Role of the Actin Ala-108–Pro-112 Loop in Actin Polymerization and ATPase Activities*

Background: Upon polymerization, the actin molecule becomes flattened, from the twisted form to the untwisted form. Results: The A108G substitution in the loop crucial for the flattening slowed down polymerization, and the P109A substitution accelerated polymerization. Conclusion: Actin polymerization involves at least two processes, the formation of the collision complex and its conformational change. Significance: This knowledge is important for understanding the polymerization mechanism. Actin plays fundamental roles in a variety of cell functions in eukaryotic cells. The polymerization-depolymerization cycle, between monomeric G-actin and fibrous F-actin, drives essential cell processes. Recently, we proposed the atomic model for the F-actin structure and found that actin was in the twisted form in the monomer and in the untwisted form in the filament. To understand how the polymerization process is regulated (Caspar, D. L. (1991) Curr. Biol. 1, 30–32), we need to know further details about the transition from the twisted to the untwisted form. For this purpose, we focused our attention on the Ala-108–Pro-112 loop, which must play crucial roles in the transition, and analyzed the consequences of the amino acid replacements on the polymerization process. As compared with the wild type, the polymerization of P109A was accelerated in both the nucleation and the elongation steps, and this was attributed to an increase in the frequency factor of the Arrhenius equation. The multiple conformations allowed by the substitution presumably resulted in the effective formation of the collision complex, thus accelerating polymerization. On the other hand, the A108G mutation reduced the rates of both nucleation and elongation due to an increase in the activation energy. In the cases of polymerization acceleration and deceleration, each functional aberration is attributed to a distinct elementary process. The rigidity of the loop, which mediates neither too strong nor too weak interactions between subdomains 1 and 3, might play crucial roles in actin polymerization.


Actin plays fundamental roles in a variety of cell functions in eukaryotic cells. The polymerization-depolymerization cycle, between monomeric G-actin and fibrous F-actin, drives essential cell processes. Recently, we proposed the atomic model for the F-actin structure and found that actin was in the twisted form in the monomer and in the untwisted form in the filament.
To understand how the polymerization process is regulated (Caspar, D. L. (1991) Curr. Biol. 1, 30 -32), we need to know further details about the transition from the twisted to the untwisted form. For this purpose, we focused our attention on the Ala-108 -Pro-112 loop, which must play crucial roles in the transition, and analyzed the consequences of the amino acid replacements on the polymerization process. As compared with the wild type, the polymerization of P109A was accelerated in both the nucleation and the elongation steps, and this was attributed to an increase in the frequency factor of the Arrhenius equation. The multiple conformations allowed by the substitution presumably resulted in the effective formation of the collision complex, thus accelerating polymerization. On the other hand, the A108G mutation reduced the rates of both nucleation and elongation due to an increase in the activation energy. In the cases of polymerization acceleration and deceleration, each functional aberration is attributed to a distinct elementary process. The rigidity of the loop, which mediates neither too strong nor too weak interactions between subdomains 1 and 3, might play crucial roles in actin polymerization.
Actin is one of the most abundant proteins in eukaryotic cells, where it plays fundamental roles in a variety of cell func-tions, such as locomotion and division (1,2). In cells, actin is present in two states: a monomeric state (G-actin) and a fibrous state (F-actin). The transitions between the two states, i.e. polymerization and depolymerization, drive several essential cellular processes. To understand the processes (29), the atomic structures of G-actin and F-actin are essential. The G-actin crystal structure was solved by Kabsch et al. (3) in 1990. On the other hand, an atomic model for the F-actin structure was first proposed by Holmes et al. (4) in 1990, based on x-ray fiber diffraction analyses. In 2009, we proposed a new model (5) and found conformational changes that are associated with the Gto F-actin transition. A recent study of the F-actin structure, using high resolution electron cryomicroscopy (6), confirmed the conformational changes.
The actin molecule has two major domains enclosing an ATP-binding cleft (3). These domains rotate relative to each other upon the G-to F-actin transition, and thus the actin molecule is flattened in F-actin (5). Within each molecule, the conformational changes are associated with the sliding of subdomain 1 relative to subdomain 3. The interface between two subdomains is formed by the side chains extending from the ␤-sheet core of subdomain 3 (green), the central ␣-helix including Gln-137-Gly-146 (wine), and the loop including Ala-108 -Pro-112 of subdomain 1 (Fig. 1, cyan) (7). The loop shifts substantially relative to subdomain 3 upon sliding. Moreover, the preceding part of the loop contacts the side chain of Gln-137, which participates in both the conformational transition of the actin molecule and the ATPase reaction (8,9). The successive part of the loop contacts the diagonal subunit in F-actin. In addition, the conformation of the loop is strongly restricted by two proline residues, Pro-109 and Pro-112, and it might behave as a stable structural unit.
The determination of the role of the Ala-108 -Pro-112 loop in actin function is essential to understand the mechanism of actin polymerization. To clarify its role, we studied the consequences of altering the Ala-108 -Pro-112 loop in terms of actin polymerization and ATPase activities. We created the A108G and P109A substitutions and prepared the two actin mutants by the use of an insect cell expression system. A108G polymerized slowly, due to an increase in the activation energy, whereas P109A polymerized rapidly, due to an increase in the frequency factor of the Arrhenius equation. P109A may affect the formation of the collision complex between a monomer and the filament end in the polymerization process, whereas A108G probably causes a conformational change of the collision complex crossing a kinetic barrier. By contrast, these substitutions have minimal effects on the actin ATPase activity. Based on these differences between the two mutants, we speculate that the rigidity of the Ala-108 -Pro-112 loop is important, through maintaining a moderate interaction between subdomains 1 and 3, for the process of actin polymerization.

EXPERIMENTAL PROCEDURES
Expression and Purification of Recombinant Actin-Skeletal muscle ␣-actin were prepared from chicken breast dry muscle according to the method of Spudich and Watt (10). Recombinant human cardiac muscle wild type ␣-actin (the WT actin) was prepared according to the method we described previously (8). The N-terminal amino acid sequence of the WT actin is MGWSHPQFEKGGIEGRDDEE; the underlined part is an additional amino acid sequence including the Strep-tag II for purification. The extra N-terminal sequence did not affect the polymerization path of the WT actin, although it modified the rate constant, and furthermore, it hardly affected the overall actin ATPase activity (8). The transfer vector was pVL1392-L21, including the enhancer sequence L21 in the pVL1392 vector (Pharmingen) (11). The transfer vectors for the expression of the actin mutants A108G and P109A were generated from pVL1392-L21, encoding the WT, by the use of a QuikChange II site-directed mutagenesis kit (Agilent Technologies). The mutagenic primers were as follows: A108G, 5Ј-CACCCTGCT-CACAGAGGGCCCGCTGAACCCCAAGG-3Ј, and P109A, 5Ј-CCTGCTCACAGAGGCCGCGCTGAACCCCAAGGC-3Ј, where the corresponding mutation sites are underlined and only the coding sequences are described. The preparation of recombinant baculovirus encoding actin and the purification of recombinant actins were also performed according to the method we described previously (8). In the case of A108G, the method was slightly modified; at the final purification step, the actin was polymerized at 25°C, rather than 4°C, because A108G polymerized extremely poorly at 4°C. The concentration of G-actin was determined from the absorbance at 290 nm, using the extinction coefficient E 290 ϭ 0.63 ml mg Ϫ1 cm Ϫ1 (12).
Biochemical Assays of Actin Polymerization-G-actin was prepared in G-buffer, composed of 10 mM Tris-HCl (pH 8.0), 0.2 mM CaCl 2 , 0.5 mM ATP, and 1 mM DTT. In assays for F-actin, G-actin was polymerized by the addition of a 20-fold concentrated polymerization solution to make final concentrations of 100 mM KCl, 2 mM MgCl 2 , and 0.5 mM ATP. Actin polymerization was detected by 90°light scattering at 660 nm (8). Ultraviolet circular dichroism (CD) spectra, time courses of scattering intensity during polymerization, critical concentrations for polymerization, and time courses of released phosphate (P i ) were measured according to the procedures described previously (8). F-actin was also observed by electron microscopy according to the previously reported procedures (8). The nucleus size for polymerization was estimated from the double logarithm plot of the maximum polymerization rate against the actin concentration (8,13). The activation energy was calculated from an Arrhenius plot of the maximum polymerization rate of 25 M actin at 3-25°C (8,14). In the present analysis, some data for the WT actin were published previously (8). The determination of the number of filaments in solution was performed according to the procedure reported by Pollard (15). Skeletal muscle ␣-actin was labeled by n-(1-pyrene)-iodoacetamide (pyrene actin), according to the procedure reported by Kouyama and Mihashi (16). We prepared two solutions, A and B, and monitored the fluorescence at 407 nm (excitation at 365 nm) after gently mixing the two solutions at 25°C. Solution A was prepared by mixing 15 l of pyrene G-actin (1 mg/ml, label ratio: 5%), 79 l of G-buffer, and 100 l of water. Solution B was prepared by mixing 12 l of high salt solution (2 M KCl, 40 mM MgCl 2 , 6 mM EGTA), 0 -9 l of expressed F-actin solution at steady state (0.4 mg/ml), and 27-36 l of G-buffer. The slope is proportional to the filament concentration of F-actin in solution.
The simplest conventional nucleation-elongation model was used for actin polymerization. Nucleation nA 3 N (A ϭ Gactin; N ϭ nucleation; n ϭ nucleus size) proceeds with the rate constant k n . Elongation N ϩ A 3 F (F ϭ F-actin) proceeds with the rate constant k ϩ . The apparent maximum polymerization rate is proportional to (k n k ϩ ) 1 ⁄ 2 [A] n/2 ϩ 1 , and the filament concentration is proportional to (k n /k ϩ ) 1 ⁄ 2 [A] n/2 . The apparent rate constants for the ATPase on F-actin (k) were calculated by the following scheme: G-actin-ATP 3 Factin-ATP 3 F-actin-ADP ϩ P i . The reverse reactions were left out of the scheme for clarity.

Preparation of Recombinant Human Cardiac
Muscle ␣-Actins in Insect Cells-Recombinant human cardiac muscle ␣-actins (the WT actin, A108G, and P109A) were expressed by the use of a baculovirus-based expression system in insect cells. They were purified by affinity chromatography on a Strep-tag II column and gel filtration on a Superdex 200 column, according to the method we described previously (8). The actins were quite reproducibly obtained with high purity, as shown in Fig.  2A. However, the average yields of the mutant actins, 1.4 mg for A108G and 0.8 mg for P109A, from 8 ϫ 10 9 cells/3.6 liters of culture, were much lower than that for the WT actin (4.2 mg).
Effects of the A108G and P109A Mutations on the Overall Structure of the Actin Molecule-To determine whether the expressed mutants adopt the canonical structure of the actin molecule, we measured the CD spectra, which reflect the secondary structure content, and the temperature dependence of the CD spectra. As shown in Fig. 2B, the CD spectra of A108G and P109A G-actin were almost identical to that of the WT actin at 25°C (Fig. 2B). The specific minima at 210 nm started to disappear as the sample temperature was gradually raised and completely disappeared at 70°C. To clarify the melting process, we monitored the mean residue ellipticity at 222 nm and determined the melting temperature (T m ) at the inflection point of the melting curve. The T m of P109A was 5°C lower as compared with those of the WT actin and A108G (Table 1). These results indicated that these mutants adopt the canonical structure of the actin molecule at room temperature, although the P109A substitution reduces the thermal stability of the actin molecule. The instability could be due to small defects in the contacts between subdomain 1 and subdomain 3. Moreover, to confirm whether the expressed actin mutants assemble into the canonical F-actin under the polymerizing conditions, we examined the preparation by electron microscopy. The neg- Each profile was obtained after averaging the measurements of at least two independent preparations. The profiles with the A108G mutant and the WT actin were indistinguishable and overlapped each other. C, electron micrographs of negatively stained F-actin. The polymerized A108G and P109A mutants at the early steady state of polymerization were negatively stained with uranyl acetate. A high salt solution, including KCl, MgCl 2 , ATP, and imidazole-HCl, was added to G-actin in G-buffer. Final concentrations were: 12.5 M actin, 100 mM KCl, 2 mM MgCl 2 , 0.5 mM ATP, and 33 mM imidazole-HCl, pH 7.4, plus G-buffer. Electron micrographs were recorded at a magnification of 40,000. The thick particle in the right panel is a tobacco mosaic virus particle, which was included to make the staining homogeneous. Crossover repeats of F-actins are shown in Table 2.
atively stained actin mutants shared the common characteristic F-actin morphology, consisting of two twisted helical strands (Fig. 2C). From the F-actin images, we measured the crossover repeats, which reflect the helical symmetry and pitches. The crossover repeats of the mutant F-actin were 37-38 nm and were identical to those of the WT F-actin ( Table 2). The results indicated that the expressed actin mutants form the canonical F-actin. We concluded that the expressed actin mutants could be utilized to study F-actin formation processes.
Polymerization of the A108G and P109A Mutants-The critical concentration for polymerization (C c ) 3 is a thermodynamic index of F-actin stability at steady state. It is defined by the ratio of the dissociation and association rate constants. The C c of P109A was almost identical to that of the WT actin at 4°C (Table 3). By contrast, the C c of A108G was ϳ10 times higher than that of WT actin, irrespective of the incubation temperature (Table 3). This result indicated that the A108G substitution destabilizes F-actin and favors G-actin.
The time courses of actin polymerization, detected by light scattering, are shown in Fig. 3A. With A108G, the apparent maximum rate of polymerization was 3.2 times slower than that of the WT actin. By contrast, with P109A, the rate was 2.4 times faster than that of the WT actin. The time courses of both mutants showed an initial short lag phase, followed by a rising phase, and a stationary phase, similar to that of WT actin. We analyzed the nucleation of the actin mutants by the conventional nucleation-elongation model (17,18). From the double logarithmic plot of the apparent maximum rate of polymerization versus the actin concentration (13,18), the nucleus sizes for polymerization were estimated to be 3.6 -4.4. These values were identical to that of the WT actin (Fig. 3B). These results suggested that the polymerization of these actin mutants occurs via processes similar to those for the WT actin, although at distinct reaction rates. To estimate the relative rate constants for nucleation and elongation, we measured the numbers of filaments in solution at the steady state. The numbers of filaments were estimated from the initial rates of muscle pyrene G-actin elongating from recombinant F-actin in solution, where polymerization reaches a steady state. The filament numbers at steady state with A108G and P109A were 0.7-and 1.1-fold, respectively, as compared with that of the WT actin. A108G slowly nucleated and elongated, whereas P109A did both rapidly, as shown in Table 4.
Activation Energy for Polymerization of A108G and P109A-The Arrhenius plots of the maximum polymerization rates of these actin mutants are shown in Fig. 3C. The slopes of the plots yield the activation energy for the polymerization reaction. For the WT actin, the activation energy was calculated as 133 kJ/mol (Table 5). This value is close to the value of 100 kJ/mol, reported previously by Kasai (14). For A108G, the activation energy was 163 kJ/mol and was higher than that for the WT actin, which indicates that the kinetic barrier for polymerization is higher (Table 5). In contrast, the plot with P109A is divided into two phases on either side of 15°C; the activation energy was 239 kJ/mol at lower temperatures and 133 kJ/mol at room temperature (Table 5). Because the value at room temperature was identical to that of the WT actin, the rapid polymerization of P109A at room temperature is due to an increase in the frequency factor of the Arrhenius equation.
Actin ATPase Rates of the A108G and P109A Mutants during Polymerization-The time courses of P i product release accompanied by actin polymerization were monitored as the total amount of P i in the assay solution by the use of an EnzChek phosphate assay kit (19) (Fig. 4). The time courses also showed an initial short lag phase followed by a rising phase and a stationary phase, similar to the polymerization time course. In all cases, the total amount of P i released for 45 min after the initiation of polymerization corresponded to the amount of F-actin. Extra amounts of P i were rarely observed. However, in the rising phase, less P i release was obtained from the slowly polymerizable A108G, whereas more was generated from the quickly polymerizable P109A as compared with that from the WT actin. The apparent maximum rate of P i release with A108G was 1.5 times slower than that with the WT actin, whereas the apparent rate with P109A was 1.5 times faster. These differences were much smaller than those observed in the polymerization experiments. Furthermore, we determined a single apparent rate constant for F-actin-ATP 3 F-actin-ADP ϩ P i . The apparent rate constants did not widely differ: 0.045 s Ϫ1 with the WT actin, 0.037 s Ϫ1 with P109A, and 0.054 s Ϫ1 with A108G (supplemental Fig. S1). These results suggested that these substitutions have minimal effects on the actin ATPase activity.

DISCUSSION
The amino acid substitutions of Ala by Gly or of Pro by Ala increase the allowed range for the dihedral angles of the resi- 3 The abbreviation used is: C c , critical concentration for polymerization.     DECEMBER 21, 2012 • VOLUME 287 • NUMBER 52 dues in the polypeptide chain, thus increasing the permissible conformations of the polypeptide chain (20). Actually, in the case of P109A, the main chain of Ala-108 -Ala-109 is deformed, and Ala-109 is separated from His-161 in subdomain 3 despite an almost invariant arrangement with Ile-163, partly because of the lack of an interaction between Pro-109 and His-161 (Protein Data Bank (PDB) codes: 3A5N (21) and 1C0F (22)). Furthermore, the main chain has slightly larger B-factors in the crystal structure of Dictyostelium P109A than that of the wild type actin. Although both A108G and P109A conferred similar perturbations of the Ala-108 -Pro-112 loop, the two substitutions altered the polymerization rates in opposite manners. P109A polymerized more rapidly at room temperature because of increases in both the elongation and the nucleation  Table 5.   rate constants. The acceleration is attributed to a slight increase in the frequency factor of the Arrhenius equation because the activation energy for the polymerization of P109A is identical to that of the WT actin (Fig. 3C). The frequency factor is related to the number of events in which G-actin molecules collide with the ends of F-actin in the proper orientation and conformation necessary to cause polymerization among the thermally fluctuating G-actin molecules; thus, under normal conditions, ϳ2% of the collisions of G-actin molecules with the F-actin end are capable of binding (23). The collision complexes become the newly revised F-actin end by crossing the kinetic barrier of the activation energy. In the case of P109A, the increase in the frequency factor is probably accounted for by the slight accumulation of monomers with a binding-capable conformation. The flattening is a molecular conformational transition mainly in the direction of the propeller-like thermal motion of G-actin. It is possible that similar kinds of motion are enhanced by the weakening of the interaction between the two major domains as suggested by the lower melting temperature. Another explanation is that the conformation with shifts of Ala-108 -Ala-109 in the crystal is close to a binding-capable conformation. Furthermore, if the fraction of monomers with a binding-capable conformation declines with decreasing temperature, then the Arrhenius plot would have a convex curve, as observed for P109A in Fig. 3C (24). Likewise, parallel Arrhenius plots were reported for the effects of salt or the pH of the actin solution on the polymerization rate (14); this is accounted for by the increase in the frequency factor, through the enhancement of the diffusion process or the favorable orientation by the attractive long range forces provided by electrostatic interactions (23,25,26). On the other hand, one explanation for the invariant activation energy with the substitution of P109A is that the energy needed for the separation of Pro-109 and His-161 is negligibly small, as compared with that required for flattening relevant to polymerization. Another explanation is that Pro-109 is almost separated from His-161 before flattening. On the other hand, the A108G mutant slowly polymerized due to decreases in both the elongation and the nucleation rate constants (Fig. 3A). The slowdown is due to an increase in the activation energy (Fig. 3C). The substitution must affect the process where the collision complex of the G-actin molecule and the F-actin end crosses a kinetic barrier, including the conformational transition relevant to polymerization, to create the newly revised end. The extra activation energy was 30 kJ/mol at most (Table 5), which corresponds to about 1.5-fold of the energy of a hydrogen bond. One explanation for the increase in the activation energy of A108G is that the substitution increases the permissible forms of Gly-108 -Pro-112, allowing extra contact at the region relevant to the flattening, for example, between Pro-109/Leu-110 and subdomain 3.

Actin Mutant with Defects in the Ala-108 -Pro-112 Loop
The Ala-108 -Pro-112 loop protrudes from the two secondary structures (␤-strand Thr-103-Glu-107 and ␣-helix Pro-112-Glu-125), and its arrangement is also stabilized by hydrogen-bonding networks among Glu-107, Arg-116, Asn-111, and Asn-115. The conformation of the loop, in which Pro-109 and Leu-110 contact subdomain 3, is probably restricted by the structural rigidity of proline. We hypothesize that the interaction sites in the restricted conformation make the contacts favorable, neither too strong nor too weak, for the attachment and detachment relevant to the flattening; hence, we might observe the opposite effects of the perturbations on polymerization. The similar role of proline has been reported for a floppy loop in arylalkylamine N-acetyltransferase (27).
Generally, polymerization occurs by the manner in which a monomer collides with the filament end, and then the collision complex adjusts its conformation to form the newly revised end. Our results experimentally demonstrated the two separate conformational changes relevant to the actin polymerization process. One is the thermal fluctuation of the conformation without activation, which probably affects the polymerization rate via the effective formation of the collision complex. The other is the conformational change crossing a kinetic barrier by extra energy. The conformational change may represent the main part of the flattening transition.