Parallel actin monomers in the 8S complex of actin–INF2

Abstract Polymerization and depolymerization of actin play an essential role in eukaryotic cells. Actin exists in cells in both monomeric (G-actin) and filamentous (polymer, F-actin) forms. Actin binding proteins (ABPs) facilitate the transition between these two states, and their interactions with these two states of actin are critical for actin-based cellular processes. Rapid depolymerization of actin is assisted in the brain and/or other cells by its oxidation by the enzyme Mical (yielding Mox-actin), and/or by the binding of Inverted Formin 2 (INF2) – which can also accelerate filaments formation. At their stoichiometric molar ratio INF2 and actin yield the 8S complex (consisting of 4 actin monomers: 2 INF2 dimer molecules). Using biochemical and biophysical methods, we investigate the structural arrangement of actin in the 8S particles and the interaction of INF2 with actin and Mox-actin. To that end, we show 2 D class averages of 8S particles obtained by negative staining electron microscopy. We also show that: (i) 8S particles can seed rapid actin assembly; (ii) Mox-actin and INF2 form 8S particles at proteins ratios similar to those of unoxidized actin; (iii) chemical crosslinkings suggest that actin monomers are in a parallel orientation in the 8S particles of both actin and Mox-actin; and (iv) INF2 accelerates the disassembly of Mox-F-actin. Our results provide better understanding of actin's arrangement in the 8S particles formed during actin depolymerization and in the early polymerization stages of both actin and Mox-actin. Communicated by Ramaswamy H. Sarma


Introduction
Actin is a vital protein in eukaryotic cells and it participates in many important cellular processes. In cells, it exists as monomeric G-actin and filamentous F-actin. Actin polymerization involves a step of nucleation followed by the elongation process. Nucleation kinetics is mainly limited by the formation and stability of actin dimers and trimers (Sept et al., 1999). This step is facilitated by several actin binding proteins (ABPs) that regulate the actin cytoskeleton (Rotty & Bear, 2014).
Formin is one of the actin-binding proteins that can accelerate actin's polymerization processes. Formin has two domains: formin homology 1 (FH1) and formin homology 2 (FH2). The acceleration of actin polymerization is facilitated by FH2 domains (Baker et al., 2015), which interact with the barbed ends of filaments (Moseley et al., 2004). Inverted formin 2 (INF2) is a formin isoform that can interact directly with actin through a region C-terminal to the FH2 (Chhabra & Higgs, 2006). In addition to FH2, INF2 contains also FH1, a C-terminal diaphanous auto-regulatory domain (DAD), N-terminal diaphanous inhibitory domain (DID), and a dimerization domain (DD). INF2 accelerates G-actin polymerization but can also accelerate F-actin depolymerization. INF2 depolymerizes actin filaments to 8S particles when the ratio of actin to INF2-FFC (containing the FH1, FH2, and C-terminal regions) is 2:1. It has been suggested that the 8S particle comprises 4 actin monomers and 2 INF2 dimers (Gurel et al., 2015). Upon their crosslinking with 1,4-phenylenebismaleimide (1,4-PBM), two possible actin dimers ("upper" and "lower" (Silv an et al., 2012)) have been observed in SDS electrophoresis assays. It has been suggested that the "upper dimer" is in a parallel arrangement, while the "lower dimer" has anti-parallel arrangement (Bubb et al., 2002). However, the in vivo existence and dominance of parallel vs anti-parallel dimer needs further clarification.
In this study we explore 8S particles formation and their structural organization for both actin and Mox-actin by utilizing biochemical, biophysical and structural biology approaches. There are two possible structural organizations of actin in the 8S particles. Actin monomers can form tetramers and retain their parallel arrangement (Figure 1(A)), similar to that found in F-actin. The other possibility, albeit less likely, is that actin monomers are in an anti-parallel arrangement in the tetramer (Figure 1(B)). To check which arrangement is favored in the 8S particles, we utilized chemical crosslinkings and spectroscopic methods. Our analytical ultracentrifugation data reveal the formation of 8S particles by both actin and Mox-actin. Our spectroscopic, fluorescence, and chemical crosslinking results suggest that actin monomers are oriented in parallel in the 8S particles of both Moxactin and actin. We also demonstrate that the 8S particles accelerate the polymerization of G-actin. Interestingly, fluorescence assays focused on the dynamics of Mox-actin reveal that INF2 accelerates the disassembly of Mox-actin at a faster rate than that of actin. In conclusion, our study: (i) shows actin organization in the 8S particles, (ii) provides novel insight into Mox-actin assembly, and (iii) demonstrates the role of 8S particles as nucleating the assembly of Mox-actin and actin. Taken together these findings pave the way for future exploration of the role of 8S particles and actin dynamics in Mical-mediated oxidation and assembly of actin.

FH2 domains of formins are highly conserved
The structures of formins FMNL3 and Bni1p, which have been solved by X-ray crystallography to near atomic resolution (Otomo et al., 2005;Thompson et al., 2013), share similar domain arrangements and phylogenetic lineage with INF2 ( Figure 2(A, B)). Sequence alignment of FH2 domains of the three proteins shows their >60% similarity (Figure 2(C)). Structural information on INF2 was insufficient, so we obtained a homology model of INF2 and compared it to the available two formin structures (Figure 2(D)). The superposition of the three structures shows their similar domain arrangements and little deviation (the root-mean-square deviation (RMSD) between INF2 and BN1P (1Y64) is 1.134 Å, between INF2 and FMNL3 (4EAH) it is 1.2 Å. The RMSD between all 3 structures is 1.1 Å), suggesting their similar interactions with actin.
Electron microscopy data support the formation of 8S particles Analytical ultracentrifugation (AUC) studies have shown that Inverted formins form the 8S particles when combined at a 1:2 molar ratio with actin (Gurel et al., 2015). We used negative stain electron microscopy to obtain images of these particles (Figure 3(A)). Representative images of micrographs and the picked particles are shown in Figure 3(B). We then performed image analysis using EMAN2 and obtained 2D class averages of the 8S particles (Figure 3(C)). A 3D model of the 8S particles (Figure 3(D)) reveals their globular shape. Additionally, the 3D model projections match well with the 2D class averages, which were used for model generation (Figure 3(E)). This is a first low-resolution imaging of the 8S particles.

8S particles aid in rapid assembly of actin filaments
A critical role of 8S particles in actin polymerization is in overcoming its nucleation barrier. 8S particles were formed by incubating G-actin, INF2, and Latrunculin A (LatA) at a 2:1:4 RSA:INF2:LatA molar ratio, at room temperature, to ensure that actin was in its monomeric form when interacting with INF2. Actin was polymerized at a critical concentration of $1 lM, at which the rate of its assembly is very slow because of the nucleation barrier. When the 8S particles were added to this actin as polymerization seeds, they accounted for 5% of the final actin concentration. The presence of such seeds allowed actin monomers to bypass the nucleation barrier and polymerize rapidly ( Figure 4(A, B)). This data confirms that the 8S particles are crucial for surpassing the nucleation barrier, and that their presence (in vitro) as seeds can accelerate actin polymerization.
Fluorescence excimer assay does not support antiparallel actin arrangement in the 8S particles Pyrene excimer fluorescence is used to observe structural changes in actin dimers by revealing the proximity of its pyrene-tagged cysteine 374 residues ( Figure 5(A)). The polymerization of G-actin is induced by the addition of polymerizing salts in G-buffer. If G-actin assembles as antiparallel dimers we should observe excimer formation in the form of high fluorescence values (Grintsevich et al., 2010). However, we did not detect any excimer, indicating the absence of antiparallel organization of actin in the dimer ( Figure 5(B)).

Cys374-Cys374 chemical crosslinkings do not occur in the 8S particles
To investigate the nature of actin dimer in the 8S particles and to map the interactions between specific actin residues in them, chemical cross-linkings were carried out. To this end, Cys374 in the C-terminus of actin was targeted for the crosslinking. Cys374-Cys374 crosslinking experiments were carried out to check whether actins in the 8S structure were anti-parallel. In such a case the two Cys residues could be linked with various length span cross-linkers: Cu (oxidation to form a disulfide bond), 1,1-Methanediyl Bismethanethiosulfonate (MTS1, $6 Angstrom) and 1,8-Octadiyl Bismethanethiosulfonate (MTS8, $13 Angstrom).
Previous studies have shown that a cross-linked anti-parallel dimer, also known as a 'lower dimer', migrates around 75kD on SDS-PAGE gels (Silv an et al., 2012). No crosslinked antiparallel lower dimer appeared in our assays when the 8S particles were crosslinked with Cu, MTS1, and MTS8 ( Figure  5(C)). Moreover, the intensity of the actin monomer band in the crosslinked actin and in the 8S particles was close to its intensity in uncrosslinked actin and in 8S particles ( Figure  5(D)). This indicated the absence of any chemical crosslinking between two Cys374 residues. Thus, our crosslinking and fluorescence assays demonstrate that these two Cys residues are distant, do not come into close proximity, and thus are not in an antiparallel conformation.

Similar properties of 8S particles of actin and Mox-actin
Similar to MTS crosslinkings carried out with unmodified actin, different cross-linkers (Cu, MTS1, and MTS8) were used to test the possibility of antiparallel Cys374-Cys374 arrangement of actin in Mox-actin 8S particles. No cross-linked antiparallel (lower) dimer was detected in these reactions ( Figure  6(B)). Also, the intensity of Mox-actin monomer on SDS gels ( Figure 6(B, C)) was unchanged by cross-linking reactions with the 8S particles (Figure 6(C)).
We also tested for a transient presence of antiparallel actin dimer confirmation in 8S particles under conditions of actin filaments assembly. This was done by using Cys374 pyrene labeled actin in the 8S particles and checking for any excimer formation during actin polymerization. As shown in Figure 6(D) there was no fluorescence increase (excimer formation) in the reactions containing Mox-actin and INF2 compared to Mox-actin alone.
Mox-actin behaves similarly to wild type actin when provided as an 8S seed particle, such that Mox-actin elongation is accelerated in the presence of these seeds. Notably, our 8S particles were formed by incubating monomeric Mox-actin, INF2, and Latrunculin A at a 2:1:4 Mox-actin:INF2:LatA molar ratio (at room temperature for 30 min) to ensure that Moxactin was retained in its monomeric form when interacting with INF2 to form the 8S particles. Mox-actin was polymerized at a critical concentration of $1 lM, and the presence These models are based on (Gurel et al., 2014(Gurel et al., , 2015. The D-loop (deep pink) of actin and amino acid 374 C (red) are highlighted to provide easy distinction between parallel (A) and antiparallel (B) orientations of the 4 actin subunits. The size of the particles is approximately 45 nM. The distance between 374 C residues in the models of parallel and anti-parallel dimers is 55.9 Å (Merino et al., 2018) and 8.2 Å (Reutzel et al., 2004), respectively. Actin structure was obtained from PBD (6DJN). Our data suggest that parallel arrangement of actins is favored in the 8S particles. INF2 is not included in A and B for image clarity.
of 8S seeds allowed actin monomers to bypass the nucleation barrier and polymerize rapidly (Figure 4(A)). Clearly, the action of small amounts of 8S particles as seeds demonstrates their ability to induce actin assembly.

Mox-actin filaments disassemble more rapidly in the presence of INF2 than actin filaments
Fluorescence assays of actin filaments depolymerization (Figures 6(E) and S1) show that INF2 can interact with Moxactin filaments and depolymerize them much faster than non-oxidized actin filaments. More specifically, in these experiments the polymerization of 4 lM pyrene-labeled Gactin and 4 lM pyrene-labeled Mox-actin was monitored, and upon its completion INF2 was added to the reaction mixture. The observed fluorescence changes documented that the rate of Mox-actin filaments formation was considerably slower than that of actin filaments ( Figure S1). With that, the disassembly of Mox-actin filaments was much faster than that of actin filaments (Figures 6(E) and S1). Pelleting assays have a limited value here since Mox-actin alone does not yield filaments that can be pelleted ( Figure S2) under our conditions. Such pelleting experiments document well the binding of INF2 to F-actin and reveal only a small amount of INF2 and Mox-F-actin present in the pellets ( Figure S2).   (Edgar, 2004). Residues with 90% identity are shown in blue in (C). (C) Phylogenetic tree (Neighbor-joining) of FH2 domains and their relationship to each other. (D) The isoforms of FH2 domain are illustrated structurally in (i), (ii), and (iii); (iv) shows the superimposed FH2 domains. The structural models of Bni1p (1Y64) and FMNL3 (4EAH) were obtained from PDB. The structural model of INF2 was generated by homology modelling using PHYRE-2 (Kelley et al., 2015). To confirm the model generated by Phyre2 we used Swiss Model (Waterhouse et al., 2018) and DTU server (Nielsen et al., 2010) to generate additional models (data not shown) and those were similar to the one generated by PHYRE-2. 2015). The area that received less attention is the interaction of INF2 with monomeric G-actin, despite its constant presence in the cells. Thus, it has been important to look in some detail at that interaction. This work addresses that task. Because INF2 has two binding sites for actin it has been expected that it would form a complex with two actin monomers, leading to a question of what is the form of the resulting actin dimer in a complex with INF2.

Discussion
An antiparallel actin dimer has been regarded as a key intermediate during the nucleation stage of actin polymerization (Grintsevich et al., 2010;Steinmetz et al., 1997). While it was previously shown that polylysine arrests actin in the antiparallel dimer conformation, whether INF2 exhibits similar properties was unknown (Bubb et al., 2002). In this study, we tested for a hypothetical antiparallel actin arrangement in the actin-INF2 8S complex.
Our polymerization results (Figure 4) showed that the 8S particles aid in the rapid assembly of actin filaments. This result suggested a likely parallel arrangement of actins in the 8S complex since such an arrangement is more favorable to elongation of actin filaments than the anti-parallel one. In a test of actin arrangement in the 8S complex, an excimer assay was conducted using Cys-374 pyrene labeled actin, but no excimer was found in this complex ( Figure 5(B)). This result is consistent with a parallel arrangement of actin monomers in the model of 8S particles (Figure 1(A)) and inconsistent with their anti-parallel arrangement. In the anti-parallel 8S particles model structure the distance between two 374 C residues would have been 8.2 Å (as taken from the previously solved structure of anti-parallel actin dimer (Reutzel et al., 2004)). Thus, in the case of antiparallel arrangement of actin monomers (Figure 1(B)), with pyrene labels positioned within 16 Å from each other, an excimer would have been detected. To support the conclusion on the absence of antiparallel actin arrangement in the 8S complex, crosslinking of actin Cys374 residues was attempted with reagents of different length  (Tang et al., 2007). The approximate size of the particles is $40nM. (D) A 3D model generated from the 2D class averages shown in different orientations along the X,Y axis. (E) Examples of class averages matching the model projections.
span, but none was observed ( Figure 5(C, D)). This confirms the result of our excimer experiments that actin monomers are not arranged in an antiparallel fashion in the 8S particles. A parallel arrangement of actins in these particles is consistent with their ability to nucleate actin filaments assembly and to be incorporated in them.
Little was known about the interaction of Mox-actin with INF2, and whether Mox-actin can form the 8S complex was unknown. In this study, the formation of Mox-actin:INF2 8S complex was confirmed by our sedimentation velocity experiments (Figure 6(A)).
In analogy to the actin-INF2 complex, Mox-actin-INF2 also did not show any excimer formation when using pyrene labeled actin. This result, and the cross-linking tests carried out in the same way as with the actin-INF2 8S complex, confirmed that Mox-actin monomers, similar to (unoxidized) actin, were arranged in a parallel way in the 8S particles ( Figure  1(A)). Thus, the structural arrangement of the 8S particles of actin-INF2 and Mox-actin-INF2 is the same, or very similar.

Rapid disassembly of Mox-actin filaments by INF2
MICALs and their ability to oxidize and disassemble F-actin have emerged as an important means to disassemble F-actin in cells (Alto & Terman, 2018;Fremont et al., 2017;Manta & Gladyshev, 2017;Vanoni, 2017). MICALs trigger disassembly of F-actin and also generate Mox-actin that has a reduced ability to polymerize (Grintsevich et al., 2016;Hung et al., 2010Hung et al., , 2011Wu et al., 2018). We now add to these results by uncovering the rapid disassembly of Mox-actin filaments by INF2 (Figure 6(E)). In particular, under identical conditions in the presence of INF2, Mox-actin filaments disassemble much faster than actin filaments. Interestingly, work in vivo also supports a role of INF2-type formins in working with Micalsuch that regulation of INF levels in vivo generates defects that resemble similar changes to Mical (see Grintsevich et al., 2021). Our results therefore also add to our recent observation that Mox-F-actin is disassembled by profilin (and at lower profilin concentrations than F-actin) (Grintsevich et al., 2021). Thus, in conclusion, these results also add to an important emerging theme that Mical and its interaction with other ABPs (e.g. Grintsevich et al., 2016Grintsevich et al., , 2021, target actin filaments for rapid disassembly.

Protein preparation and purification
Rosetta 2 DE3 Escherichia coli were used for the expression of INF2-FFC. After breaking the cells and getting the supernatant, a GST (glutathione S-transferase) column was first used to purify INF2. ProTEV was used after that for cleavage of GST. A gel filtration column was then used to purify INF2. The purified protein was dialyzed vs Hepes buffer (10 mM Hepes pH ¼ 7.4, 50 mM KCl, 1 mM MgCl 2 , 1 mM EGTA) and was stored at À81 C for further use. Rabbit skeletal muscle actin (RSA) was purified from acetone powder of muscles (Spudich & Watt, 1971) and Mical oxidized actin (Mox-Actin) was generated as described by (Grintsevich et al., 2016). N-(1-pyrene)-maleimide was obtained from Molecular Probes (Eugene, OR) or AnaSpec Inc. (San Jose, CA). RSA labeling with pyrene maleimide was carried out in thiol-free GB2 supplemented with 2 mM MgCl 2 and 100 mM KCl, at 1:2.5 (actin:dye) molar ratio, for 1 h on ice. The resulting pyrene-labeled F-actin was pelleted, depolymerized (GB2), and gel-filtered on Superdex S200 16/60 column (Kouyama & Mihashi, 2005). Mical-oxidized RSA (Mox-actin) was prepared and purified as described (Grintsevich et al., 2016).

Negative stain electron microscopy
Protein samples were applied to 400-mesh carbon-coated copper grids coated with formvar films (EM Sciences). After 60 s of adsorption, the grids were blotted dry and treated with 1% uranyl acetate for 45 s. The grids were examined in a Technai T12 electron microscope operated at 120 kV. The collected images were analyzed using IMAGE J software (Schneider et al., 2012). Image processing and 2D class averages were done with EMAN2 (Tang et al., 2007). Around 5397 particles that were picked manually were used to generate 2D class averages with EMAN2, and each class average consisted of >200 particle images. Individual classes were evaluated, and an initial 3D model was generated from the relevant class averages. Each model was then evaluated by inspection of the class vs projection images to select the final model.

Analytical ultracentrifugation
Stoichiometric mixtures of unlabeled G-actin, Mox G-actin and INF2 that are known to form 8S particles were mixed under polymerization conditions. Sedimentation velocity experiments were carried out at 20 C in a Beckman Optima XL-A analytical ultracentrifuge equipped with a photoelectric scanning system. Sedimentation boundaries of actin recorded at the beginning of the run, at 3000 rpm, provided the information on a total 8S concentration in the solution. The plateau regions of the boundaries were recorded at the top run speeds (45,000 rpm). Sedimentation coefficients distributions were determined from g(s) plots using the Beckman Origin-based software (Version 3.01).

Fluorescence excimer assays
The time course of pyrene-labeled G-actin polymerization was recorded using fluorescence signal detection in a TECAN instrument (with 365 nm excitation and 385 nm emission). The polymerization of 4 lM pyrene-labeled G-actin was induced with the addition of polymerizing salts. Excimer formation during actin polymerization was monitored using 360 nm excitation and 407 nm emission wavelengths in the TECAN instrument (Grintsevich et al., 2010).

Fluorescence based actin nucleation assay
To stabilize the monomeric form of actin prior to filaments assembly, 2 lM actin was mixed with Latrunculin A (at 1:2 molar ratio of G-actin:LatA) (Sigma). After addition of MgCl 2 (to replace Ca in G-actin) and a 3 min incubation period, the F-actin buffer was added as a polymerization salt and INF2 was added to the reaction well at the molar ratio of 2:4 (INF2:actin). 8S complex seeds were formed by incubating Gactin, INF2, and Latrunculin A (LatA) at a 2:1:4 RSA:INF2:LatA molar ratio, in the dark, for 30 min. The seeds were diluted prior to their addition to yield 5% of $1 lM pyrene-labeled G-actin. This procedure was repeated with Mox-actin (Grintsevich et al., 2016;Wioland et al., 2021).

Crosslinking experiments
Cys374-Cys374 crosslinking: G-actin was dialyzed overnight vs buffer containing 10 mM Hepes pH ¼ 7.4, 50 mM KCl, 1 mM MgCl 2 , and 1 mM EGTA. After the dialysis, actin (at a concentration of 3 lM) was incubated for 20 min with 1X KMEH and 1.5 M INF2 to form 8S particles. After the 8S particles were formed, crosslinkings were done for 30 min, with Cu (oxidation to form a disulfide bond), MTS1 (crosslinking distance of $6 Å) and MTS8 (crosslinking distance of $13 Å). These reactions were stopped by using 2-mercaptoethanol. Samples were then analyzed by SDS PAGE and the intensity of SDS gel bands was quantified by ImageJ (Schneider et al., 2012). All crosslinking regents were obtained from Toronto Research Chemicals Inc., North York (Ontario, Canada).

High-speed pelleting assays
For high-speed pelleting assays, Mg-ATP F-actin (10 mM) was prepared by incubating Ca-ATP-G-actin for 3 min in a polymerization ME exchange buffer (0.05 mM MgCl 2 and 0.2 mM ethylene glycol-bis(2-aminoethylether)-N, N, N', N'-tetraacetic acid [EGTA]). Actin stock was then diluted to 2 mM with 1xKMEH 7.4 buffer (50 mM KCl, 1 mM MgCl 2 , 1 mM EGTA, 10 mM HEPES, 0.2 mM ATP, 1 mM DTT, pH 7.4) in the presence of 1 mM INF2, followed by its incubation at room temperature for 1 h. These reactions samples were then subjected to high-speed centrifugation (TLA100 rotor, 80,000 rpm, 4 C, 20 min). High speed supernatants and pellets were analyzed by SDS-PAGE. The gels were stained with Coomassie Blue and quantified with ImageJ.

Statistics and reproducibility
All experiments were repeated at least three separate independent times. At least three independent protein purifications and multiple independent actin biochemical experiments were performed with similar results. Figure  legends list the sample size for each experiment. To the best of our knowledge the statistical tests are justified as appropriate. No cell lines were used in this study.