Purification of human β- and γ-actin from budding yeast

The actin cytoskeleton is an essential, highly conserved and abundant component of cells. Simple eukaryotes tend to express a single actin isoform, whereas humans display tissue and cell specific expression patterns. Although closely related, actin isoforms can subtly differ in biochemical properties related to polymer formation, nucleotide hydrolysis and exchange, and interactions with one or more essential regulatory proteins (Allen et al., 1996; Moradi et al., 2017; Namba et al., 1992; Perrin and Ervasti, 2010). Humans express six actin isoforms: α1, α2, α-cardiac and γ2 (encoded by ACTA1, ACTA2, ACTC1 and ACTG2), which occur predominantly in skeletal, cardiac and smooth muscle cells, and β (ACTB) and γ1 (γ henceforth; ACTG1) found predominantly in non-muscle cells and considered cytoplasmic isoforms of actin. Human β- and γ-actin are structurally divergent (Arora et al., 2023), yet differ by only four amino acids located within their first ten amino acids (Fig. 1A). Various muscle tissues are the most common sources of actin for use in biochemical studies. Actin purified in this manner is present as a mixture of muscle isoforms and minor amounts of β- or γ-actin. Characterizing specific isoforms of actin has been limited by accessibility and is critically important for understanding mechanisms of disease (Parker et al., 2020). Thus, it has been challenging to discern the biochemical properties of either β- and γ-actin in biochemical assays.

Producing recombinant human actin outside of eukaryotic cells is difficult due to the complex network of chaperones needed to properly fold actin; however, several systems have been developed (Geissler et al., 1998; Grantham, 2020; Millán-Zambrano and Chávez, 2014; Schafer et al., 1998; Valpuesta et al., 2002). The pCold system permits the bacterial synthesis of recombinant tagged β-actin (Tamura et al., 2011). Actin isoforms expressed and purified from popular eukaryotic systems produce relatively large quantities of biochemically active actin (Bergeron et al., 2010; Bookwalter and Trybus, 2006; Ohki et al., 2009; Rutkevich et al., 2006; Yamashiro et al., 2014). Although these and related purification methods have been adopted for various studies of normal and mutant versions of actin, preparations are often contaminated with low amounts (5–15%) of host actin (Hundt et al., 2014; Müller et al., 2012, 2013; von der Ecken et al., 2016). Other systems use a combination of affinity tags and a direct fusion to the actin monomer-binding protein thymosin-β4 (Tβ4; also known as TMSB4) to prevent the spurious polymerization or aggregation of recombinant actin and to facilitate isoform specific purification (Mu et al., 2020; Hatano et al., 2018; Hatano et al., 2020; Kijima et al., 2016; Lu et al., 2015; Noguchi et al., 2007). With the addition of NAA80 or SETD3 enzymes, this approach permits the isolation of actin in specific post translationally modified states, including N-acetylation, N-arginylation and methylation of a conserved histidine residue (Arora et al., 2023; Hatano et al., 2018, 2020). Additional approaches even permit the close preservation of native secondary modifications (Ceron et al., 2022). Although each system requires post-purification processing, they provide a source of non-muscle, normal or mutant (including non-polymerizable), actin isoforms.

Here, we have taken a different approach to generate pure human β-or γ-actin, engineering the yeast Saccharomyces cerevisiae to produce either isoform as their only source of actin. This technique was first pioneered to purify chicken β-actin from yeast, relying on hydroxylapatite chromatography to separate host and recombinant actin (Karlsson, 1988). A follow-up study showed that yeast could survive, albeit not well, with this as their sole actin source, although β-actin was not purified from these yeast strains (Karlsson et al., 1991). We have taken codon-optimized genes for human β- and γ-actin and expressed them in yeast lacking the resident actin gene ACT1. The resulting strains grow considerably slower than wild-type yeast, yet provide significant, low-cost and easily obtainable yields of homogeneous β- or γ-actin. Although actin purified by this approach does not undergo standard mammalian N-terminal processing, each actin isoform polymerizes and interacts with several actin-binding proteins including profilin, Tβ4, the formin mDia1 (also known as DIAPH1) and fascin. Thus, these engineered strains provide convenient sources of pure β- or γ-actin for biochemical studies where conventional α-actin sources might be unsuitable.

We expressed codon-optimized human cytoplasmic β- or γ-actin in Saccharomyces cerevisiae strains lacking the sole yeast actin gene (ACT1). The strain expressing human β-actin grew slowly (Fig. S1A) and displayed heterogeneous morphology (Fig. S1B–D), similar to what has been found for yeast expressing chicken β-actin (Karlsson et al., 1991). The strain expressing γ-actin grew even more slowly (Fig. S1A). However, both strains could be maintained without additional selection for the essential plasmid-borne actin genes. To confirm the identity and assess the posttranslational state of both isoforms, each recombinant actin was purified by affinity chromatography and subjected to mass spectrometry. This analysis showed 89% (β-actin), 95% (γ-actin) and 98% [α-actin; from rabbit muscle actin (RMA) as a control] coverage of the predicted peptide profiles. We took special note of the state of N-terminal peptides, which are the only peptides expected to differ between β- and γ-actin (Fig. 1A). We detected N-terminally acetylated β- and γ-actin (at Met1) (Fig. 1C,D), although not all intact peptides were N-acetylated. We also detected non-acetylated β- and γ-actin peptides truncated by one to five amino acids (Tables S1–S3) For comparison, all N-terminal peptides on α-actin (control) were truncated by two to six amino acids, present with and without acetylation on Asp3 (Fig. 1B; Tables S1–S3).

Two other important modifications of metazoan versions of actin include the methylation of His73, which might affect P_i release following ATP hydrolysis (Terman and Kashina, 2013), and Lys84, which might block myosin binding (Li et al., 2013). Each of these modifications were present in the α-actin (control) dataset but were absent in the β- and γ-actin spectra (Table S1). Finally, N-terminal arginylation of β-actin on Asp3 can influence cellular and biochemical functions (Chin et al., 2022; Karakozova et al., 2006). However, N-arginylation of β- or γ-actin on the third residue cannot be reliably distinguished from N-acetylation on the second residue (Tables S1 and S2) (Karakozova et al., 2006; Xu et al., 2009; Drazic et al., 2022). Not surprisingly, the modification profiles of β- and γ-actin reflect processing that normally occurs for yeast actin. Yeast processing includes the acetylation of N-terminal Met1 without further processing, and yeast lack additional methylating enzymes (Fig. 1; Tables S1 and S2; Kalhor et al., 1999).

The ability of both β- and γ-actin to support yeast growth suggests protein functionality (Fig. S2). To explore this idea further, we used classic biochemical assays to characterize each cytoplasmic actin. We began by assessing the ability of β- and γ-actin to polymerize into filaments with conventional pelleting assays (Fig. 2A,B). Polymerization of each actin was indistinguishable, with filaments pelleting to similar degrees at multiple concentrations (Fig. 2A–C). Each actin remained in the supernatant under controls performed in G-buffer (Fig. S2). To assess whether β- and γ-actin could co-polymerize, we performed epifluorescence microscopy to visualize actin filaments polymerized before or after mixing unlabeled and Oregon Green (OG)-labeled β- or γ-actin bound to rhodamine (Rh)–phalloidin (Fig. 2D,E). Rh–phalloidin stabilization of actins polymerized before mixing resulted in a mixed population of filaments including Rh–phalloidin-decorated unlabeled actin (only visible in the Rh–phalloidin channel) and OG-labeled filaments co-labeled with Rh–phalloidin (visible in both channels) (Fig. 2D,E, top rows). In contrast, mixing of actins prior to polymerization, followed by Rh–phalloidin stabilization, produced uniformly labeled actin filaments in both wavelengths, consistent with co-polymerization (Fig. 2D,E, bottom rows). Notably, filament co-labeling occurred regardless of which actin was OG labeled, consistent with co-polymerization of β- and γ-actin. To further assess the actin assembly properties of β- and γ-actin, we used total internal reflection fluorescence (TIRF) microscopy to directly observe single actin filament polymerization (Fig. 3A; Movie 1). Unlabeled 1 µM β-actin, γ-actin or control α-actin (RMA) was visualized with 10% fluorescently labeled RMA (Fig. 3A) (Chin et al., 2022; Hatano et al., 2018). Under these conditions, α-, β- or γ-actin showed similar means of nucleation ranging from 25.5±5.5 to 59.7±19.3 filaments per field of view (mean±s.e.m.; P≥0.45) (Fig. 3B). Control α-actin filaments (RMA) elongated at a rate of 10.0±0.3 subunits s⁻¹ µM⁻¹ (mean±s.e.m.), consistent with other studies (Fig. 3C) (Liu et al., 2022; Pimm et al., 2022). β-actin and γ-actin filaments polymerized at mean rates of 11.1±0.3 subunits s⁻¹ µM⁻¹ and 12.7±0.2 subunits s⁻¹ µM⁻¹, respectively (Fig. 3C). The mean elongation rate for γ-actin filaments was significantly faster than α-actin, β-actin or a 1:1 mixture of cytoplasmic isoforms (10.9±0.4 subunits s⁻¹ µM⁻¹; P=0.02) (Fig. 3C).

Cellular functions of actin are further modulated through interactions with many regulatory proteins. Therefore, we assessed the activities of several classic regulators of actin polymerization dynamics in the presence of either actin isoform. In mammalian cells, Tβ4 sequesters actin monomers to regulate available subunits for filament polymerization (Skruber et al., 2018, 2020). We assessed each human actin made in yeast for Tβ4 binding using fluorescence polarization (Fig. 4A). GFP–Tβ4 bound to β-actin [K_D=1.05±0.30 nM (mean±s.e.m.)] and γ-actin (K_D =0.94±0.09 nM) with similar affinities, which were each significantly stronger than skeletal muscle actin (K_D=7.71 nM±0.40; P<0.002) (Fig. 4A) (Pimm et al., 2022). This reinforces the notion that actin-binding proteins may have different functions with specific isoforms of actin.

To explore this idea further, we assessed the activities of profilin with each isoform of actin. Profilin directly inhibits α-actin assembly (Ferron et al., 2007; Pimm et al., 2020; Skruber et al., 2018). We began using pelleting assays to assess whether profilin (referring to profilin-1, PFN1) preferentially blocked the assembly of β- or γ-actin filaments (Fig. 4B). Profilin bound each actin isoform and strongly reduced signals in the pellet fractions of each cytoplasmic actin. We noted that profilin blocked the assembly of β- and γ-actin more effectively than muscle actin (P≤0.01) (Fig. 4C) (Antoku et al., 2019; Kinosian et al., 2000). However, no significant differences were observed for profilin-mediated nucleation in TIRF assays (P>0.99) (Fig. 4D,E).

When present with the actin polymerization-stimulating formin (mDia1), profilin enhances actin filament elongation (Chesarone et al., 2010; Li and Higgs, 2003; Valencia and Quinlan, 2021). Thus, to assess whether β- or γ-actin promotes formin-based filament assembly, we used TIRF microscopy to monitor actin filament polymerization in the presence of the constitutively active formin mDia1(FH1-C). Each actin isoform stimulated nucleation to similar levels (P>0.91) (Fig. 4D,E) and stimulated actin filament elongation when combined with formin and profilin (Fig. 4D,F; Fig. S4A–C), with γ-actin filaments elongating significantly faster than the other isoforms (see Fig. S4 for all statistical comparisons and rescaled plots). Compared to α-actin filaments, which grew at 73.6±4.5 subunits s⁻¹ µM⁻¹ (mean±s.e.m.), the mean elongation rate for β-actin and γ-actin with profilin and formin was not significantly slower at 59.3±4.2 or 61.2±4.5 subunits s⁻¹ µM⁻¹, respectively (P≤0.051) (Fig. 4F; Fig. S4A–C, Movie 2). In summary, α-actin from muscle and both cytoplasmic actins display similar properties in the presence of profilin and the formin mDia1, albeit to different levels.

Finally, the higher-order organization of cellular actin arrays is commonly achieved through the association of proteins that cross-link or bundle filaments. We produced actin bundles with fascin, stained filaments with fluorescent phalloidin and used TIRF microscopy to compare the higher-order structures made by each isoform (Fig. 5A). As actin filaments bundle, the overall area covered by pixel signal decreases (Fig. 5A). Meanwhile, the distribution of pixel intensities shifts to brighter pixels (Higaki et al., 2010; Khurana et al., 2010). Thus, we quantitatively measured the extent of actin filament bundling using length and intensity-based metrics to assess total bundling (Fig. 5B,C). As actin filaments coalesced into bundles, the number of objects detected in ridge analysis decreased to similar levels, regardless of actin isoform (P=0.98) (Fig. 5B). In contrast, the fluorescence intensity of β-, γ- and α-actin filaments crosslinked by fascin were similar to each other but each was significantly brighter than fascin-absent controls (P<0.01) (Fig. 5C). In conclusion, this demonstrates actin filaments of each isoform can be bundled by fascin.

We developed budding yeast strains for the expression and purification of human cytoplasmic β- and γ-actin, and have demonstrated that these actin isoforms polymerize and interact with several canonical actin-binding proteins. There are several benefits to purifying recombinant actin from this system including: yield (0.5–1 mg/l starting culture), no special growth or tissue culture requirements, use of conventional purification reagents and protocols, no additional post-purification processing (e.g. cleavage of fusion tags), and no contaminating ‘host’ actin. Even with these advantages, we note that for both yeast and human actin purifications, a significant amount of protein (>50%) is lost following post-elution dialysis. This likely reflects the denaturing activity of the formamide used to elute actin from the DNaseI column and might be improved by supplementing our approach with an alternative 6×His–gelsolin fragment-based purification (Ceron et al., 2022; Ohki et al., 2009).

Comparisons of actin isoforms or mixes that reflect other species and non-muscle cells are crucial to our understanding of how actin and its regulators truly function. For example, the formin FHOD1 displays differential interactions with actin from different sources, specifically enhancing the nucleation phase of actin assembly with cytoplasmic actin but preventing the assembly of filaments generated from RMA (Antoku et al., 2019; Patel et al., 2018). Profilin and cofilin proteins each bind cytoplasmic actin with higher affinity than to muscle actin (Antoku et al., 2019; De La Cruz, 2005; Kinosian et al., 2000). Furthermore, specific myosin motors differentially prefer muscle or non-muscle sources, some even preferring specific β- over γ-actin isoforms for movement (Müller et al., 2013). Our results identify Tβ4 as an additional actin regulatory protein influenced by actin source, binding β- or γ-actin with higher affinity than to actin purified from muscle.

Over 100 post-translational modifications (PTMs) of cytoplasmic and muscle actins have been described, with common reports of acetylation, arginylation, methylation and phosphorylation (Mu et al., 2020; Terman and Kashina, 2013; Varland et al., 2019). Post-translationally modified actin might influence the rate of polymerization, filament–filament interactions, actin-binding protein interactions, cellular localization and locomotion (Arnesen et al., 2018; Varland et al., 2019; Yamashiro et al., 2014). Perhaps most relevant for the discussion of cytoplasmic actin isoforms is N-terminal acetylation, as β- and γ-actin differ by four residues in this region (Fig. 1A). We found that yeast-produced β- and γ-actin were each acetylated on the N-terminal methionine (Tables S1–S3) (Cook et al., 1991). One notable missing modification from actin purified in this system is methylation of His73, which is important for the nucleotide-exchange of actin (Wilkinson et al., 2019). Introducing NAA80 or SETD3 to this system might ameliorate these challenges and expand the versatility of our budding yeast-based purification system (Arora et al., 2023; Hatano et al., 2020).

Most mammalian actin is processed to remove the N-terminal methionine and then acetylated or less commonly (and only for β-actin) arginylated at Asp3 (Varland et al., 2019). Studies in yeast and other systems further suggest that N-terminal arginylation occurs on clipped or deacetylated proteins, which might target them for proteasomal degradation (Drazic et al., 2022; Kumar et al., 2016; Nguyen et al., 2019). N-terminal differences in actin isoforms might be important for some applications and less consequential for others (Cook et al., 1992; Hatano et al., 2018). Notably, several studies have successfully purified and utilized versions of human actin lacking N-terminal modifications other than truncations (Mu et al., 2020; Lu et al., 2015). For those utilizing muscle actin to study interactions with non-muscle actin-binding proteins, differences between actin isotypes (primary amino acid composition) might prove more important than the exact nature of secondary modifications. Our system provides easily obtainable and relatively inexpensive sources of bulk cytoplasmic actin that will work well for many applications and that can be useful partners to systems that allow tighter control of specific PTMs or the synthesis of mutated versions of actin (Ceron et al., 2022; Hatano et al., 2020).

Unless otherwise specified, chemicals and supplies were purchased from Thermo Fisher Scientific (Pittsburgh, PA). DNaseI for affinity columns was purchased from Worthington Biochemical (Lakewood, NJ). Cloning reagents were obtained as follows: restriction enzymes and DNA ligase (New England Biolabs; Ipswich, MA); Primestar HS DNA polymerase (Takara Bio USA; San Jose, CA); and oligonucleotides (Eurofins Genomics; Louisville, KY).

Human α1-actin (ACTA1; NCBI Gene ID: 58) and β-actin (ACTB; NCBI Gene ID: 60) genes flanked by BamHI and HindIII sites were codon-optimized for expression in budding yeast (GenScript; Piscataway, NJ). Codon-optimized γ1-actin (ACTG1; NCBI Gene ID: 71) was generated from the ACTB sequence with the following mutagenic primer (5′-CTCGGATCCATGGAAGAAGAAATTGCTGCATTGGTTATTGATAATGGTTCTGGCATGTG-3′). The TDH3 promoter was PCR amplified as an EcoRI-BamHI fragment from yeast genomic DNA with 5′-GTGAGAATTCTCAGTTCGAGTTTATCATTA-3′ and 5′-CTCGGATCCTTTGTTTGTTTATGTGTGTTT-3′. TDH3 and actin genes were inserted into YEp351 (Hill et al., 1986). The yeast strain expressing only β-actin (BHY845) is a haploid segregant of an ACT1/act1Δ heterozygous diploid that carried the β-actin expression plasmid. The yeast strain that expresses only γ1-actin (BHY848) was generated by introducing the γ1-actin expression plasmid into a haploid act1Δ strain carrying the yeast ACT1 gene on a URA3-based plasmid (Haarer et al., 2007). We used 5-fluoroorotic acid (FOA) to select for yeast that lost the ACT1 (URA3) plasmid but remained viable. We were unable to recover yeast expressing human α1-actin, consistent with previous reports (McKane et al., 2005, 2006). We re-isolated and sequenced plasmids from β- and γ-actin strains at the time of harvest to confirm sequence fidelity. Each actin gene sequence was identical to the starting yeast-optimized ACTB and ACTG1 genes.

Human β- and γ-actin were prepared as described in Aggeli et al. (2014), with the following modifications: yeast were grown in 1 l of 1% yeast extract, 2% bactopeptone and 2–4% glucose (YPD), harvested by centrifugation at 8000 g for 7 min, washed in 25 ml 10 mM Tris-HCl pH 7.5, 0.2 mM CaCl₂. Pellets were collected by centrifugation at 4000 g for 10 min and stored frozen at −80°C. Pellets were suspended in 20 ml G-buffer (10 mM Tris-HCl pH 7.5, 0.2 mM CaCl₂, 0.5 mM ATP and 0.2 mM DTT) supplemented with protease inhibitors [0.1 mM PMSF and 1:500 Cocktail IV (Calbiochem; San Diego, CA)] and lysed by two passes through a French press at 1000 psi (6.9 MPa). Lysates were bound to 3 ml DNaseI–Sepharose, then beads were washed in five column volumes of each of the following: (1) 10% formamide in G-buffer, (2) 0.2 M NH₄Cl in G-buffer, and (3) G-buffer alone. Actin was eluted with 50% formamide in G-buffer, and then immediately diluted with 1–2 ml G-buffer and dialyzed overnight against 2 l G-buffer. To remove contaminating proteins, actin underwent at least two polymerization–depolymerization cycles alternating between 0.6 M KCl and pipetting with dialysis against G-buffer.

Rabbit muscle actin (RMA; α-actin), mDia1(FH1-C) (amino acids 571–1255), GFP–Tβ4, profilin-1 and fascin used in TIRF or anisotropy assays were purified as described previously (Jansen et al., 2011; Liu et al., 2022; Pimm et al., 2022). Actin was labeled with Oregon Green iodoacetamide or Alexa Fluor NHS-Ester (Aggeli et al., 2014; Hertzog and Carlier, 2005; Kuhn and Pollard, 2005). Proteins were aliquoted, flash-frozen in liquid nitrogen and stored at −80°C. Proteins used in TIRF assays are shown in Fig. S5.

Sample preparation, operation and mass spectrometry (MS) analyses were performed by the Upstate Medical University Proteomics and Mass Spectrometry Core Facility. Samples (15–30 µg) were treated with 5 mM tris(2-carboxyethyl)phosphine (TCEP) and then 15 mM iodoacetamide for 30 min in the dark. Trypsin digestion was performed at 37°C with 0.66 µg Trypsin Platinum (Promega, Madison, WI) overnight. Samples were acidified and then desalted using 2-core MCX stage tips (Rappsilber et al., 2003). Peptides were eluted with 75 µl of 65% acetonitrile (ACN) with 5% ammonium hydroxide, and dried. Samples were dissolved in 65 µl 2% ACN and 0.5% formic acid in water, then 0.5 µg was injected onto a pulled tip nano-LC column held at 50°C, with 100 µm inner diameter packed to 32 cm with 1.9 µm, 100 Å, core shell Magic2 C18AQ particles (Premier LCMS, Penn Valley, CA). Peptides were separated with a 3–28% ACN gradient over 60 min, followed by an increase to 85% ACN over 5 min. The inline Orbitrap Lumos was operated at 2.3 kV in data-dependent mode with a cycle time of 2.5 s. MS1 scans were collected at 120,000 resolution with a maximum injection time of 50 ms. Dynamic exclusion was applied for 15 s. Stepped HCD fragmentation of 34, 38 and 42% collision energy was used followed by two MS2 microscans in the Orbitrap at 15,000 resolution with dynamic maximum injection time.

The core facility used SequestHT in Proteome Discoverer (version 2.4; Thermo Fisher Scientific) to search three MS databases: S. cerevisiae (Uniprot, 6816 entries, retrieved 2017), Rabbit O. cuniculus (Uniprot, 41462 entries, retrieved 2023), a list of common contaminants (Thermo Fisher Scientific, 298 entries), and a custom list of human β-actin, γ-actin and rabbit α-actin. Additional entries for each actin were made with truncations of the first one, two and three N-terminal residues. Enzyme specificity was set to semi-tryptic with up to two missed cleavages. Precursor and product ion mass tolerances were set to 10 ppm and 0.02 Da. Cysteine carbamidomethylation was set as a fixed modification and the following modifications were set as variable: methionine oxidation, protein N-terminal acetylation, peptide N-terminal arginylation, lysine acetylation or methylation of lysine, histidine and arginine. The output was filtered using the Percolator algorithm. We visualized datasets with software at http://www.interactivepeptidespectralannotator.com/PeptideAnnotator.html (Brademan et al., 2019). Full data sets have been deposited at the PRIDE Database (ProteomeXchange identity: PXD040174).

Polymerization of 1 or 2 µM actin was induced with the addition of concentrated (20×) F-buffer (final 1× concentration: 10 mM Tris (pH 7.5), 25 mM KCl, 4 mM MgCl₂, 1 mM EGTA, 0.5 mM ATP). Samples were incubated for 30 min at room temperature, then subjected to centrifugation at 200,000 g for 30 min at 20°C. Supernatants were removed and pellets and supernatant samples were brought to equal volumes in protein loading buffer. Equal volumes were loaded on SDS-polyacrylamide gels and stained with Coomassie Brilliant Blue. Indicated concentrations of profilin-1 were mixed with 2 µM actin monomers in G-buffer and then incubated for 30 min at room temperature. Actin polymerization was then induced with F-buffer for an additional 30 min prior to centrifugation as above.

Actin isoform co-polymerization reactions were assessed by epifluorescence microscopy (Zeiss Imager.Z1, Oberkochen, Germany). 2 µM unlabeled or OG-labeled β- or γ-actin were polymerized in F-buffer, either individually or as a 1:1 mixture, then bound to 1.1 µM rhodamine–phalloidin. Samples were viewed individually or as mixtures bound to phalloidin prior to mixing. Samples were flowed into slide chambers prepared by laying a cover glass onto a slide with two intervening strips of double-sided tape, then visualized with DsRed and FITC filter sets. Views were obtained near fluid–air boundaries.

We prepared and visualized TIRF imaging chambers on a DMi8 TIRF microscope (Leica Microsystems; Wetzlar, Germany) as in Henty-Ridilla (2022), with the following modifications: reactions were executed in a different TIRF buffer [20 mM imidazole pH 7.4, 50 mM KCl, 1 mM MgCl₂, 1 mM EGTA, 0.2 mM ATP, 10 mM DTT, 40 mM glucose and 0.25% methylcellulose (4000 cP)], with minimal (5–7%) labeled RMA. Some TIRF experiments with β- or γ-actin were supplemented with 10% RMA to enable visualization of filaments, as noted in figure legends. Experiments performed with fascin used unlabeled actin isoforms and were visualized with 130 nM Alexa Fluor 488-conjugated phalloidin. Filament elongation rates (subunits s⁻¹ µM⁻¹) were determined by measuring filament lengths from at least five frames, with a conversion factor of 370 subunits/µm (Kuhn and Pollard, 2005). Total filament or bundle length was calculated using the FIJI Ridge Detection plugin with settings that minimized background signals but permitted the detection of faint filaments without image saturation and applied identically to all images (Steger, 1998; Wagner et al., 2017). The skewness parameter was measured from FIJI measurements (Higaki et al., 2010; Khurana et al., 2010; Schindelin et al., 2012).

Fluorescence polarization determination of GFP–Tβ4 binding to actin (Liu et al., 2022) was carried out in 1× PBS (pH 8.0) supplemented with 150 mM NaCl. 10 nM β- or γ-actin were mixed with various concentrations (0.1 pM to 10 µM) of GFP–Tβ4 and incubated at room temperature for 15 min. Fluorescence polarization was measured in a plate reader equipped with a monochromator, with excitation at 440 nm and emission intensity detection at 510 nm (bandwidth set to 20 nm) (Tecan; Männedorf, Switzerland). Three technical replicates were carried out on the same plate.

GraphPad Prism (version 9.5.0; GraphPad Software, San Diego, CA) was used to plot all data and perform statistical tests. All experiments were repeated at least three times. Individual data points are presented as dots in each figure and histograms represent mean±s.e.m. (unless noted otherwise). All one-way ANOVA tests performed compared all means with Tukey post-hoc analysis and passed tests for normality. P-values presented are those from ANOVA tests across tested conditions (under the line), except for Fig. S4 where all comparisons are listed. The threshold (P≤0.05) was used to determine significance throughout this work. Specific comparisons are described in each figure legend. Figures were made in Adobe Illustrator 2023 (version 27.1.1; Adobe, San Jose, CA).

We thank Stephan Wilkens and Rebecca Oot for the use of their FPLC and assistance with protein labeling. Tom Duncan, Stewart Loh and Leszek Kotula for the use of equipment. Marcela Alcaide Eligio for assistance with fascin purification. Marc Ridilla for critically reading this work and D. D. Johnson for a greater appreciation of muscle actin. The Orbitrap Lumos used by the SUNY Upstate Medical University Proteomics and Mass Spectrometry core facility was obtained through NIH instrumentation grant 1S10OD023617-01A1.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260540.reviewer-comments.pdf