Report

Rotational Movement of the Formin mDia1 Along the Double Helical Strand of an Actin Filament

See allHide authors and affiliations

Science  07 Jan 2011:
Vol. 331, Issue 6013, pp. 80-83
DOI: 10.1126/science.1197692

Abstract

Formin homology proteins (formins) elongate actin filaments (F-actin) by continuously associating with filament tips, potentially harnessing actin-generated pushing forces. During this processive elongation, formins are predicted to rotate along the axis of the double helical F-actin structure (referred to here as helical rotation), although this has not yet been definitively shown. We demonstrated helical rotation of the formin mDia1 by single-molecule fluorescence polarization (FLP). FLP of labeled F-actin, both elongating and depolymerizing from immobilized mDia1, oscillated with a periodicity corresponding to that of the F-actin long-pitch helix, and this was not altered by actin-bound nucleotides or the actin-binding protein profilin. Thus, helical rotation is an intrinsic property of formins. To harness pushing forces from growing F-actin, formins must be anchored flexibly to cell structures.

Actin polymerization is regulated by factors such as Rho family guanosine triphosphatases (GTPases) and actin nucleators, and can rapidly remodel the cortical cytoskeleton and generate force for cell edge protrusion (1). Formin proteins are the major actin nucleators and function in diverse processes such as cell migration, polarity formation, and cytokinesis (2, 3). Formins share the conserved formin homology 1 and 2 domains (FH1 and FH2, respectively), where FH2 enhances F-actin nucleation. FH2 remains associated with the growing barbed end of F-actin and processively elongates the assembling actin filament (4, 5). FH1 binds the G-actin–binding protein profilin, and this interaction facilitates great enhancement of adenosine 5′-triphosphate (ATP)–actin assembly at the FH2-bound filament tip (6). FH2 forms a ring-shaped dimer that binds around the barbed end of F-actin (7). The subunits in actin filaments form two strands arranged in a right-handed, long-pitch double helix (8). If FH2 tracks processively along the long-pitch helix as the filament elongates, FH2 will rotate on the barbed end of the filament.

Several studies have reported functions of formins that might involve organelle transport, such as regulation of endosome motility by hDia2C (9) and transport of meiotic chromatin by Formin-2 (10). Anchoring growing actin ends to the cell cortex is another possible function of formins. In animal cells, Rho GTPase and its effector, the formin protein mDia1, induce the formation of actin stress fibers (11, 12). The truncated FH1-FH2 mutant of mDia1 processively moves over several tens of micrometers at a speed of 2 μm/s in cells (4). However, among wild-type mDia1 visualized as single molecules, processively moving mDia1 constitutes a minor population (4.3%), and half of mDia1 (47.5%) is stationary (13). Yeast formins for3p and Bni1p are thought to processively assemble actin cables while remaining at the cell tip (14, 15). In vitro, actin polymerization from an immobilized formin produces a pushing force in the piconewton range (16). However, it has not been established that formins can transmit force produced by actin polymerization for the movement of cellular structures. It is thus important to elucidate how the actin polymerization force acts on formins.

We analyzed the rotational movement of F-actin assembled by recombinant FH1-FH2 domains of mDia1 [amino acids 543 to 1192, glutathione S-transferase (GST)–mDia1ΔN3] (17) using fluorescence polarization of single dye molecules on the filament (18). GST-mDia1ΔN3, the presumptive structure of which is a dimer harboring GST-GST and FH2-FH2 interactions, was immobilized on a glass surface within protein aggregates formed by antibodies to GST and secondary antibodies (4). We observed F-actin, which contains tetramethylrhodamine-labeled actin (TMR-actin) at a low density. The emission moment of fluorescence polarization on TMR-F-actin lies at an angle of 45° to the filament axis (18). Thus, the orientation of TMR-actin in a diagonally aligned filament can be monitored by separating the fluorescence into vertically (FLV) and horizontally (FLH) polarized components (Fig. 1).

Fig. 1

Detection of the polarized fluorescence from single-molecule tetramethylrhodamine (TMR) on growing actin filaments. F-actin was nucleated by mDia1 and TMR-G-actin (1.5%). TMR emits polarized fluorescence at an angle of 45° to the filament axis (18). When the filament lies diagonally, the orientation of TMR-actin subunits can be monitored by separating the fluorescence into vertically (FLV) and horizontally (FLH) polarized components with a polarizing beam splitter.

Under normal actin elongation conditions with ATP–monomeric actin (G-actin; 0.5 μM), the intensities of FLV and FLH alternated periodically (Fig. 2, A and C, and movie S1). The fluorescent spot moved directionally (Fig. 2B), and the fluorescence polarization [FLP = (FLV − FLH)/(FLV + FLH)] of the spot alternated over a long distance (Fig. 2D). The periodic alternation in FLP was not observed when processive actin elongation was arrested (fig. S1). Analysis of the spot displacement yielded 36.1 ± 0.33 nm (mean ± SEM) per inversion between FLV and FLH, which corresponds to the half-pitch length of the F-actin long-pitch helix (Fig. 2E). These results indicate that mDia1 undergoes helical rotation during processive filament elongation.

Fig. 2

Helical rotation of ATP-F-actin growing from immobilized mDia1. F-actin was nucleated by incubating 1 μM TMR-ATP-G-actin (1.5%) with protein aggregates containing GST-mDia1FH1-FH2 (mDia1ΔN3) adsorbed on a glass surface. After 1 min, free TMR-G-actin was washed out, and the elongation of F-actin was observed in the presence of 0.5 μM unlabeled ATP-G-actin. (A) Merged images of FLV (red) and FLH (green) from TMR-actin at 2-s intervals. The alternation between the intensity of FLV and FLH of the spot (arrowheads) is apparent. (B and C) Displacement (B) and time-lapse images of FLV and FLH at 0.5-s intervals (C) of the spot in (A). (D) Fluorescence polarization (FLP) (blue) and distance from mDia1ΔN3 aggregates (pink) of the spot in (A). (E) Distance per half-rotation of ATP-F-actin (mean ± SEM; 19 independent experiments, n = 28 spots, total 428 alternations). Scale bars (A and C), 1 μm.

Previous studies concluded that formins slip around the filament like a bearing when both the formin and the distal end of the filament are anchored to the slide (16, 19). Kovar and Pollard (16) observed that F-actin elongating from Bni1p(FH1-FH2) nonspecifically adsorbed on glass did not form a supercoil when the pointed end of the filament was fixed. However, they left it uncertain whether slippage occurs between FH2 and the glass surface (3, 6). We hypothesized that when helical rotation is faithfully coupled to elongation, the buckling frequency should be lower when FH2 is immobilized tightly than loosely. We therefore compared the F-actin buckling frequency obtained by the previous immobilization method for FH2 (16) with that obtained by our method. When a portion of F-actin was trapped through the interaction between biotinylated actin and streptavidin-coated glass, the buckling frequency with mDia1-antibody aggregates was smaller than that with mDia1 nonspecifically adsorbed to the glass surface (fig. S2). During buckled actin elongation, we did not observe periodic alternation of FLP (fig. S3 and movie S2). We also verified helical rotation of F-actin elongating from Bni1p(FH1-FH2) (fig. S4 and movie S3). The different buckling frequencies resulting from different immobilization methods of FH2 indicate that torsional stress can be relaxed through slippage between FH2 and the glass surface in both the previous experiments by Kovar and Pollard (16) and the present study. We believe that formins undergo helical rotation faithfully along the long-pitch helix, although our data do not exclude the possibility that FH2 might rotate like a bearing under strong torsional stress.

Profilin promotes ATP-actin elongation at the FH1-FH2 bound barbed end, which could affect the actin polymerization force. We therefore examined whether helical rotation is retained in the presence of profilin. Consistent with a previous report (6), the addition of 6 μM profilin accelerated mDia1-mediated actin elongation in the presence of 1 μM ATP-G-actin from 7.8 to 48 subunits s−1. To reduce the elongation rate for FLP detection, we needed to use 15 μM profilin and 0.5 μM actin. Under this condition, actin elongation is suppressed because free profilin competes with the profilin-actin complex for binding to FH1 (6). We detected a periodic alternation of FLP on processively elongating F-actin in the presence of profilin (Fig. 3A and movie S4). The distance per half-rotation was 34.6 ± 0.77 nm (mean ± SEM) (Fig. 3B). Thus, the interaction between FH1 and profilin does not interfere with the helical rotation of FH2.

Fig. 3

The effects of ADP and profilin. (A) Helical rotation of ATP-F-actin in the buffer containing 2 mM ATP, 0.5 μM ATP-G-actin, and 15 μM profilin. (B) Distance per half-rotation of ATP-F-actin elongating from mDia1 in the presence of 15 μM profilin (mean ± SEM; four independent experiments, n = 7 spots, total 64 alternations). (C) The elongation rate of ADP-F-actin by mDia1ΔN3 in buffer containing 2 mM ADP and 5 μM ADP-G-actin. Each column shows the elongation rate of ADP-G-actin alone (black), ADP-G-actin with 20 mM Pi (white), ADP-G-actin with 6 μM profilin (green), and ADP-G-actin with 6 μM profilin and 20 mM Pi (blue). Error bars indicate SD (n = 20 to 25 filaments for each condition). (D) Helical rotation of ADP-actin elongating from mDia1ΔN3 with 5 μM ADP-G-actin alone. (E) Distance per half-rotation (mean ± SEM) of ADP-actin elongating from mDia1 (four independent experiments, n = 9 spots, total 116 alternations).

We next tested whether ADP-G-actin supports helical rotation. Although profilin accelerates elongation of ATP-actin, its effect on ADP-actin remains controversial. Repeatedly, mDia1-mediated elongation of ADP-actin has been shown to be slow with ADP-G-actin alone (~36% of that without mDia1) (6) and almost negligible in the presence of profilin (20). Because prolonged depletion of ATP during preparation of ADP-G-actin may deteriorate the conformation of assembled F-actin (21), we carefully prepared ADP-G-actin. In addition, ATP hydrolysis may be involved in the aging process of F-actin (22). Recent studies (23) reported that transiently (~2 min) after polymerization, actin forms filaments, which structurally deviated from the Holmes model (24). Taking advantage of our system, we also investigated the effect of nucleotides on the helical twist of assembling actin filaments.

ADP-G-actin alone processively elongated mDia1-nucleated F-actin (Fig. 3C). The addition of profilin decreased the elongation rate by 53% (Fig. 3C). These data differ from those reported in a previous study (6), in which profilin accelerated the elongation of ADP-actin at the mDia1FH1-FH2–bound barbed end. This discrepancy arises from the difference in the elongation rate without profilin. Our data do not favor the view that profilin accelerates mDia1-associated actin elongation regardless of the actin-bound nucleotides (6) (see below).

F-actin rotated periodically during elongation from immobilized mDia1 with ADP-G-actin (Fig. 3D and movie S5). Thus, helical rotation proceeds without ATP hydrolysis. We did not detect any change in the distance per half-rotation with ADP-actin (Fig. 3E) and ATP-actin (Fig. 2E).

Our data also revealed that profilin lowers the ADP-actin elongation rate not by capping, as previously concluded (20), but by enhancing the off-rate of actin at the barbed end. Direct observations show that profilin promotes processive depolymerization of the formin-bound barbed end (Fig. 4A). Profilin enhanced the processive depolymerization rate of the mDia1-bound barbed end from 1.4 to 11.6 s−1. This enhancement was observed at a lower concentration of profilin than that required for the promotion of depolymerization of the free barbed end (25), which was also found in a previous observation of bulk filament disassembly (20). FH1 and profilin thus accelerate both ATP-actin elongation and ADP-actin depolymerization. The ADP-F-actin rotates as it shortens in the presence of 20 μM profilin (Fig. 4B and movie S6). The distance between the spot and the barbed end decreased by 36.8 nm per half-rotation (Fig. 4C).

Fig. 4

Effects of profilin and inorganic phosphate (Pi) on elongation, depolymerization, and helical rotation of mDia1ΔN3. (A) The dose-dependent effect of profilin on depolymerization of ADP-F-actin bound to mDia1ΔN3. (B) Helical rotation of ADP-F-actin during profilin-accelerated depolymerization at the mDia1ΔN3-bound barbed end. Depolymerization of ADP-F-actin was induced by perfusion of a buffer containing 2 mM ADP and 20 μM profilin. (C) Distance per half-rotation (mean ± SEM) of ADP-F-actin bound to mDia1 during depolymerization with 20 μM profilin (three independent experiments, n = 8 spots, total 72 alternations). (D) The dose-dependent effect of Pi on profilin (5 μM)–enhanced depolymerization of ADP-F-actin bound to mDia1ΔN3 at pH 7.0. (E) The dose-dependent effect of Pi on mDia1ΔN3-mediated elongation of ADP-actin (5 μM) in the presence of 5 μM profilin. (F) Helical rotation of ADP-F-actin elongating from mDia1ΔN3 in the presence of 5 mM Pi. (G) Distance per half-rotation (mean ± SEM) of ADP-F-actin elongating from mDia1 in the presence of 5 mM Pi (three independent experiments, n = 6 spots, total 88 alternations).

Inorganic phosphate (Pi) decreases the off-rate of ADP-F-actin at the free barbed end (26). Pi also blocks the binding of F-actin and ADF/cofilin (27), which increases the filament twist (28). We therefore examined the effect of Pi on the processive elongation and twist of mDia1-nucleated ADP-actin. We found that Pi abolished the profilin-enhanced actin depolymerization at the mDia1-bound barbed end. This inhibition occurs in the submillimolar range of Pi, which is two orders of magnitude lower than the dissociation constant of Pi and G-actin (26). Thus, binding of Pi to F-actin inhibits profilin-induced depolymerization (Fig. 4E).

ADP-G-actin (5 μM) elongated mDia1-bound F-actin faster in the presence of 20 mM Pi than in its absence (Fig. 3C). This effect of Pi was more prominent in the presence of profilin than in its absence (Fig. 3C). The decrease in the actin off-rate (Fig. 4D) corresponds well with the increase in the ADP-actin elongation rate by 3 to 20 mM Pi (Fig. 4E). We thus suggest that Pi cancels the inhibitory effect of profilin on ADP-actin elongation (Fig. 3C) by abolishing the enhanced barbed end off-rate. The discrepancy of the effect of 1 mM Pi on depolymerization and assembly (Fig. 4, D, and E) is because slow dissociation of Pi prebound to a fraction of the ADP-F-actin subunits [dissociation constant (Kd) ≈1.5 mM] may limit terminal subunit dissociation during depolymerization (26), but not dissociation of assembling ADP-actin, which is mostly free from Pi. Profilin thus allows processive elongation of the FH1-FH2–bound barbed end regardless of the actin-bound nucleotide, but attenuates ADP-actin elongation by increasing the barbed end off-rate. Our results urge reconsideration of the ATP-specific acceleration mechanism for formin-associated actin elongation.

Helical rotation of mDia1 was observed during processive ADP-actin elongation in the presence of Pi (Fig. 4F and movie S7). The distance per half-rotation was 35.8 nm (Fig. 4G).

Our data demonstrate continuous rotation of mDia1-bound filaments during both elongation and depolymerization. The distance per half-rotation of F-actin is in the range of 34.6 to 36.8 nm regardless of the actin-bound nucleotide and presence of Pi and profilin (Figs. 2 to 4). These findings indicate that helical rotation of FH2 is an intrinsic property derived from the helical structure of F-actin. Cellular actin filaments are highly cross-linked as evidenced by single-molecule observations showing movement of actin subunits with no change in their relative positions (29). Therefore, formins must rotate in the cell. The rotation speed of for3p at the cell tip and processively moving mDia1 can reach 250 and 1700 rpm, respectively. If anchoring the growing end of F-actin is the function of formins, the link between formins and cellular structures must be flexible. Alternatively, formin-mediated actin elongation may be regulated by torsional stress in F-actin.

Conversely, formins might modulate the stability of F-actin by helical rotation. Torsional stress induces destabilization of the filament (30). Cofilin, the major actin depolymerizing factor, twists the strand of F-actin, which is thought to contribute to actin disassembly (28). Our data have opened up the possibility that actin elongation and remodeling could be regulated by axial torsion in the filament. Our findings should help elucidate the actin turnover mechanism regulated by formins in the cell.

Supporting Online Material

www.sciencemag.org/cgi/content/full/science.1197692/DC1

Materials and Methods

Figs. S1 to S4

References

Movies S1 to S9

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank H. Honda for rabbit muscle acetone powder and T. M. Watanabe for customizing the G-track software. This work was supported by Grants-in-Aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT) and grants from the Uehara Memorial Foundation (N.W.) and the Human Frontier Science Program (N.W.).
View Abstract

Navigate This Article