Actin disassembly by cofilin, coronin, and Aip1 occurs in bursts and is inhibited by barbed-end cappers

Abstract

Turnover of actin filaments in cells requires rapid actin disassembly in a cytoplasmic environment that thermodynamically favors assembly because of high concentrations of polymerizable monomers. We here image the disassembly of single actin filaments by cofilin, coronin, and actin-interacting protein 1, a purified protein system that reconstitutes rapid, monomer-insensitive disassembly (Brieher, W.M., H.Y. Kueh, B.A. Ballif, and T.J. Mitchison. 2006. J. Cell Biol. 175:315-324). In this three-component system, filaments disassemble in abrupt bursts that initiate preferentially, but not exclusively, from both filament ends. Bursting disassembly generates unstable reaction intermediates with lowered affinity for CapZ at barbed ends. CapZ and cytochalasin D (CytoD), a barbed-end capping drug, strongly inhibit bursting disassembly. CytoD also inhibits actin disassembly in mammalian cells, whereas latrunculin B, a monomer sequestering drug, does not. We propose that bursts of disassembly arise from cooperative separation of the two filament strands near an end. The differential effects of drugs in cells argue for physiological relevance of this new disassembly pathway and potentially explain discordant results previously found with these drugs.

Figure 1.

Actin filaments disassemble in bursts in cofilin, coronin, and Aip1. (A) Time-lapse wide-field epifluorescence images of fluorescently labeled actin filaments in the presence of 2 μM cofilin, 1 μM coronin, 200 nM Aip1, 5 μM of actin monomer, and 2 mM ATP. Filaments shorten and disappear from the field of view. Bar, 3 μm. (B) Successive time-lapse images showing a single actin filament (f1) over time, along with kymographs drawn along the contours of representative filaments (f1–f4). The red lines on the image of f1 at t = 0 denote the contour on which the kymograph was drawn. Time is given on the x axis of the kymograph, whereas the position along the filament contour is given on the y axis. Mean integration time for a single image was 400 ms for f1–f3 and 16 ms for f4. Triangles denote endwise bursting (f1–f3); yellow triangles denote initial burst (f1–f3), red triangles denote successive proximal bursts (f1 and f2), and green triangle denotes a successive distal burst (f3). Same-side bursts occurred more frequently (78%) than opposite-side bursts (22%; P < 0.001, one-tailed z test). The square denotes internal disassembly event counted as a severing event (f3). Bar, 1 μm. (C) Histogram of filament burst size. The mean burst size was 260 subunits. (D) Histogram of waiting times between successive bursts (red), fit to a single exponential (black). Single exponential fit gave characteristic decay time of τ = 14 s.

Figure 2.

Filament bursting is distinct from cofilin-mediated severing. (A) Time-lapse images of two actin filaments disassembling in 10 μM cofilin, 5 μM of actin monomer, and 2 mM ATP. Long filaments were chosen to illustrate the occurrence of multiple severing events within a single filament. Bar, 1 μm. (B) Bar graph showing the fraction of disassembly events scored as bursting (red) or severing (green), either in the full depolymerizing system (left) or in cofilin alone (right). In the full system, bursting occurred with significantly higher frequency than severing (χ² = 190, df = 1, P < 0.01). However, in the presence of high concentrations of cofilin, severing occurred with significantly higher frequency than bursting (χ² = 55, df = 1, P < 0.01) and was the predominant disassembly mechanism observed under these conditions. We note that mean initial filament length did not differ significantly between different experiments and was not the cause of the differences observed.

Figure 3.

Actin filaments disassemble with similar kinetics from both ends. (A) Time-lapse images of actin filament bundles grown off fragments of L. polyphemus acrosomal processes. Filaments in the long bundle have exposed barbed ends (b), whereas filaments in the short bundle have exposed pointed ends (p). The two bundles shown were elongated from opposite ends of the same L. polyphemus acrosomal fragment. The brightness of the shorter bundle was increased relative to the longer one for ease of visualization. At 0 s, filament bundles were perfused with 2 μM cofilin, 1 μM coronin, 300 nM Aip1, 5 μM of actin monomer, and 2 mM ATP. Bar, 1 μm. (B) Total actin polymer mass in the filament bundles as a function of time measured for all bundles (red), bundles with exposed barbed ends (green), and bundles with exposed pointed ends (blue). To compare the rates of rapid disassembly, the slowly varying component of each decay curve was removed (Fig. S1, available at http://www.jcb.org/cgi/content/full/jcb.200801027/DC1; see Materials and methods). (C) Bar graph comparing decay rates for bundles with exposed barbed ends (b) with those with exposed pointed ends (p). Data represent mean and standard deviation of three independent experiments.

Figure 4.

Bursting disassembly generates barbed ends not recognized by CapZ. Fluorescence images show Alexa 647 actin (left) and Alexa 488 CapZ (right) in filament bundles. Barbed ends are oriented toward to the right. The bar graph shows the CapZ/actin filament ratio, a measure of the fraction of capped barbed ends in the bundle. Data represent mean and standard deviation of eight filament bundles. (A) Filament bundles polymerized from L. polyphemus acrosomal fragments were incubated directly with Alexa 488 CapZ. (B) Filament bundles were partially disassembled with 8 μM cofilin for 90 s then incubated with CapZ. (C) Filament bundles were partially disassembled with 2.5 μM cofilin, 1.5 μM coronin, and 50 nM Aip1 for 25 s, then incubated with CapZ. (D) Filament bundles were partially disassembled with cofilin, coronin, and Aip1 as in C, incubated in buffer for 10 s, then incubated with CapZ. Bar, 1 μm.

Figure 5.

Barbed end–capping factors inhibit actin disassembly by cofilin, coronin, and Aip1. (A) Mean filament length over time under the following conditions: 2 μM cofilin, 1 μM coronin, 5 μM of actin monomer, and 2 mM ATP, as well as 200 nM Aip1 (red), no Aip1 (green), 5 μM CapZ (yellow), 200 nM Aip1 + 1 μM CytoD (black), or 200 nM Aip1 + 5 μM CapZ (blue). Fast monomer-insensitive disassembly required Aip1 (red vs. green, yellow). CytoD and CapZ inhibited disassembly in the full system (black, blue vs. red). (B) Bar graph showing the percentage of filaments that underwent endwise bursting or severing. The number of filaments analyzed is given below the bars. The percentages of filaments that either underwent endwise bursting or severing increased significantly in the presence of Aip1 (red vs. green, yellow) and decreased significantly when the barbed-end cappers CytoD and CapZ were added to the full system (black, blue vs. red; severing frequency of −CapZ vs. +CapZ, χ² = 4.9, df = 1, P < 0.05; all other pairwise comparisons, χ² > 40, df = 1, P < 0.001). (C) Polymer mass decay for bundles with either exposed barbed ends or pointed ends in 3 μM CytoD (brown and purple) or 10 μM CapZ (blue and green), as well as polymer mass decay for all filament bundles in the absence of barbed-end cappers (red). Disassembly of bundles with either exposed barbed ends or pointed ends were inhibited by CytoD/CapZ with equal efficacy. (D) Fraction of polymer mass disassembled as a function of CytoD concentration (red circles). Best fit of the data to a hyperbola (red curve) shows an IC₅₀ of 90 nM for inhibition of disassembly. Length of filament bundles polymerized for a fixed period of time in the presence of varying concentrations of CytoD is also shown (blue squares). Hyperbolic best fit (blue curve) shows an IC₅₀ of 30 nM for inhibition of polymerization. (E) Fraction of polymer disassembled as a function of CapZ concentration (red squares). The titration curve did not reach saturation; the red curve denotes best fit of the data to a straight line. The length of filament actin bundle polymerized in the presence of varying concentrations of CapZ is also shown (blue circles). Polymerization conditions were identical to those in D. Hyperbolic best fit (blue curve) gave an IC₅₀ of 30 nM for inhibition of polymerization. Error bars indicate SD. (F) CytoD does not inhibit cofilin-mediated severing. The bar graph shows the percentage of filaments that severed during a 400 s video either in the absence of CytoD (left) or in the presence of 1 μM CytoD (right). Conditions: 8 μM cofilin and 5 μM of actin monomer in assay buffer.

Figure 6.

Barbed end–capping drugs inhibit actin disassembly in mammalian tissue-culture cells. (A) Images of suspension HeLa S3 cells showing F-actin stained with TRITC-phalloidin (top) and DNA stained with Hoechst (bottom). Images show untreated cells (left), cells treated with 10 μM CytoD for 480 s (middle), and cells treated with 10 μM LatB for 480 s (right). All images used identical acquisition settings and contrast levels. Bar, 10 μm. (B) The level of F-actin per cell at different times after treatment with either CytoD (blue) or LatB (red). F-actin levels remained high after CytoD treatment. In contrast, F-actin levels fell rapidly upon LatB treatment. Each point represents the mean and SD of at least 10 fields of cells. (C) Images of actin comet tails in L. monocytogenes–infected BSC-1 cells taken at successive times after drug treatment. In the absence of any drugs (control), comet tails disassembled within 50 s. Upon LatB treatment, comet tails also disassembled within 50 s, which suggests that there is no effect of LatB on comet tail disassembly kinetics. In contrast, comet tails did not disassemble in CytoD- and KabC-treated cells for at least 50 s after drug addition, which implies that CytoD and KabC inhibited comet tail disassembly. All drugs were added at time 0. Bar, 5 μm. (D) Curves showing decay of actin polymer mass from L. monocytogenes comet tails. In the absence of drugs, actin disassembled at an initial rate of k_control = (2.3 ± 0.2) × 10⁻²s⁻¹. LatB did not change the disassembly rate, with k_LatB = (2.6 ± 0.2) × 10⁻². In contrast, CytoD reduced the L. monocytogenes comet tail decay rate to k_CytoD = (1.9 ± 6) × 10⁻⁴s⁻¹. Data represent the mean of polymer mass in multiple tails (control, 27 tails; CytoD treatment, 40 tails; LatB treatment, 27 tails). Disassembly rates were obtained by fitting the initial slope of the decay curve to a straight line.

Figure 7.

Barbed-end cappers inhibit disassembly of dynamic actin at the ruffling cell edge. (A–C) Time-lapse images of ruffling edges in BSC-1 cells expressing mRFP-PAGFP-actin taken before CytoD addition (A), after LatB addition (B), or after CytoD addition (C). The red channel shows bulk actin and the green channel shows photoactivated fluorescence. Images were taken before activation (left), immediately after activation (middle), and 100 s after activation (right). (D–F) Kymographs taken along the lines shown in A, B, and C, respectively. The color kymograph shows bulk actin in red and photoactivated actin in green. The black and white kymograph shows photoactivated actin alone. Arrows show the time of CytoD addition (D and F) or LatB addition (E). (A and D) Before drug treatment, photoactivated subpopulations of actin decayed within 100 s. (B and E) LatB did not inhibit actin disassembly. Both bulk and photoactivated subpopulations of actin disappeared within 100 s. (C and F) CytoD inhibited disassembly of dynamic actin assemblies at the ruffling edge. Photoactivated actin subpopulations remained visible for >100 s, and bulk actin fluorescence persisted at the ruffling cell edge. Bar, 5 μm.

Figure 8.

Candidate mechanisms for actin filament disassembly by cofilin, coronin, and Aip1. (A) Cooperative strand separation. (B) Filament severing. Black lines denote filament strands. The blue box denotes CapZ. The red bar denotes inhibition of CapZ binding. Successive illustrations depict reaction intermediates in each disassembly pathway.