effect of platelet-derived growth factor on morphol- W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). ogy and motility of human glial cells. J. Muscle Res. Crystallography & NMR system: a new software Cell Motil. 4, 589–609. suite for macromolecular structure determination. 74. Rozelle, A. L., Machesky, L. M., Yamamoto, M., Acta Crystallog. sect. D, 54, 905–921. Driessens, M. H., Insall, R. H., Roth, M. G. et al. (2000). 94. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Phosphatidylinositol 4,5-bisphosphate induces actin- (1997). Refinement of macromolecular structures by based movement of raft-enriched vesicles through the maximum-likelihood method. Acta Crystallog. WASP-Arp2/3. Curr. Biol. 10,311–320. sect. D, 53, 240–255. 75. Ackermann, M. & Matus, A. (2003). Activity-induced 95. Brunger, A. T. (1992). Free R value: a novel statistical targeting of profilin and stabilization of dendritic quantity for assessing the accuracy of crystal spine morphology. Nature Neurosci. 6, 1194–2100. structures. Nature, 355, 472–475. 76. Chiarugi, P., Pani, G., Giannoni, E., Taddei, L., 96. Emsley, P. & Cowtan, K. (2004). Coot: model- Colavitti, R., Raugei, G. et al. (2003). Reactive oxygen building tools for molecular graphics. Acta Crystallog. species as essential mediators of cell adhesion: the sect. D, 60, 2126–2132. 348 Redox Regulation of β-Actin

97. Winn, M. D., Isupov, M. N. & Murshudov, G. N. the quality of macromolecular structure-factor data (2001). Use of TLS parameters to model anisotropic and their agreement with the atomic model. Acta displacements in macromolecular refinement. Acta Crystallog. sect. D, 55, 191–205. Crystallog. sect. D, 57, 122–133. 101. Hooft, R. W., Vriend, G., Sander, C. & Abola, E. E. 98. Schuttelkopf, A. W. & van Aalten, D. M. (2004). (1996). Errors in protein structures. Nature, 381, 272. PRODRG: a tool for high-throughput crystallogra- 102. Jones, S. & Thornton, J. M. (1996). Principles of phy of protein-ligand complexes. Acta Crystallog. sect. protein-protein interactions. Proc. Natl Acad. Sci. D, 60, 1355–1363. USA, 93,13–20. 99. Lovell, S. C., Davis, I. W., Arendall, W. B., 3rd, de 103. Krissinel, E. & Henrick, K. (2005). Detection of Protein Bakker, P. I., Word, J. M., Prisant, M. G. et al. (2003). Assemblies in Crystals. In Protein Interfaces, Surfaces Structure validation by Calpha geometry: phi,psi and and Assemblies Service PISA at European Bioinformatics Cbeta deviation. Proteins: Struct. Funct. Genet. 50, Institute. CompLife 2005. Springer, Berlin-Heidelberg. 437–450. 104. Kabsch, W., Mannherz, H. G., Suck, D., Pai, E. F. & 100. Vagin, A. A., Richelle, J. & Wodak, S. J. (1999). Holmes, K. C. (1990). Atomic structure of the actin: SFCHECK: a unified set of procedures for evaluating DNase I complex. Nature, 347,37–44.

Edited by I. Wilson

(Received 8 February 2007; received in revised form 18 April 2007; accepted 18 April 2007) Available online 4 May 2007