Actin Polymerization-Driven Molecular Movement of mDia1 in Living Cells

See allHide authors and affiliations

Science  26 Mar 2004:
Vol. 303, Issue 5666, pp. 2007-2010
DOI: 10.1126/science.1093923


mDia1, a Rho effector, belongs to the Formin family of proteins, which shares the conserved tandem FH1-FH2 unit structure. Formins including mDia1 accelerate actin nucleation while interacting with actin filament fast-growing ends. Here our single-molecule imaging revealed fast directional movement of mDia1 FH1-FH2 for tens of microns in living cells. The movement of mDia1 FH1-FH2 was blocked by actin-perturbing drugs, and the speed of mDia1 FH1-FH2 movement appeared to correlate with actin elongation rates. In vitro, mDia1 FH1-FH2 associated persistently with the growing actin barbed end. mDia1 probably moves processively along the growing end of actin filaments in cells, and Formins may be a molecular motility machinery that is independent from motor proteins.

Formin homology proteins (Formins) play essential roles in the regulation of specific actin-based structures (1, 2), e.g., the contractile ring in cytokinesis (3, 4); actin cables in budding yeasts (57); and actin fibers in animal cells (8). The structure of Formins is conserved in most of the C-terminal half including the FH1 and FH2 domains. The FH1 domain consists of poly-proline repeats that bind to profilin, and it is followed by conserved sequences of ∼400 amino acids including the FH2 domain. This structural feature unique to Formins, called the FH1-FH2 unit (FH1-FH2) (8), implies a certain evolutionarily conserved function of FH1-FH2.

Yeast Formins (912) and mDia1 (13) have an actin nucleation-promoting activity in their FH1-FH2 or FH2 fragments. Nucleated actin filaments elongate at the fast-growing, barbed end. Formins have thus emerged as potential cellular actin nucleators. However, in marked contrast to the Arp2/3 complex, another actin nucleator, Formins bind the barbed end of actin filaments. Two models for how Formins allow barbed end elongation while binding the same site have been proposed. One is the profilingated mechanism for the barbed end capping by Schizosaccharomyces pombe Cdc12p (12). Alternatively, Bni1p and mDia1 function as “a leaky cap,” and it has been postulated that Bni1p might “walk” along the barbed end of an elongating filament (11, 14).

To explore the biochemical properties of Formins in living cells, we took an approach similar to single-molecule actin speckle analysis (15). We expressed mDia1 mutants fused to an enhanced green fluorescent protein (EGFP) (Fig. 1A) in XTC fibroblasts and carried out live-cell imaging of mDia1 expressed at a low level (16). mDia1△N3, an FH1-FH2 mutant, induced the formation of massive thin actin fibers, reminiscent of Rho-induced actin stress fibers (Fig. 1B) (8). Speckle microscopy revealed fast, directional molecular movement of EGFP-mDia1△N3 (Fig. 1C, Movie S1). Fluorescence intensity of moving signals was comparable to that of single-molecule EGFP-actin (15), which indicated that they consisted of single EGFP or a few EGFP molecules coupled to multimeric mDia1 (13). mDia1△N3 movement persisted along a long path that often spanned the radius of the cell (from the center to the edge). mDia1△N3 speckles could be tracked along a line or a wide arc, which suggests an association with certain cytoskeletal filaments. At the cell periphery, the majority of mDia1△N3 speckles moved outward along filopodium-like protrusions, accumulating in the tip region (Fig. 1B, inset; Movie S1). We also observed movement of mDia1F2 speckles. mDia1F2 is a mutant that is defective in promoting actin fiber formation (Fig. 1B) (8), and it weakly activates serum response factor (17). mDia1F2 speckles displayed slower rates of movement and traveled over shorter distances (Fig. 1D, Movie S2). The average speed of mDia1△N3 and mDia1F2 movement was 2.0 μm/s and 0.13 μm/s, respectively (Fig. 1E). The speed of mDia1△N3 movement is two orders of magnitude faster than the flow rate of the actin network (typically ∼0.025 μm/s in lamellipodia of XTC cells) (15).

Fig. 1.

Fast movement of EGFP-mDia1 mutants as revealed by single-molecule observations in living XTC cells. (A) Structures of mDia1 mutants. (B) Massive actin fibers observed in cells expressing EGFP-mDia1△N3 (top), but not in cells expressing EGFP-mDia1F2 (bottom). Actin filaments (red) and EGFP signals (green) are shown. Scale bar, 10 μm. Note the faint accumulation of EGFP-mDia1△N3 at the tip of filopodium-like processes (inset, scale bar, 2 μm). (C and D) Representative time-lapse images of mDia1△N3 (C) and mDia1F2 (D) speckles. Images of EGFP-mDia1 mutants expressed at a very low density were acquired. Each colored square indicates the same speckle followed over time. Scale bars, 5 μm. (E) Distribution of the speed of mDia1△N3 (left) and mDia1F2 (right) speckles. The speeds of mDia1△N3 and F2 movement (means ± SD) were 2.0 ± 0.30 μm/s, n = 67, three cells, and 0.13 ± 0.10 μm/s, n = 67, four cells, respectively. (F) Fast molecular movement of mDia1Full induced by RhoA-V14 microinjection. Time-lapse images of EGFP-mDia1Full were taken before (left) and 2 min after (right) RhoA-V14 microinjection. Arrowheads indicate mDia1Full, which emerged as a fast-moving speckle. Scale bar, 5 μm.

Unlike mDia1△N3 and mDia1F2, an EGFP-fusion of full-length mDia1, mDia1Full hardly exhibited any fast directional movement (Fig. 1F, left; Movie S3, top). We then tested whether administration of Rho affects the mobility of mDia1Full, as Rho binding to the mDia1 N terminus has been postulated to disrupt the autoinhibitory intramolecular interaction of mDia1, by exposing FH1-FH2 and thus permitting its activity on actin (8, 13, 18). As early as 2 min after microinjection of recombinant RhoA-V14 (a constitutively active mutant), we frequently observed directional movement of mDia1Full similar to that of mDia1△N3 (Fig. 1F, right; Movie S3, bottom). The average speed of mDia1Full movement was 0.97 μm/s after microinjection. These results support a role for Rho binding in the molecular activation of mDia1 (8) within cells and demonstrate spatial and temporal molecular dynamics mediating a signal between Rho and actin reorganization (19).

We next examined which cytoskeletal filaments are required for mDia1 movement. Treatment of cells with 10 μM nocodazole depolymerized most microtubules, whereas mDia1△N3 movement was unchanged even in the area devoid of microtubules (Fig. 2A, Movie S4). In contrast, three actin-perturbing drugs, cytochalasin D (CD), latrunculin B (LatB), and jasplakinolide (Jas), decelerated and finally stopped mDia1△N3 movement completely (Fig. 2B, Movie S5, Fig. 3). Thus, the fast movement of mDia1 depended on the integrity or dynamics of actin filaments, but not of microtubules.

Fig. 2.

mDia1 FH1-FH2 movement is dependent on actin, but not on microtubules. (A) Time-lapse images of EGFP-mDia1△N3 were taken before and 60 min after cells were treated with 10 μM nocodazole. Cells were fixed 61 min after the treatment and stained with antibody against β-tubulin (right) and fluorescent phalloidin (middle right). Arrows indicate the tracks of representative mDia1△N3 speckles. Speckles continued to move in an area devoid of microtubules after the treatment. The average speed of mDia1△N3 movement before and 60 min after the treatment was 1.1 μm/s, n = 15, and 1.0 μm/s, n = 15, respectively. (B) Arrows indicate the track of representative mDia1△N3 speckles (left). Speckle movement was not observed 14 min after 1 μM Jas treatment (right). Scale bars, 5 μm.

Fig. 3.

Two types of mDia1△N3 speckle deceleration after treatment with three actin-perturbing drugs. (A) Cells expressing EGFP-mDia1△N3 were treated with 1 μM LatB. Each plot represents the speed of individual speckles measured at the intervals of 0.4 s. Speed of speckles decreased gradually and synchronously. Arrow shows the beginning of perfusion. (B) Cells expressing EGFP-mDia1△N3 were treated with 1 μM Jas. Due to the slow decrease in the speckle speed, measurement was carried out intermittently. Triangles indicate the speed of individual speckles, and bars indicate the average speed of analyzed speckles at indicated time. Speed of speckles decreased gradually. (C) Cells expressing EGFP-mDia1△N3 were treated with 1 μM CD. Speckle speed was measured at the intervals of 0.4 s. Each plot represents the speed of individual speckles. Speckles stopped suddenly within 2 consecutive images.

We next found evidence of an actin polymerization–driven mechanism that did not involve myosin facilitating the fast mDia1 movement. The three actin-perturbing drugs, CD, LatB, and Jas, attenuated EGFP-mDia1△N3 movement, but careful measurement revealed two types of change in the speed of individual mDia1△N3 speckles. A gradual deceleration was observed in cells treated with LatB (Fig. 3A, Movie S6) or with Jas (Fig. 3B, Movie S7), whereas sudden stoppage, which occurred within two consecutive images, was observed in CD-treated cells (Fig. 3C, Movie S8). The speckle stoppage occurred at various times and places in cells after CD treatment, and before arrest, speckles continued to move as fast as they did before drug administration. This clearly differed from the synchronous, gradual deceleration of all speckle movement after LatB or Jas treatment.

These changes in the speed of mDia1△N3 speckles correspond well to the predictable actin elongation rate change. Latrunculins bind actin monomers and decrease polymerization-competent monomers (20). For the other drugs, actin filament turnover, the t1/2 of which is ∼30 s (15), must be taken into consideration. Jas inhibits actin depolymerization (21), which would result in a gradual loss of monomers. CD binds the barbed end and inhibits filament elongation directly (22). Because the actin elongation rate is nearly proportional to the sum of the free monomer and profilin-actin concentrations (23), LatB or Jas treatment would decrease the elongation rate, whereas CD treatment would not decrease the elongation rate until it blocked barbed end growth. All of the speckle speed data (Fig. 3) fit with this estimated elongation rate change. Synchronous speckle deceleration in LatB- or Jas-treated cells might reflect gradual retardation of actin elongation as a consequence of actin monomer depletion. In CD-treated cells, mDia1△N3 speckles stopped suddenly, presumably when CD bound the barbed end, which occurred stochastically at different filaments. CD treatment did not abolish the association of mDia1△N3 with actin structures. Thus, the speed of mDia1△N3 movement is defined by the actin elongation rate in cells. Myosin involvement is unlikely, because actin filaments remained for the observed duration; also, the flow and contraction of the actin network, perhaps myosin-driven events, were not blocked by CD, LatB, or Jas. Taken together with the proposed biochemical property of a leaky cap of FH1-FH2 (11, 13, 14), our data suggest that mDia1 FH1-FH2 processively associates with the growing barbed end of actin filaments in cells.

In order to obtain direct evidence that mDia1 can move by a mechanism other than myosin, we reconstituted mDia1-catalyzed actin assembly using purified components and observed filament elongation relative to mDia1 FH1-FH2 under the microscope (16). We labeled recombinant glutathione S-transferase mDia1ΔN3 fused to (GST-mDia1△N3), using antibodies, which resulted in the formation of large protein aggregates containing GST-mDia1△N3. After actin and profilin were mixed with the labeling mixture, we observed marked outgrowth of actin filaments emanating from mDia1△N3-containing aggregates (Fig. 4, A to D; Movie S9). Speckle labeling of the filaments revealed the filament growth occurring at the aggregates but not at the distant end of the filaments (Fig. 4, E and F; Movies S10 and S11). We also observed growth of a single isolated filament from small mDia1△N3-containing aggregates (Fig. 4, G and H; Movie S12), which demonstrated continuous anchoring of the growing filament end to mDia1△N3. Filament growth at the rate of 0.040 μm/s (Fig. 4F) can occur only at the barbed end under these conditions. Thus, mDia1△N3 is capable of persistently binding the growing barbed end. This finding provides further evidence for myosin-independent movement of mDia1 FH1-FH2.

Fig. 4.

Actin filament growth emanating from recombinant mDia1△N3. GST-mDia1△N3 was labeled using a GST-specific goat antibodyand an Alexa594antibody directed against goat IgG, and then mixed with 1.9 μM actin and 1.3 μM profilin in buffer [20 mM imidazole (pH 7.0), 2 mM Tris HCl, 0.2 mM ATP, 2.1 mM MgCl2, 2 mM EGTA, 100 mM KCl, 10.5 mM DT T, 0.1 mg/ml glucose oxidase, 3 mg/ml glucose, 20 μg/ml catalase, 0.5% methylcellulose, 83 nM Oregon Green-phalloidin]. (A to D) Time-lapse images of actin filament outgrowth (A to C). Merged image of F-actin (green) and GST-mDia1△N3 (red) (D). Time relative to (A) is shown. (E and F) Speckle labeling of the filaments revealing fast filament elongation at GST-mDia1△N3. Merged image of GST-mDia1△N3 (red) and F-actin (green). A part of the phalloidin label image (square) is enlarged [(E), inset, scale bar, 2 μm]. Actin filaments moved initially at ∼0.040 μm/s. Conditions were the same except for 4.2 nM Oregon Green-phalloidin. Scale bar, 10 μm. Kymograph (F) shows cross-sectional view of the intensityvalues along the dashed line. Distance is from arrow in (E). (G and H) Persistent growth of a single actin filament anchored to mDia1△N3. (G) F-actin (green) and GST-mDia1△N3 (red). Scale bar, 5 μm. (H) Time-lapse images of phalloidin-staining (dashed square) are paneled. Time is in minutes. Scale bar, 2 μm.

From our finding of long-distance mDia1 FH1-FH2 movement clinging to the growing barbed end of actin filaments, we may infer several possible roles of Formins. Their role may be to protect barbed ends from capping proteins (14). mDia1 FH1-FH2 frequently moved over a distance of >10 μm, although barbed end capping by capping proteins is estimated to occur within less than 1 s (24, 25). In cooperation with profilin (26), Formins may play a role in generating long actin filaments, which may be a prerequisite for assembling actin stress fibers, cables, and the contractile ring.

Another possible role may be to accelerate barbed end growth. The speed of mDia1△N3 in cells, 2.0 μm/s, corresponds to the actin elongation rate by ∼70 μM profilin-actin complex. This value seems slightly higher than a cellular profilin concentration, ∼30 μM (27). However, any acceleration of barbed end growth has not been detected so far (13).

The third possible role is to anchor the barbed end while allowing actin monomer exchange. In budding yeasts, live imaging of GFP-Abp140 detected fast extension of actin cables away from the bud (28), implying pointed end–directed actin translocation supported by certain barbed end factors. In animal cells, actin filaments do not freely move because of cross-linking, as proven by the constant relative position between actin speckles (15). Therefore, the anchoring mechanism, if it applies, might locally hold a cluster of growing filaments such as Rho-dependent actin fibers found at the Shigella entry into epithelial cells (29).

Finally, FH1-FH2 has the potential to act in cells as a motor driven by actin polymerization. The directional movement of mDia1 over tens of microns is a property known only for motor proteins. Actin polymerization has been regarded as a “motor” because it converts chemical energy to mechanical energy to drive cell edge protrusion (30). mDia1 can elicit such actin polymerization–driven motion at the molecular level, which raises the possibility that Formins comprise a molecular motility machinery fueled by actin polymerization.

Supporting Online Material

Materials and Methods


Movies S1 to S12

References and Notes

View Abstract

Navigate This Article