Distinct VASP tetramers synergize in the processive elongation of individual actin filaments from clustered arrays

Abstract

Ena/VASP proteins act as actin polymerases that drive the processive elongation of filament barbed ends in membrane protrusions or at the surface of bacterial pathogens. Based on previous analyses of fast and slow elongating VASP proteins by in vitro total internal reflection fluorescence microscopy (TIRFM) and kinetic and thermodynamic measurements, we established a kinetic model of Ena/VASP-mediated actin filament elongation. At steady state, it entails that tetrameric VASP uses one of its arms to processively track growing filament barbed ends while three G-actin-binding sites (GABs) on other arms are available to recruit and deliver monomers to the filament tip, suggesting that VASP operates as a single tetramer in solution or when clustered on a surface, albeit processivity and resistance toward capping protein (CP) differ dramatically between both conditions. Here, we tested the model by variation of the oligomerization state and by increase of the number of GABs on individual polypeptide chains. In excellent agreement with model predictions, we show that in solution the rates of filament elongation directly correlate with the number of free GABs. Strikingly, however, irrespective of the oligomerization state or presence of additional GABs, filament elongation on a surface invariably proceeded with the same rate as with the VASP tetramer, demonstrating that adjacent VASP molecules synergize in the elongation of a single filament. Additionally, we reveal that actin ATP hydrolysis is not required for VASP-mediated filament assembly. Finally, we show evidence for the requirement of VASP to form tetramers and provide an amended model of processive VASP-mediated actin assembly in clustered arrays.

Keywords: Ena/VASP; TIRF; actin; cluster; formin.

Fig. 1.

VASP-mediated filament elongation correlates directly with N in solution. (A) Schematic domain organization of the generated VASP-oligomerization mutants. Abbreviations: EVH1, Ena/VASP-homology domain 1; PRD, proline-rich domain; EVH2, Ena/VASP-homology domain 2 containing the G-actin-binding site (GAB) from Dictyostelium, the F-actin binding site (FAB), and the oligomerization domain. Abbreviations: F, FAB; G, GAB. (B) Sedimentation velocity analysis of the used VASP-oligomerization mutants fused to MBP. Differential sedimentation coefficient distribution analysis showed the proteins to sediment with s_20°,w = 4.0 S (VASP-1M), 5.7 S (VASP-2M), 7.3 S (VASP-3M), 8.8 S (VASP-4M), and 9.9 S (VASP-6M) and is consistent with their predicted oligomerization status. (C) Time-lapse micrographs of TIRFM assays—corresponding to Movie S1—for determination of elongation rates. The actin (1 μM, 23% Alexa 488 labeled) was polymerized in the absence or presence of 200 nM of the indicated VASP construct in TIRF buffer, respectively. Time is given in seconds. (Scale bar, 20 µm.) (D) Dependence of the acceleration of actin-filament elongation on the oligomerization state of VASP. The elongation rates derived from TIRFM movies as shown in C were divided by the rate of spontaneous actin assembly to obtain the acceleration factors for each VASP oligomerization construct at the concentrations indicated. Concentrations are given in monomers. Fifteen actin control filaments and 30 filaments for all other data points in presence of VASP were measured. (E) Acceleration of VASP-mediated actin assembly at 100 nM VASP oligomer increased as a function of N (=total GABs − 1) as predicted by the kinetic model (9). Red line shows linear fit.

Fig. S1.

Analytical ultracentrifugation reveals a linear relationship of the Stokes radius on the expected molar mass for the different oligomeric forms of VASP in a log–log plot. Logarithm of the Stokes radii (circles) as calculated from the sedimentation coefficients from Fig. 1B plotted against the expected molar masses calculated from amino acid composition of the different oligomeric forms of VASP (VASP-1M, 77.7 kg/mol; VASP-2M, 163.9 kg/mol; VASP-3M, 246.4 kg/mol; VASP-4M, 332.8 kg/mol; and VASP-6M, 489.5 kg/mol). Linear regression (solid red line) resulted in a slope m = 0.474, an intercept b = −0.226, and a correlation coefficient R² = 0.993.

Fig. S2.

Dependence of the oligomerization state of VASP on the acceleration of actin filament elongation. This figure corresponds to Fig. 1D but depicts the measured filament elongation rates instead of the acceleration factor.

Fig. 2.

Additional GABs in the same polypeptide chain of VASP are less efficient to accelerate filament elongation in solution at higher N values. (A) Schematic domain organization of generated chimeric VASP mutants containing additional Dictyostelium GABs in the EVH2 domain. (B) Time-lapse micrographs of TIRFM assays for determination of elongation rates. The actin (1 μM, 23% Alexa 488 labeled) was polymerized in absence or presence of 200 nM of the indicated VASP construct in TIRF buffer, respectively. Time is given in seconds. (Scale bar, 5 µm.) (C) Comparison of the actin-filament elongation rates derived from TIRFM movies as shown in B in the presence of different VASP mutants at the concentrations indicated. Concentrations are given in monomers. Data correspond to means ± SD. Fifteen actin control filaments and 30 filaments for all other data points in presence of VASP were analyzed. (D) Comparison of the acceleration factors of VASP oligomerization construct and VASP constructs containing multiple GABs at 100 nM (relating to monomer concentration). Red line shows linear fit. Note less augmented filament elongation rates with VASP constructs containing multiple GABs within the same EVH2 domain.

Fig. 3.

Clustered VASP promotes processive filament elongation at a fixed rate irrespective of its oligomerization state. (A) Actin assembly of 1 µM actin (23% Alexa 488 labeled) in 1× TIRF buffer in the presence of 40 nM CP on 2 µm Ø benzylguanine beads derivatized with the VASP constructs indicated. Red arrows heads highlight processively growing, buckling filaments with their barbed ends attached to the bead surface. Time is given in seconds. (Scale bar, 20 µm.) (B and C) Determination of the elongation rates from TIRFM time-lapse movies corresponding to Movie S2 revealed a single, fixed rate for all tested VASP constructs. For each mutant, the fastest filaments growing on coated beads were analyzed. Data correspond to means ± SD. n = 30. n.s., nonsignificant. Color coding as in Fig. 1.

Fig. 4.

VASP does not require actin ATP hydrolysis of terminal subunits for filament elongation. (A) TIRF images of 4 µM ADP-actin (23% Alexa 488 labeled) polymerized in 1× ADP-TIRF buffer in the absence (Top) or presence (Lower) of 200 nM VASP-4M. Time is given in seconds. (B) Elongation rates of ADP-actin in the presence of increasing concentrations of VASP-4M in solution. (C) Comparison of elongation rates of ADP-actin or ATP-actin in absence or presence of 200 nM VASP-4M. (D) TIRFM images of 4 µM ADP-actin (Top) and 1.0 µM ATP-actin (Lower) polymerized in presence of VASP-4M–coated microspheres and 40 nM CP. (E) Comparison of elongation rates from ADP-G-actin and ATP-G-actin in the presence VASP-4M–derivatized beads. Note similar decrease in presence of ADP-actin for spontaneous and VASP-mediated actin assembly. (A and D) Time is given in seconds. (Scale bars, 20 µm.) (B, C, and E) Data correspond to means ± SD. Fifteen actin control filaments and 30 filaments for all other data points in presence of VASP-4M were analyzed.

Fig. 5.

Profilin does not accelerate VASP-mediated filament elongation but is required to increase speed of formin-driven actin assembly. (A) Time-lapse micrographs of single TIRFM assays for determination of elongation rates in solution. The actin (1 μM, 23% Alexa 488 labeled) was polymerized in the absence or presence of either 5 µM PFN1, 10 nM VASP-4M, or 5 nM mDia1-C in TIRF buffer, respectively. Time is given in seconds. (Scale bar, 5 µm.) (B) Quantification of the elongation rates in solution derived from TIRFM experiments as shown in A. Bars represent means ± SD from at least three independent experiments. (C) Time-lapse micrographs of single TIRFM assays for determination of elongation rates in clustered arrays on beads. The actin (1 μM, 23% Alexa 488 labeled) was polymerized in absence or presence of either 5 µM PFN1 or PFN2a in TIRF buffer, respectively using SNAP-tag derivatized VASP-4M or mDia1-C beads. The spontaneous assembly of actin was inhibited by addition of 40 nM CP. No filament growth from mDia1-C–derivatized beads in the absence of PFN was observed. Time is given in seconds. (Scale bar, 10 µm.) (D) Quantification of the elongation rates on beads derived from TIRFM experiments as shown in C. Bars represent means ± SD from at least three independent experiments.

Fig. 6.

Dependence of VASP oligomerization and multiple GABs on processivity. (A) Assembly of 1 µM actin (23% Alexa 488 labeled) in 1× TIRF buffer in the presence of 40 nM CP on 2 µm Ø benzylguanine beads derivatized with the SNAP-tagged VASP constructs indicated. Red circles indicate regions with short capped filaments that detached from the beads. Time is given in seconds. (Scale bar, 20 µm.) (B) The actin (1 μM, 23% Alexa 488 labeled) was polymerized in 1× TIRF buffer in the presence of 25 nM SNAP-Surface 549-labeled VASP constructs indicated. White arrowheads indicate single VASP molecules (red) surfing on growing filament barbed ends (green). Note different timescales. Time is given in seconds. (Scale bar, 5 µm.) (C) Dot plot summarizing barbed-end dwell times of the tested VASP constructs. Red lines indicate mean values. n, sample size.

Fig. 7.

Context-dependent mechanisms of VASP-mediated filament elongation. (A) The rates of VASP-mediated actin filament elongation in solution are directly proportional to the number of free GABs (N). (B) The rates of VASP-mediated actin filament elongation after clustering on beads are fixed, irrespective of the oligomerization state or the number of additional GABs, and correspond to the rate of the canonical VASP tetramer using three free GABs (N = 3) as predicted by the kinetic model (9). Because after dense clustering, even monomeric VASP-1M can drive processive filament elongation at the same rate, the three free GABs must be provided by distinct VASP molecules. The constructs harboring additional GABs within the same polypeptide chain were omitted for reasons of clarity. Clustering of constructs is not shown for reasons of clarity. (C) Scheme for the proposed cellular arrangement of VASP and accessory proteins in the paracrystalline lattice beneath the plasma membrane. Top view is shown. Upon macroscopic clustering, mediated by multivalent interaction of the two C-terminal SH3 domains of IRSp53 and the four PRD regions within the VASP tetramer, the spacing of neighboring VASP molecules becomes close enough to reach a sufficiently high density of VASP at the surface to trigger long-lasting processive filament elongation in the presence of CP using three free GABs from neighboring VASP tetramers.