Figure 7. Overexpression of Twinfilin suppresses morphological defects caused by CARMIL hyperactivity.

(A) Representative images of F-actin staining in untreated B16F10 cells (control), and cells transfected with Flag-CARMIL1, full-length (FL)-myc-mTwf1, or both. Scale bar, 20 μm. (B) Average Microspike density in cells treated as in (A). Box and whisker plots show mean, first and third quartile, and the maximum and minimum values. Data averaged from two experiments (n = 19–25 cells per condition). Data averaged from two experiments. From Left to right: n = 19, 25, 20, and 25; mean microspike density 0.75, 1.13, 0.62, 0.58 filopodia per 10 μm of cell cortex. Error bars, s.e.m. ***p≤0.001, n.s. p>0.05 by one-way ANOVA with Tukey post hoc test. (C) ‘Earlier’ model for CP regulatory cycle, adapted from Fujiwara and colleagues (Fujiwara et al., 2014). Proposed steps in model: (1) V-1 globally inhibits Capping Protein (CP) in the cytoplasm, (2) membrane-associated CARMIL (at the protruding cell edge) catalyzes dissociation of V-1 from CP, (3) the resulting CARMIL-CP complex is partially active, binding weakly to free barbed ends to provide capping function, (4) an unknown factor or mechanism promotes dissociation of CARMIL from CP, allowing V-1 to rebind CP and complete the cycle. (D) Our revised working model for the CP regulatory cycle. We propose that V-1 functions to maintain a cytosolic reservoir of inactive CP, from which Twinfilin and CARMIL activate CP, generating two distinct forms of active CP in cells: Twinfilin-CP complexes and CARMIL-CP complexes. Twinfilin-CP complexes are fully active and support stable capping of barbed ends. In contrast, CARMIL-CP complexes have ~100 fold reduced affinity for barbed ends, and may therefore more transiently cap barbed ends, permitting restricted network growth at the cell membrane where CARMIL localizes. CARMIL and Twinfilin directly compete with each other for binding CP (shown in close up of Transition state), which may result in the displacement of CP from Twinfilin. This would leave Twinfilin at the barbed end to catalyze depolymerization, or alternatively return filaments back to the original state of assembly.

Figure 7—figure supplement 1. Structural model for a ternary complex formed by Twinfilin, Capping Protein and the barbed end of an actin filament.

The three actin protomers at the barbed end are shown in gray, the CP heterodimer is shown in yellow, and Twinfilin is shown in blue. The structure of CP bound to the barbed end was solved by cryo-EM (Narita et al., 2006). We docked the Twinfilin ADFH domains onto this structure using the following criteria. The Twinfilin N-terminal ADFH (N-ADFH) lobe, shown bound between the first and third subunits from the barbed end of the actin filament, is based on the structure of this domain bound to G-actin (PDB 3DAW) (Paavilainen et al., 2008), and assumes that this domain also uses its F-site to bind to F-actin in a manner similar to cofilin, which was solved by cryo-EM (PDB 5YU8) (Tanaka et al., 2018). The Twinfilin C-terminal ADFH (C-ADFH) lobe, shown bound to the first subunit of the actin filament barbed end, is based on the solution structure of this domain bound to G-actin via is G/F-site. The Twinfilin linker sequence (connecting the two ADFH domains) is depicted as an unstructured chain, 25 residues long, which if unstructured (87 Å) is sufficient in length to allow each ADFH domain to bind a different actin protomer. There is no structure available for the CPI motif-containing C-terminal tail of Twinfilin; therefore, in this model we used the known structure of the CPI region of CARMIL bound to the stalk of CP (PDB 3AAE) (Hernandez-Valladares et al., 2010). Attempts to model Twinfilin bound to the opposite strand of F-actin did not allow sufficient length between the tail (binding CP) and the C-terminal ADFH domain binding to actin.