Introduction

Cytoskeletal β– and γ–actin are the basic building blocks of the actin cytoskeleton in human non– muscle cells. The spatio–temporal regulation of these two actin isoforms, mediated by distinct actin– binding proteins (ABPs), drives the establishment and maintenance of specialized cytoskeletal compartments that are defined by their network architecture and the presence of specific filamentous actin–associated proteins (Pollard, 2016; Plastino & Blanchoin, 2019). The regulated ATP–dependent turnover of these actin networks as well as the coordinated interplay of actin with network–specific ABPs contribute to essential cellular processes like directed migration, adhesion, division and mechanotransduction (Ponti et al, 2004; Abercrombie et al, 1970, 1971; Ridley et al, 2003; Pollard et al, 2002; Lauffenburger & Horwitz, 1996).

β– and γ–actin differ by only four amino acids in the N–terminal region of the molecule but show isoform–specific spatial organization in studies performed with monoclonal antibodies. In motile and stationary cells, β–actin is found in contractile stress fibers and the cell periphery (Dugina et al, 2009; Latham et al, 2013; Dugina et al, 2015). In contrast, γ–actin localization is less static. During cell movement, γ–actin is recruited to the leading edge of the cell, resulting in an enrichment of the isoform in the lamellipodium together with β–actin (Dugina et al, 2009). In non–moving cell states, γ–actin is co–localized with β–actin in stress fibers (Dugina et al, 2009). In order to reconcile the highly similar protein sequences with the varying localization, several hypotheses on isoform regulation were put forward. These include the preferential interaction of ABPs with a particular isoform depending on the minor differences in the protein sequence, post–translational modifications as a means of isoform– specific regulation, and differences in the translation speed of ACTB and ACTG1, the genes encoding cytoskeletal β–actin and γ–actin (Zhang et al, 2010; Vedula et al, 2017; Pavlyk et al, 2018; Müller et al, 2013).

Diseases associated with heterozygous pathogenic variants in ACTB and ACTG1 are typically referred to as non–muscle actinopathies. These include a spectrum of rare syndromes that manifest themselves as more or less pronounced developmental disorders in patients (Verloes et al, 2015; Di Donato et al, 2014; Latham et al, 2018; Rivière et al, 2012; Baraitser & Winter, 1988; Cuvertino et al, 2017). Baraitser–Winter–Cerebrontofacial–Syndrome (BWCFF, (Baraitser & Winter, 1988)) is a condition that affects the development of multiple organs, but particularly results in facial anomalies and brain malformations. BWCFF occurs due to various missense mutations in ACTB or ACTG1 (Verloes et al, 2015; Di Donato et al, 2014; Rivière et al, 2012; Di Donato et al, 2016). BWCFF–patients show a recognizable facial gestalt, with variants in ACTB generally resulting in a stronger facial phenotype than mutations in ACTG1 (Di Donato et al, 2014). BWCFF–patients regularly suffer from cortical malformations that lead to severe developmental and mental impairments. In addition, symptoms such as hearing loss, dystonia, hypotonia and polydactyly occur with increased frequency (Di Donato et al, 2014; Rivière et al, 2012). Variants in ACTB alone are responsible for two other well–defined syndromes. ACTB–associated pleiotropic malformation syndrome and ACTB–associated syndromic thrombocytopenia can be caused by single–point missense mutations, truncating mutations, frameshift mutations or deletion (Latham et al, 2018; Cuvertino et al, 2017). These ACTB–associated syndromes differ from BWCFF in that they cause only mild craniofacial malformations and developmental disorders (Cuvertino et al, 2017; Latham et al, 2018). Selected single–point missense mutations in ACTG1 lead to a non–syndromic form of hearing loss (Rendtorff et al, 2006; van Wijk et al, 2003; Yuan et al, 2016).

How exactly mutations in the ubiquitously produced actin isoforms of the cytoskeleton lead to various syndromes and mutation-specific symptoms remains unclear. Currently, four main disease mechanisms are under discussion (Greve et al, 2022; Costa et al, 2004). First, mutations in the actin isoform can lead to functional haploinsufficiency, where the produced molecule is either intrinsically unstable and therefore degraded by the proteasome or unable to associate with other actin molecules and thus unable to partake in the necessary steps involved in the dynamic remodeling of the actin cytoskeleton. This results in a reduced amount of “active“ actin molecules, which cannot fully be compensated for by the remaining allele (Vang et al, 2005). Second, the mutant protein is produced in normal quantities and integrated into filaments but interferes with actin filament turnover (Costa et al, 2004). Third, the mutation perturbs interactions with actin binding proteins, either by direct or allosteric disturbance of actin–actin or actin–ABP interaction interfaces (Greve et al, 2022; Hundt et al, 2014; Bryan et al, 2006; Müller et al, 2012). Fourth, the mutant protein shows an increased propensity for forming toxic rod–like oligomeric structures, which can potentially harm the cell (Ilkovski et al, 2004; Schröder et al, 2004).

Therefore, one of the significant challenges is elucidating the relationship between the disturbed molecular mechanisms of the cell’s actin dynamics and disease patterns observed in patients. In order to achieve this, it is necessary to resolve the structure–function relationship of the various actin mutations at the molecular level. Here, we report the biochemical characterization of p.E334Q, a cytoskeletal γ–actin variant, which causes a syndromic form of non–muscle actinopathy. Features associated with the production of p.E334Q include mild facial anomalies and an abnormal gyral pattern of the cerebral cortex (Di Donato et al, 2016). At four years (age of diagnosis), a patient carrying this mutation showed developmental delay and severe muscle hypotonia (Di Donato et al, 2016). Although the patient was initially reported within the ACTG1–associated BWCFF cohort, careful reassessment of the phenotype revealed that the clinical phenotype cannot be classified as BWCFF and is therefore assigned as an unclassified NMA. Structural analysis of the disease–causing mutation shows that residue E334 is located in the hinge region of the actin molecule, which connects the two subdomain pairs of the protein. The region’s flexibility and ability to act as a pivot point are essential for the subdomain rearrangement when actin monomers polymerize into filaments (Figure 1A). Furthermore, E334 is part of a contact region on the F–actin surface that interacts with the so-called cardiomyopathy–loop in the motor domain of cytoskeletal myosin isoforms and contributes to the formation of the actin–cofilin complex (Figure 3). We present experimental evidence for the impact of changes involving residue E334 with respect to productive interactions with cytoskeletal myosin isoforms, actin polymerization dynamics, and the formation of stable actin-cofilin complexes. We show that mutation E334Q does not adversely affect protein folding. The thermal stability of p.E334Q resembles that of the wild–type protein, both as a monomer and following integration into filaments. The mutant protein efficiently copolymerizes with wildtype (WT) monomers and integrates into linear and branched filament networks. In contrast, the interaction of mutant filaments with non-muscle myosin-2A/C and myosin-5A is strongly impaired, as shown by in vitro motility experiments and solution kinetics. Furthermore, p.E334Q filaments are less susceptible to cofilin–mediated disassembly, while monomers are more susceptible to cofilin–mediated nucleation.

Figure 1

Results

Production of cytoskeletal γ–actin and its variant p.E334Q

We produced untagged cytoskeletal γ–actin and its variant p.E334Q in insect cells, free of contaminating endogenous S. frugiperda actin, using a method that was originally designed for the production of toxic actin mutants in Dictyostelium discoideum and later adapted for use with the baculovirus/Sf9 insect cell expression system (Noguchi et al, 2007; Lu et al, 2015). The monomeric actin constructs were initially produced fused to the actin-monomer sequestering protein thymosin-β4. Purification of the fusion protein by immobilized metal affinity chromatography, followed by chymotrypsin–mediated cleavage of C–terminal linker and tag sequences, results in homogeneous protein without non–native residues. The method worked equally well for both WT and mutant actin. Starting with 2 × 10⁹ cells, typical yields ranged from 5–7 mg of purified protein for cytoskeletal γ–actin and p.E334Q (N=3 for WT and N=4 for p.E334Q).

Impact of mutation E334Q on protein folding and thermal stability

Mutation–induced misfolding of actin has been identified as one of the main causes for the development of actinopathies (Bryan et al, 2006; Vang et al, 2005). Residue E334 is located in the hinge region, formed by residues 137–145 and 333–338 (Figure 1A), which is strongly involved in the structural reorganization of the monomer upon incorporation into a growing filament (Dominguez & Holmes, 2011). Previous studies have shown that an intact hinge region is crucial for the interactions with the eukaryotic chaperonin CCT/TriC and myosin (Noguchi et al, 2012; McCormack et al, 2001b, 2001a). We first performed a DNase–I inhibition assay to analyze the effect of mutation E334Q on the folding of the protein. Previous studies have shown that the activity of DNase I is only effectively inhibited by correctly folded actin (Schüler et al, 2000). We observed a 10% increase in the apparent IC₅₀–value of p.E334Q (25.7± 0.6 nM) compared to the WT protein (22.7± 0.7 nM) (Figure 1B). This is in good agreement with CD–spectroscopy results showing nearly identical secondary structure compositions for WT and mutant proteins (Suppl. Figure 1). Thermal denaturation temperatures (T_M) analyzed by differential scanning fluorimetry (DSF) gave values of 56.77 ± 0.24°C and 56.19 ± 0.30°C for monomeric WT actin and p.E334Q. The values obtained for WT and E334Q filaments were 65.00 ± 0.76°C and 64.83 ± 0.25°C, respectively (Figure 1C, Table 1). Overall, these results show that the mutant actin is correctly folded and has comparable stability to the WT protein.

Table 1

As the hinge region is in close proximity to the nucleotide–binding cleft, we examined a potential effect of the mutation on the interaction with nucleotides and the turnover of ATP. Initial tests gave near identical rates of 0.00089 ± 0.00005 s^-1 and 0.00088 ± 0.00004 s^-1 for the steady–state turnover of ATP by filaments formed from cytoskeletal γ–actin or the p.E334Q variant (Table 1). Next, we determined the nucleotide dissociation rates k_-T and k_-D by monitoring the decrease in fluorescence amplitude associated with the displacement of ε–ATP or ε–ADP after rapid mixing with a large excess of unlabeled ATP. With p.E334Q-actin, we observed rates of k_-T=0.0972 ± 0.0019 s^-1 and k_-D=0.3199 ± 0.0027 s^-1. These rates are about 2-fold and 1.5-fold faster than those observed in experiments performed with WT monomers (k_-T=0.0473 ± 0.0013 s^-1, k_-D= 0.2155 ± 0.0017 s^-1 (Figure 1D, Table 1). The observed nucleotide dissociation rates for human cytoskeletal γ–actin are in good agreement with values obtained for a mixture of pyrene-labeled cytoskeletal γ–actin (15%) and β–actin (85%) purified from human platelets in experiments performed at the same ionic strength (De La Cruz & Sept, 2010).

p.E334Q shows normal polymerization behavior and is efficiently integrated into linear and branched actin structures

The regulated assembly and disassembly of filamentous actin drive various cellular processes, including directed cell migration, adhesion and endocytosis. A slight conformational change of the actin monomer accompanies the integration of a monomeric actin molecule into a growing filament. A small change in the hinge region leading to a 12° to 13° rotation of the inner subdomain pair relative to the outer subdomain pair results in a flatter actin molecule within the filament compared to the free actin monomer (Dominguez & Holmes, 2011; Oda et al, 2009). Mutations in the hinge region can thus potentially affect filament assembly by impairing the required structural reorganization of the monomer, as it attaches to the growing filament end. TIRF microscopy–based tracking of the polymerization of monomeric Atto–655–labeled WT or mutant actin in vitro revealed a small but significant reduction in the elongation rate from 12.6 ± 0.3 nm s^-1 for WT to 11.1 ± 0.3 nm s^-1 for mutant filaments (Figure 2A, B). To detect any differences in nucleation efficiency, we determined the time-dependent increase in the number of filaments (Figure 2C). Aside from short-term fluctuations, the linear increase in the number of filaments recorded for WT (0.039 ± 0.002 s⁻¹) and p.E334Q (0.041 ± 0.004 s⁻¹) was similar over the full 1000 s duration of the experiment. Overall, p.E334Q shows only a mild polymerization defect.

Figure 2

Since actin polymerization in vivo is tightly regulated in a spatio–temporal manner by different ABPs, we analyzed the interaction of p.E334Q with one of the key–regulators of actin polymerization, the Arp2/3 complex (Goley & Welch, 2006). The Arp2/3 complex promotes the nucleation of new actin filaments from the sides of existing filaments and caps their pointed ends. Thereby, it promotes the formation of branched actin networks and contributes to essential processes that depend on the dynamic remodeling of the actin cytoskeleton, such as cell migration, endocytosis, and organelle positioning (Nicholson-Dykstra et al, 2005; Mullins et al, 1998; Blanchoin et al, 2000; Loisel et al, 1999). The ability of p.E334Q to form branched actin networks was probed in TIRF microscopy–based experiments in the presence of increasing concentrations of VCA–activated human Arp2/3 complex (Figure 2D). At the lowest Arp2/3 concentration used (0.01 nM), we observed the appearance of the first branches after approximately 250 s. In the presence of 0.1 nM and 0.2 nM of the VCA-activated Arp2/3 complex, mutant and WT actin continued forming branched networks with comparable branch formation and branch elongation rates. Similar results were obtained with WT actin, mutant actin or with 1:1 mixtures of WT and mutant actin, when we analyzed the effect of adding increasing concentrations of VCA–activated human Arp2/3 in pyrene–actin based bulk–polymerization assays (Figure 2E, F). The results are summarized in Table 1.

Impact of E334Q mutation on interactions with ADF/cofilin family proteins

Homology models of the complexes formed by human cofilin–1 with filamentous cytoskeletal γ–actin and p.E334Q were generated to evaluate the functional impact of mutation E334Q. Cofilin-1 binds to actin filaments by making simultaneous contact with two adjacent monomers. Actin residue E334 is located halfway between two adjacent cofilin molecules, but only makes direct contact with residue L111 of the cofilin–1 molecule on the pointed end side (Figure 3A). Cofilin–1 residue L111 is located in an α-helix (L111-L128) that is essential for actin–binding (Pope et al, 2000). Mutation E334Q maintains the van der Waals interaction with L111, but reduces the electrostatic attraction between the interacting surfaces, with likely effects on the stability and allosteric coupling of the complex (Figure 3B, C).

Figure 3

To evaluate the impact of mutation E334Q on the elastic properties of F–actin in the presence and absence of cofilin–1 binding, we performed normal mode analysis using bare, partially decorated and fully decorated filaments (Suppl. Figure 2A). Since residue E334 is located on the outer surface of the actin filament, the mutation produces no significant change in the normal mode frequencies of bare filaments. Similarly, fully decorated filaments show only minor differences. Substantial changes in the low-frequency modes associated with global conformational changes were observed when comparing the partially cofilin-decorated filaments of WT and mutant filaments, where the bare and cofilin– decorated regions move as rigid bodies and the interface between the two regions acts as a hinge (modes 7, 8) or fulcrum (mode 9) (Suppl. Figure 2B).

To test the validity of our modeling results, we performed biochemical and functional assays probing the interactions of human cofilin–1 with monomeric and filamentous cytoskeletal γ–actin and p.E334Q. The effect of cofilin–1 on filament assembly was assessed by TIRF microscopy experiments. The polymerization of 1 µM actin was induced in the presence of 0, 50 or 100 nM cofilin–1. Following an initial lag phase of 200 to 300 seconds, the filament formation rate increased with the cofilin–1 concentration (Figure 4). In experiments with p.E334Q-actin, de novo filament nucleation proceeded faster (Figure 4A, B). Movies showing filament formation by de novo nucleation in the presence of cofilin-1 are available in the supplement (Suppl. Video 1–6).

Figure 4

Next, we investigated the effect of cofilin–1 on the disassembly of preformed cytoskeletal γ–actin and p.E334Q filaments. The effect of cofilin–1 on existing WT and mutant filaments was initially tested by performing pyrene–based dilution–induced disassembly experiments with actin filaments polymerized for 16 hours and containing predominately ADP–actin. Cofilin–1 has previously been shown to increase the bulk–disassembly rate of actin in this experimental setup (Breitsprecher et al, 2011; Greve et al, 2022). In the absence of cofilin, the changes in pyrene–fluorescence are best described by single exponential functions with k_obs values of 0.0022 ± 0.0002 s^-1and 0.0029 ± 0.0003 s^-1 for WT and p.E334Q filaments, respectively (Figure 5A, Suppl. Figure 3). The addition of cofilin–1 increased the observed rate of the dilution–induced disassembly in a concentration–dependent manner for both mutant and WT filaments. At the highest cofilin–1 concentration used (40 nM), the change in pyrene–fluorescence with p.E334Q filaments is still best described by a single exponential function with a k_obs value of 0.0059 ± 0.0008 s^-1. In contrast, the change for WT filaments is best-fitted to a bi-exponential function. Here, the k_obs values for the fast and slow phases are 0.0163 ± 0.0028 s^-1 and 0.0002 ± 0.00003 s^-1, respectively (Figure 5A, Suppl. Figure 3). Additional experiments with heterofilaments showed milder but significant differences (Table 1).

Figure 5

The pyrene-based bulk experiments show an evident change in the p.E334Q-cofilin interaction. However, they cannot distinguish between changes affecting filament binding and filament disassembly. To analyze the perturbed interaction of p.E334Q with cofilin–1 in more detail, we first performed high-speed co–sedimentation assays with constant concentrations of F-actin (5 µM) and increasing concentrations of cofilin-1 (0.0625 – 20 µM) at pH 6.5 and pH 7.8. Cofilin-1 binds faster but severs filaments less efficiently under slightly acidic pH conditions, whereas a weakly basic environment favors cofilin–mediated severing (Wioland et al, 2019). We first performed experiments at pH 6.5. This endpoint experiment showed that cofilin–1 binds to and disassembles p.E334Q and WT filaments (Figure 5B, C). Most pellet fractions from the experiments conducted with WT filaments were found to contain less cofilin-1, as higher disassembly efficiency results in more cofilin–1 being bound to actin monomers and retained in the supernatant fraction. We repeated the experiment at pH 7.8 with an equimolar and a 2–fold excess of cofilin, (Figure 5B, C). In experiments with equimolar cofilin–1, we found significantly less WT actin in the pellet fraction than at pH 6.5, consistent with a higher disassembly–efficiency at pH 7.8. The shift to pH 7.8 did not have the same enhancing effect on the disassembly of p.E334Q filaments, as indicated by an increase of p.E334Q actin in the pellet fraction. The endpoint measurements show that cofilin–1 binds to p.E334Q filaments but severs them less efficiently. Next, we performed fluorescence quench experiments with pyrene-labeled WT and p.E334Q filaments, as it was previously shown that cofilin-1 efficiently quenches the fluorescence of pyrene-labeled actin filaments (Blanchoin & Pollard, 1999; De La Cruz, 2005). We determined the change in fluorescence amplitude when 2 µM pyrene-labeled p.E334Q or WT filaments were mixed with increasing concentrations of cofilin-1. Fitting a quadratic equation to the obtained experimental values gave only slightly different dissociation equilibrium constants of K_{D, p.E334Q} = 1.83 ± 0.13 µM and K_{D, WT} = 1.61 ± 0.07 µM (Figure 5D). These K_D values are in the range of values previously determined for muscle α-actin filaments (Blanchoin & Pollard, 1999; De La Cruz, 2005), actin filaments from Acanthamoeba castellanii (Blanchoin & Pollard, 1999) and filaments containing a mixture of human ψ– and β–actin (De La Cruz, 2005) using the same method.

To better understand the underlying changes leading to the differences in cofilin–mediated filament disassembly efficiency, we performed two–color TIRF microscopy experiments using Atto–655 labeled WT or mutant filaments and eGFP–cofilin–1. This setup allows us to observe the binding of eGFP– cofilin–1 to individual filaments and the resulting disassembly of the filaments in real–time. eGFP– cofilin–1 was shown to bind to actin filaments and disassemble actin as efficiently as WT cofilin–1 (Wioland et al, 2019, 2017; Breitsprecher et al, 2011). Immediately after mixing 300 nM “aged” actin filaments, capped at their barbed end by human capping protein (CP), with 100 nM eGFP–cofilin–1, the reaction mixture was flushed into a flow–chamber and imaging was started. The delay between mixing and the start of imaging was approximately 15 seconds for each sample. At the end of the experimental dead time, almost twice as many cofilin clusters had formed in the field of view on WT filaments (151 ± 29) as on p.E334Q filaments (82 ± 6) (Figure 6A), the number of both WT and mutant actin filaments had doubled from initially about 95 to about 200 filaments (Figure 6B), and the Atto– 655 fluorescence of the WT filaments had decreased to 30% of the starting value, whereas the Atto– 655 fluorescence of p.E334Q filaments remained unchanged (Figure 6C). During the following 500 s, total eGFP-cofilin-1 fluorescence increased linearly, with the rate of increase being approximately twice as fast for WT as for p.E334Q filaments (Figure 6D). While the total eGFP-cofilin-1 fluorescence in the presence of p.E334Q filaments continued to increase linearly over the next 1000 seconds, the signal for WT filaments began to saturate and reached a plateau value (Figure 6D). The number of cofilin clusters on WT filaments decreased steadily during the observation period of 1500 s. In contrast, their number increased in the presence of p.E334Q filaments, reaching a maximum after 500 s, followed by a plateau of 400 s and a subsequent slow decrease (Figure 6A). The changes in the number of actin filaments showed similar time dependencies, a steady decrease for WT filaments and an initial increase for p.E334Q filaments followed immediately by a linear decrease without an intervening plateau phase (Figure 6B). For WT and p.E334Q filaments, the size of eGFP–cofilin–1 clusters on the steadily shrinking filaments expanded continuously until either severing occurred or the actin filament was fully decorated. No severing events were observed with fully decorated filaments.

Figure 6

Impact of E334Q mutation on interactions with myosin family proteins

The precise nature of the six actin-binding elements in the myosin motor domain varies considerably among myosin isoforms (Sellers, 2000; Cope et al, 1996). The associated sequences have conserved hydrophobic and electrostatic properties, variations of which contribute to the tuning of kinetic and functional properties. Actin residue E334 interacts with a positively charged patch in the cardiomyopathy loop (CM–loop) of the myosin upper 50 kDa domain of NM2C and Myo5A (Pospich et al, 2021; Von Der Ecken et al, 2016). In the case of NM2C, the electrostatic interaction of E334 with K429 in the CM–loop is further stabilized by van der Waals interactions with I420 and V427 (Figure 7). A lysine residue corresponding to K429 in NM2C is present in all human myosin–2 and myosin–5 isoforms. During force-generating interactions with mutant actin filaments, the weakening of complementary electrostatic interactions in this region of the actomyosin interface is likely to attenuate the binding affinity of class 2 and 5 myosins as well as the coupling between their actin and nucleotide binding sites.

Figure 7

Cytoskeletal γ–actin filaments move with an average sliding velocity of 195.3 ± 5.0 nm s⁻¹ on lawns of surface immobilized NM2A–HMM molecules (Figure 8A, B). Filaments formed by p.E334Q actin move 5-fold slower, resulting in an observed average sliding velocity of 39.1 ± 3.2 nm/s. Filaments copolymerized from a 1:1 mixture of WT and p.E334Q actin move with an average sliding velocity of 131.2 ± 10 nm s⁻¹ (Figure 8A, B).

Figure 8

Figure 9

While we observed similar densities of surface-attached WT and mutant filaments, we noticed increased rapid dissociation and association events for p.E334Q filaments (Suppl. Video 7-9). This prompted us to investigate the impact of the mutation on actomyosin–affinity using transient–kinetic approaches. We determined the dissociation rate constants using a single–headed NM2A–2R construct (Figure 8C). NM2A–2R is a constitutively active motor–domain construct generated by fusing the motor–domain of NM2A to two spectrin–like repeats serving as a rigid lever arm extension (Anson et al, 1996). The construct is ideally suited for the kinetic characterization of the NM2–isoforms (Heissler & Manstein, 2011). The dissociation rate constant k_-A is 5.6–fold increased from 0.08 × 10^-3 s^{− 1} to 0.45 × 10^-3 s⁻¹ for NM2A–2R when p.E334Q filaments are used. A similar change from 0.31 × 10^-3 s⁻¹ to 1.70 × 10^-3 s⁻¹ was observed for the analogous NM2C–2R construct (Figure 8D). Next, we determined the affinities of NM2A–2R for WT and p.E334Q filaments in the absence of nucleotide (K_A) and the presence of saturating concentrations of ADP (K_DA). We performed affinity–titration experiments with pyrene– labeled, phalloidin–stabilized filaments and NM2A–2R (Kurzawa & Geeves, 1996). The values measured under rigor conditions show a 9.9-fold reduced affinity of NM2A–2R for p.E334Q filaments (K_A = 16.9 ± 1.3 nM) compared to WT filaments (K_A = 1.7 ± 0.3 nM) (Figure 8E). When the experiment was performed in the presence of saturating concentrations of ADP, we obtained values for K_DA of 353.0 ± 20.6 nM in the presence of p.E334Q filaments and 122.6 ± 11.4 nM with WT filaments (Figure 8F). The apparent ATP–affinity (1/K₁) of NM2A in complex with WT and p.E334Q filaments showed no significant differences (Table 1).

Next, we investigated the effects of the E334Q mutation on the motile activity of the processive motor Myo5A. Since the lysine residue corresponding to K429 in the CM loop of NM2C is also present in all human myosin 5 isoforms, the resulting changes in binding strength and thermodynamic coupling are expected to reduce the motor activity of Myo5A upon interactions with p.E334Q filaments. Structural studies have shown that the interaction between actin residue E334 and the CM-loop in the Myo5A motor domain is maintained from the rigor to the ADP–bound state (Pospich et al, 2021). We performed unloaded in vitro motility assays to evaluate the effect of the mutation on the interaction with Myo5A–HMM. The rapid transition from weak to strong binding with a slow and rate-limiting ADP release leading to a high working ratio (De La Cruz et al, 1999) allows the in vitro motility assay to be performed on surfaces with a low motor density. In initial experiments, we found a Myo5A–HMM surface density of 770 molecules µm⁻² to be optimal for experiments with WT filaments. Movement was steady and filament breaks were practically non–existent at 770 molecules µm⁻², while higher surface densities led to increased filament breaks, resulting in rapid fragmentation. At all Myo5A– HMM surface densities used, the sliding velocity was significantly reduced for p.E334Q compared to WT filaments. In experiments with p.E334Q filaments, a 4-fold higher Myo5A–HMM surface density was required to achieve a similar actin surface decoration and optimal motile activity. Furthermore, filament breaks occurred less frequently, indicating a less efficient actomyosin interaction. The number of moving filaments and their sliding velocity decreased sharply below surface densities of 770 molecules/µm² for WT filaments and 3100 molecules µm⁻² for p.E334Q filaments. Below a surface density of 380 Myo5A–HMM molecules µm⁻², no productive attachment events were observed with p.E334Q filaments. Thus, mutation E334Q in cytoskeletal ψ−actin leads to an impairment of the interaction of the actin filament with Myo5A, similar to the changes observed with NM2A and NM2C, but with additional manifestations specific to processive motors.

Discussion

At present, the exact pathogenesis of non-muscular actinopathies remains poorly understood. A rigorous molecular characterization of the contribution of disturbances in actin dynamics, contractility and transport processes is necessary to develop an understanding of the diverse disease patterns.

In the past, the functional characterization of disease–associated mutations in cytoskeletal ψ−actin focused on mutations p.T89I, p.K118N/M, p.E241K, p.T278I, p.P264L, p.P332A, and p.V370A, which all cause non–syndromic hearing loss. Two mutants showed abnormal polymerization behavior, indicated by a reduced polymerization rate (p.V370A) (Bryan et al, 2006) or an increased propensity for bundle formation (p.E241K) (Morín et al, 2009). The mutants p.K118N/M showed a perturbed interaction with the Arp2/3 complex, reducing branching efficiency (Kruth & Rubenstein, 2012; Jepsen et al, 2016). A significantly reduced protein stability was observed for two mutant proteins (p.P264L, p.P332A) (Bryan et al, 2006). Perturbed interactions with cofilin were observed for a range of mutations. p.T89I, p.E241K, p.T278I, and p.V370A are characterized by an increased sensitivity to cofilin–mediated filament severing (Bryan & Rubenstein, 2009). Interestingly, mutation p.P332A, close to the analyzed mutation p.E334Q, enhances resistance towards cofilin−mediated severing (Bryan & Rubenstein, 2009). The perturbed cofilin interaction is the only overlap between the spectrum of molecular phenotypes observed for syndromic deafness−associated mutations and mutation p.E334Q, which results in a non−syndromic actinopathy. It has to be noted that all cited studies relied on the use of actin purified from budding yeast for their biochemical studies, which shows 91% sequence identity to human cytoskeletal ψ−actin (Ng & Abelson, 1980).

Here, we describe mutation E334Q in cytoskeletal γ–actin that leads to a syndromic actinopathy with hypotonia and is not associated with non–syndromic hearing loss. Our biochemical analysis shows that the mutation does not affect the folding of the actin monomer or the stability of the monomer and the resulting filaments, demonstrating that mutations in the hinge region do not necessarily affect protein stability. The mutant protein displays a robust interaction with the VCA-activated Arp2/3 complex. Its polymerization kinetics are equivalent to those of the WT protein, with which it effectively copolymerizes in vitro. Based on these observations, we exclude functional haploinsufficiency as a potential disease mechanism and favor a mechanism in which p.E334Q integrates into the cytoskeletal actin networks of the cell but shows defects in the interaction with key binding partners. Consistent with this hypothesis and based on several structural studies (Pospich et al, 2021; Behrmann et al, 2012; Robert-Paganin et al, 2021; Risi et al, 2021; Fujii & Namba, 2017; Gyimesi et al, 2020; Gong et al, 2022; Gurel et al, 2017; Von Der Ecken et al, 2016) as well as our initial modeling of complexes formed by p.E334Q–filaments, we observed perturbed interactions with cofilin–1 and human cytoskeletal myosin isoforms NM2A, NM2C and Myo5A (Table 1).

We interpret the observed changes in the interaction with NM2A, NM2C and Myo5A to primarily reflect the weakening of complementary interactions between actin residue E334 and a core contact triad in the myosin CM–loop, which in the case of NM2C consists of K429 and hydrophobic residues I420 and V427 (Figure 7). Analogous changes in the interaction of mutant filaments with Myo1D, Myo1G, NM2B, Myo5B, Myo5C and Myo19 are expected due to the presence of the core contact triad residues in their CM–loops (Figure 7B). In contrast, the absence of a basic residue corresponding to K429 in NM2C and additional changes affecting their CM loop structure make it difficult to predict the consequences for the interaction of p.E334Q–filaments with the cytoskeletal myosin isoforms Myo1A, Myo1C, Myo1E, Myo1F, Myo3A, Myo3B, Myo15, Myo18A, Myo18B and Myo19.

In addition to primary consequences, reflected in the observed changes in the binding interactions with NM2A and cofilin–1, our results point to significant changes that result from allosteric perturbations. The results from our transient–kinetic analysis of the complexes formed by p.E334Q with NM2A–2R indicate that reduced allosteric coupling between the actin and nucleotide binding sites within the myosin motor domain contributes to the observed decrease in motor activity. The coupling efficiency is given by the ratio of the dissociation equilibrium constants for actin binding in the absence (K_A) and presence of ADP (K_DA). The value for K_DA/K_D obtained for the interaction of NM2A–2R in the presence of p.E334Q (20.9) is 3.5–fold lower than the value measured in the presence of WT actin (Table 1). Allosteric perturbations are also major contributors mediating the impact of mutation E334Q to the changes in elastic properties and the rate of actin filament disassembly in the presence of cofilin–1. Our results are consistent with the E334Q mutation slowing the initial complex formation due to less favorable charge interactions and allosteric perturbations having a major impact on subsequent steps.

Patients carrying mutations in ACTB or ACTG1 predominately suffer from developmental disabilities, which can be directly linked to cortical malformations of varying severity. Therefore, neuronal cells and/or neuronal progenitor cells are assumed to be the most affected cell population in the patients (Verloes et al, 2015; Di Donato et al, 2016; Rivière et al, 2012). The actin composition of the neuron is thought to reach ratios between 1:1 and 1:2 for ψ−actin to β−actin in the mature state after fluctuations during neuronal development (Cheever & Ervasti, 2013). Studies of the spatial localization of the cytoskeletal actin isoforms in the mature neuron are contradictory. A study showing that β− and ψ−actin strongly colocalize in hippocampal and cerebral neurons (Cheever et al, 2012) appears to contradict previous studies, which reported spatial enrichment of β−actin in dynamic structures such as the dendritic spine and exclusion from less dynamic structures such as the axon shaft, where ψ−actin was found to be enriched (Weinberger et al, 1996; Micheva et al, 1998; Bassell et al, 1998). Based on our results, we conclude that p.E334Q actin is readily produced and integrated into the cytoskeletal actin networks of neuronal cells, where it perturbs interactions with cytoskeletal myosin isoforms that play essential roles in the function and maintenance of healthy neurons. The functions of myosin-5 isoforms in neuronal cells are all linked to their role as actin barbed-end directed processive cargo transporters (Hammer & Wagner, 2013). Several of the myosin-5 driven transport-processes are important contributors to synaptic plasticity, i.e. the ability of the neurons to change synaptic strength based on previous activities (Hammer & Wagner, 2013).

The different NM2 isoforms are found in a wide range of cell types in the brain, but they show varying spatial expressions (Murakami & Elzinga, 1992; Golomb et al, 2004; Rochlin et al, 1995; Kawamoto & Adelstein, 1991). NM2A is predominately enriched in the vasculature of the brain, consistent with the importance of NM2A for angiogenesis (Ma et al, 2020). NM2B and NM2C are mainly found in growth cones and dendritic spines of neuronal cells (Murakami & Elzinga, 1992; Golomb et al, 2004; Rochlin et al, 1995). NM2 isoforms are known to contribute to various neuronal processes, but unlike for Myo5A, their exact role is less well understood. NM2B was shown to determine the maturation and morphology of the dendritic spines in hippocampal neurons and was additionally linked to basal synaptic function and plasticity (Hodges et al, 2011; Ma et al, 2004; Rex et al, 2010). NM2C and NM2B contribute to the differentiation of oligodendrocytes (Domingues et al, 2020). The NM2 isoforms compete with proteins of the ADF/cofilin family for binding to F–actin and this competition is essential for stable actomyosin contractility and an intact cell cortex (Wiggan et al, 2012).

In line with this, ADF/cofilin family members are highly abundant in growth cones and dendritic spines and essential for spine morphology and determination of spine length by modulating actin turnover (Star et al, 2002; Hotulainen et al, 2009; Namme et al, 2021). Cofilin-mediated severing and nucleation were previously proposed to synergistically contribute to global actin turnover in cells (Andrianantoandro & Pollard, 2006; Du & Frieden, 1998). Our results show that the mutation affects these different cofilin functions in actin dynamics in opposite ways. Cofilin-mediated filament nucleation is more efficient for p.E334Q monomers, while cofilin-mediated severing of filaments containing p.E334Q is significantly reduced. The interaction of both actin monomers and actin filaments with ADF/cofilin proteins involves several distinct overlapping reactions. In the case of actin filaments, cofilin binding is followed by structural modification of the filament, severing and depolymerizing the filament (De La Cruz & Sept, 2010). Cofilin binding to monomeric actin is followed by the closure of the nucleotide cleft and the formation of stabilized “long-pitch” actin dimers, which stimulate nucleation (Andrianantoandro & Pollard, 2006).

While the number of disease–associated mutations reported for ACTB and ACTG1 is steadily increasing, the total number of patients affected remains low compared to patients with disease–associated mutations in muscle actin isoforms (Parker et al, 2020). High sequence and structure conservation across all actin isoforms enable us to extrapolate our findings obtained with cytoskeletal γ–actin to diseases associated with mutations of E334 in other actin isoforms. Mutations of residue E334 in α– skeletal actin (encoded by ACTA1) are reported to result in nemaline myopathy (E334K, (Laing et al, 2009)) or core myopathy (E334A, (Kaindl et al, 2004)) (reviewed in (Parker et al, 2020)). The mutations either lead to the inversion of the charge at residue 334 (E334K) or the complete loss of charge (E334A). Our study shows that loss of charge at actin residue 334 leads to the loss of a conserved core contact between actin and the myosin motor domain of cytoskeletal class 2 myosin. Skeletal muscle myosin–2 is a central component of the sarcomere. The myosin forms the thick filaments that interact with the thin filaments formed by actin and regulatory proteins, generating force that powers muscle contraction (Sweeney & Hammers, 2018). Because the core contact triad that interacts with actin residue E334 is present in all class 2 myosin isoforms, it is reasonable to assume that mutations of residue E334 in α-skeletal actin affect the interaction with skeletal muscle myosin 2 to a similar extent, resulting in less efficient force generation and thereby contributing to the development of myopathy in patients.

Interestingly, a recent study analyzing the prevalence of cytoskeletal actin mutations in cancer using the cBioPortal database found that mutations in ACTB and ACTG1 are associated with lymphoid cancers, specifically mutations in ACTB with diffuse large B–cell lymphoma and mutations in ACTG1 with multiple myeloma. When the authors extended the analysis to all cancer types, they found that the observed mutations in the two genes were scattered across the entire actin sequence, with no apparent preference for a particular region. Nevertheless, the group identified residue E334 in ACTB as a potential hotspot, as the missense variants p.E334Q/p.E334K are more frequent in all analyzed datasets (Witjes et al, 2020). In this paper, we investigated the impact of the E334Q exchange on the function of cytoskeletal γ–actin in the context of developmental diseases. However, the high sequence similarity between the cytoskeletal actin isoforms makes it highly likely that the E334Q mutation in cytoskeletal β–actin results in similar molecular perturbations of actin dynamics but with potentially different consequences for the organism due to varying enrichment of the two isoforms in functional compartments of the actin cytoskeleton. Although future studies are needed to determine the role of ACTB and ACTG1 mutations as driver or passenger mutations in different cancers, our data show a possible link between impaired cytoskeletal actomyosin dynamics, cofilin–mediated actin turnover and cancer development.

Material and Methods

Plasmids

The coding sequence of human cytoskeletal γ–actin (Uniprot-ID: P63261) was fused via a C–terminal linker to a His₆–tagged thymosin β4 moiety (Uniprot-ID: P62328), as previously described for yeast actin (Uniprot-ID: P60010) (Noguchi et al, 2007). The fusion protein was cloned into the multiple cloning site of the pFastBac–Dual vector under the control of the polyhedrin promotor. The plasmid encoding the mutant protein p.E334Q was generated by performing site-directed mutagenesis on the WT plasmid. Motor domain carrying fragments of human NM2A, NM2C, and Myo5A were produced using MYH9, MYH14, and MYO5A heavy chain sequences, respectively (Uniprot-IDs: P35579, Q7Z406, Q9Y4I1). For the production of HMM-like constructs of human NM2A and human Myo5A, expression vectors derived from pFastBac were used, as described previously (Hundt et al, 2016; Reindl et al, 2022). NM2A–HMM was co–produced with human light chains MYL6a (Uniprot-ID: P60660) and MYL12b (Uniprot-ID: O14950). Myo5A HMM was co–produced with human calmodulin (Uniprot-ID: P0DP23). NM2A–2R and NM2C–2R, single–headed, constitutively active motor domain constructs of human NM2A and NM2C–2C0 with an artificial lever arm region, were produced using pFastBac– derived expression vectors, as described previously (Heissler & Manstein, 2011; Müller et al, 2013). Human cofilin–1 (Uniprot-ID: P23528) was cloned into the expression vector pGEX–6P–2 (Greve et al, 2022). The human Arp2/3 complex consisting of subunits Arp2 (Uniprot-ID: P61160), Arp3 (Uniprot-ID: P61158), ArpC1B (Uniprot-ID: O15143, ArpC2 (Uniprot-ID: O15144), ArpC3A (Uniprot-ID: O15145), ArpC4 (Uniprot-ID: P59998), ArpC5 (Uniprot-ID: O15511) and a C–terminal FLAG tag on ArpC3A was produced using the expression vector pBIG2abc, provided by Roberto Dominguez (University of Pennsylvania, Philadelphia, USA). The VCA domain of human N–WASP (Uniprot-ID: O00401) was cloned into pGEX–6P–2. Human capping protein (subunits α1/β2, Uniprot-ID: P52907/P47756) in pET– Duet–1 and eGFP–tagged human cofilin–1 (Uniprot-ID: P23528) in pGEX–6P–2 vector were provided by Jan Faix (Medizinische Hochschule Hannover, Hannover, Germany). Recombinant bacmids and viruses for the production of proteins in the Sf9 system were produced according to the Bac–to–Bac baculovirus expression system protocol (Thermo Fisher Scientific, Waltham, MA, USA).

Protein production and purification

We produced recombinant human cytoskeletal γ–actin and p.E334Q mutant protein as actin–thymosin β4–His₆ fusion protein using the baculovirus/S. frugiperda insect cell expression system. The C-terminal native actin residue is a phenylalanine, which provides a cleavage site for chymotrypsin to remove downstream tag sequences completely. The tag sequence employed consists of a 14-amino acid linker (ASSGGSGSGGSGGA), 43 thymosin-β4 residues, followed by 6 histidine residues. To produce WT and mutant fusion proteins, 2×10⁹ cells were harvested 72 hours after infection. Cells were pelleted, washed with PBS, shock frozen in liquid nitrogen and stored at –80 °C. Cell pellets from two liters of Sf–9 culture were thawed for a typical preparation. For lysis, cells were resuspended in lysis buffer–A (10 mM Tris pH 7.8, 5 mM CaCl₂, 10 mM imidazole, 1.25% Triton X–100, 100 mM KCl, 7 mM β–mercaptoethanol, 1 mM ATP, 1 mM PMSF, 100 μg/mL TAME, 80 μg/mL TPCK, 2 μg/mL pepstatin, 5 μg/mL leupeptin) and sonicated. The lysate was cleared by centrifugation at 30,000 × g (30 min, 4°C). The cleared lysate was loaded onto 3 mL equilibrated PureCube 100 NiNTA agarose (Cube Biotech, Manheim am Rhein, Germany) and incubated for two hours under constant rotation. Afterwards, the material was washed extensively with 10 mM Tris pH 7.8, 0.2 mM CaCl₂, 10 mM imidazole, 50 mM KCl, 1 mM ATP and the actin–thymosin β4 fusion protein was eluted with 10 mM Tris pH 7.8, 250 mM imidazole, 0.2 mM CaCl₂, 1 mM ATP. Peak fractions were pooled and dialyzed overnight against G– buffer (10 mM Tris pH 7.8, 0.2 mM CaCl₂, 0.1 mM ATP, 0.5 mM DTT). The next day, protein aggregates were removed by centrifugation at 15,000 × g (20 min, 4 °C). The supernatant was pooled and α– chymotrypsin (Sigma-Aldrich, Schnelldorf, Germany) was added at a w/w ratio of 1:300 (chymotrypsin:actin) to remove the tag-sequences. The optimal digestion time varied from 45 to 80 min and needs to be determined for each batch of α–chymotrypsin. PMSF was added to a final concentration of 0.2 mM to stop the digestion. Actin was separated from the thymosin-His tag using polymerization-depolymerization steps. Unless otherwise stated, all protein purification steps were carried out at 4°C. F-actin was stored on ice. G-actin was concentrated to approximately 150 µM, flash frozen in liquid nitrogen and stored at –80°C in G–buffer. Clarified G-actin, free of aggregates and oligomers, was always prepared fresh. G-actin stocks were thawed, diluted in G-buffer to the appropriate concentration, and centrifuged at 136,000 × g (15 min, 4°C). The top 2/3 of the centrifuged sample was carefully removed for use in further experiments. All myosin constructs were produced in the baculovirus/S. frugiperda insect cell expression system and purified as previously described for NM2A–HMM (Greve et al, 2022; Hundt et al, 2016; Pathan-Chhatbar et al, 2018), Myo5A–HMM (Reindl et al, 2022), the single–headed, constitutively active constructs NM2A–2R (Heissler & Manstein, 2011; Müller et al, 2013) and NM2C–2R (Heissler & Manstein, 2011; Müller et al, 2013). Production of the human Arp2/3 complex in insect cells followed the procedure described in Zimmet et al. (Zimmet et al, 2020). Human cofilin–1 and human eGFP–cofilin–1 were produced as GST–fusion proteins in Rosetta2 (DE3) cells. Purification was carried out as previously described for human cofilin–1 (Greve et al, 2022).

Human GST–N–WASP–VCA was produced in ArcticExpress (DE3) cells. The cells were lysed by treatment with lysozyme followed by sonification in lysis buffer–W (25 mM Tris pH 7.4, 50 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.5% Triton X–100, 1 mM PMSF). The cleared cell lysate was loaded onto a self–packed GSH–Sepharose column. The column was then washed with five column volumes of wash buffer–W (25 mM TRIS pH 7.4, 50 mM NaCl, 1 mM EDTA, 1 mM DTT). GST–N–WASP–VCA was eluted from the column by applying 10 mM glutathione in wash buffer–W. The eluted protein was concentrated and loaded onto a S75 16/600 size exclusion chromatography column (GE Healthcare, Chicago, USA). After chromatography, fractions containing pure protein were pooled, concentrated and the concentration was determined via absorbance at 280 nm. The GST–N–WASP–VCA protein was frozen in liquid nitrogen and stored at – 80°C.

Human capping protein (α1/β2) carrying a 6×His–tag was produced in Rosetta2 (DE3) cells. For purification, cells were lysed by lysozyme followed by sonification in lysis buffer-CP (25 mM Tris pH 7.8, 250 mM NaCl, 1 mM EDTA, 1 mM PMSF, 1.5 % Triton X–100, 10 mM imidazole, 7 mM beta– mercaptoethanol). The cleared lysate was loaded onto PureCube 100 NiNTA agarose (Cube Biotech, Monheim am Rhein, Germany) in a gravity–flow column. The agarose was washed with wash buffer– CP (25 mM Tris pH 7.8, 250 mM NaCl, 30 mM imidazole) and the protein was then eluted with elution buffer–CP (25 mM Tris pH 7.8, 50 mM NaCl, 250 mM imidazole). Fractions with pure capping protein were pooled, concentrated and loaded onto a S200 16/600 size exclusion chromatography column (GE Healthcare, Chicago, USA). The purified protein was frozen in liquid nitrogen and stored at –80°C.

α–skeletal actin (Uniprot-ID: P68139) was prepared from chicken pectoralis major muscle, as previously described for rabbit α–skeletal actin (Lehrer & Kerwar, 1972).

Unless otherwise stated, all purification steps were performed at 4°C.

DNase–I inhibition assay, DSF and nucleotide exchange assay

The DNase–I inhibition assay and DSF to study actin folding and thermal stability were performed as previously described (Greve et al, 2022).

Nucleotide exchange experiments were performed with freshly clarified G–actin as follows. The G– actin solution was passed through a Zeba™ spin desalting column (Thermo Fisher Scientific, Waltham, MA, USA) equilibrated with G–buffer to remove excess nucleotide. The sample was then incubated with a 2–fold excess of ε–ATP or ε–ADP (Jena Bioscience, Jena, Germany) for two hours on ice. After incubation, excess nucleotide was removed by passing the sample through a Zeba™ spin desalting column equilibrated with 10 mM TRIS pH 8.0 and 10 µM ε–ATP/ ε–ADP. The concentration of the Ca²⁺– ε–ATP/ε–ADP–G–actin solution was determined using the Coomassie protein assay (Thermo Fisher Scientific, Waltham, MA, USA). Prior to the nucleotide exchange experiment the Ca²⁺–ε–ATP/ε–ADP– G–actin solution was supplemented with 10× Mg²⁺-exchange buffer (1 mM MgCl₂, 10 mM EGTA) to a final concentration of 0.1 mM MgCl₂ and 1 mM EGTA. The resulting solution was rapidly mixed 1:1 with assay buffer–SF containing excess unlabeled ATP (10 mM TRIS pH 7.8, 0.2 mM MgCl₂, 0.2 mM ATP) using a HiTech Scientific SF61 stopped–flow system (TgK Scientific Limited, Bradford on Avon, UK). The ε–ATP/ε–ADP fluorescence was excited at 335 nm and monitored through a KV389 cut–off filter. The displacement of the fluorescent nucleotide analogs by ATP results in a decrease in the fluorescence amplitude, which is best described by a single exponential fit that yields the apparent rate of nucleotide dissociation

CD spectroscopy

Directly prior to the measurement, the freshly clarified G-actin was diluted to 0.05 mg/mL in 10 mM phosphate buffer pH 7.4. A buffer reference was prepared by mixing G buffer with 10 mM phosphate buffer pH 7.4 in the same ratio. CD spectroscopy was performed with a PiStar 180 spectrometer (Applied Photophysics, Leatherhead, UK). Measurements were performed at 25°C using quartz cuvettes (0.3 cm path length) and nitrogen purging. CD spectra were recorded from 176 nm to 280 nm in steps of 1 nm using a bandwidth of 4 nm. For every sample, three spectra were recorded and subsequently averaged. The averaged buffer spectrum was subtracted, and the resulting protein spectrum smoothed using a Savitsky–Golay filter (Savitzky & Golay, 1964) in two consecutive iterations. The DichroWeb online platform was used to determine the secondary structure composition (Whitmore & Wallace, 2004). Data were analyzed based on the mean residue ellipticity [𝜃]_MRE of the sample using the CDSSTR algorithm and reference set 3 (optimized for 185 nm–240 nm).

Steady–state ATPase assay

The steady–state ATPase activity of F–actin was measured using a NADH–coupled enzymatic assay (Furch et al, 1998). Experiments were performed with 10 µM F–actin at 37 °C in assay buffer–A (25 mM HEPES pH 7.3, 5 mM MgCl₂, 50 mM KCl, 0.5 mM DTT) supplemented with the enzymatic system, which generates the quantifiable readout (1 mM ATP, 0.5 mM NADH, 0.5 mM 2–phosphoenolpyruvate, 0.02 mg/mL lactate dehydrogenase from rabbit muscle (Sigma-Aldrich, Schnelldorf, Germany), 0.05 mg/mL pyruvate kinase from rabbit muscle (Sigma-Aldrich, Schnelldorf, Germany), 0.5 mM NADH). The change in absorbance at 340 nm due to oxidation of NADH was recorded in a Multiskan FC Microplate Photometer (Thermo Fisher Scientific, Waltham, MA, USA).

TIRF microscopy–based assays

To distinguish between the impact of the mutation on filament nucleation and elongation, we performed TIRF microscopy–based assays using freshly clarified protein. Unlabeled WT or mutant G– actin was supplemented with 10% Atto–655 labeled WT or mutant G–actin for visualization of the filament assembly process. Glass coverslips were cleaned, surfaces chemically treated and flow cells assembled as described previously (Greve et al, 2022). To induce polymerization, the G–actin solutions were diluted to a final concentration of 1 µM with 2× TIRF–buffer (20 mM imidazole, 50 mM KCl, 1 mM MgCl₂, 1 mM EGTA, 0.2 mM ATP, 15 mM glucose, 20 mM β–ME, 0.25 % methylcellulose, 0.1 mg/mL glucose oxidase and 0.02 mg/mL catalase. The solutions were mixed, immediately flushed into flow cells and image acquisition was started.

To follow filament disassembly, two–color TIRF experiments with Atto–655 labeled actin and eGFP– cofilin–1 were performed with preformed filaments that were “aged” and capped at their barbed end by human capping protein. WT or mutant actin was polymerized for 3 hours at room temperature and then moved to 4 °C overnight. “Aged” F–actin and human capping protein were diluted into TIRF– buffer to generate capped actin filaments. eGFP–cofilin–1 was added to final concentrations of 300 nM F–actin, 25 nM capping protein and 100 nM eGFP–cofilin–1. Immediately adding eGFP–cofilin–1, the reaction mixture was flushed into the flow cell and image acquisition was started. Images were captured using a Nikon Eclipse TI–E inverted microscope equipped with a 100×/1.49 NA Apo oil immersion objective and Ixon3 897 EMCCD cameras (Andor, Belfast, UK) under control of the NIS software (Nikon, Minato, Japan). Images were taken every 3 seconds with an exposure time of 60 ms.

To analyze actin polymerization in the presence of ABPs, the ABPs were first pre–diluted in KMEI buffer and diluted to their final concentration in TIRF buffer just before adding actin. Image acquisition was performed as described for experiments with actin only. Image analysis for all TIRF microscopy experiments was performed using ImageJ (Rueden et al, 2017). The elongation rate of actin filaments was determined by manually tracking individual filaments. Only filaments that could be tracked for at least four consecutive minutes were used to determine the elongation rate. Using these criteria, 20 to 30 filaments were tracked in each individual experiment (in a microscopic field of view of 81.41 × 81.41 µm). For quantification of two–color TIRFM–experiments, 52 × 52 µm ROIs from the entire microscopic field of view were used. The number of filaments observed throughout an experiment was determined using the Analyze Particle tool in ImageJ after background–subtraction (sliding paraboloid, rolling ball radius: 25 pixels).

Unloaded in vitro motility assay

The unloaded in vitro motility assay to study the interaction between WT and mutant actin and NM2A– HMM or Myo5A–HMM was essentially performed as previously described (Greve et al, 2022; Reindl et al, 2022; Kron & Spudich, 1986). Flow cells used in the in vitro motility assay were constructed using nitrocellulose–coated coverslips. NM2A–HMM requires phosphorylation of the associated regulatory light chain for full enzymatic activity. Purified NM2A–HMM was incubated with the GST–tagged kinase domain (residues 1425-1776) of human smooth muscle myosin light chain kinase (Uniprot-ID: A0A8I5KU53) at a molar ratio of 10:1 in 25 mM MOPS, pH 7.3, 50 mM KCl, 5 mM MgCl₂, 1 mM CaCl₂, 0.2 µM calmodulin, 3 µM regulatory light chain (MLC12b), 3 µM essential light chain (MLC6a), 1 mM DTT and 1 mM ATP for 30 minutes at 30 °C. In experiments with Myo5A–HMM, each buffer and actin solution was supplemented with 2 μM calmodulin to prevent calmodulin dissociation from Myo5A– HMM. Olympus IX70 or Olympus IX83 inverted fluorescence microscopes (Olympus, Hamburg, Germany) equipped with a 60×/1.49 NA PlanApo oil immersion objective and an Orca Flash 4.0 CMOS camera (Hamamatsu Photonics Deutschland GmbH, Herrsching, Germany) were used for image acquisition. All in vitro motility experiments were performed at a 37°C. Analysis of the recorded image series was performed in ImageJ (Rueden et al, 2017). The Plugin wrmTrck (Nussbaum-Krammer et al, 2015), developed by Jesper S. Pedersen, was used to determine the trajectories and corresponding velocities of the individual actin filaments. Only filaments that showed movement for at least 30 consecutive seconds were tracked and used to determine the average sliding velocity.

Pyrene–actin based bulk–polymerization assays

The polymerization and depolymerization of pyrene–labeled recombinant WT or mutant actin (5% pyrene-labeled WT or mutant actin) was monitored as a function of increasing or decreasing pyrene– fluorescence using a Synergy 4™ microplate reader (BioTek Instruments, Winooski, VT, USA) and the built–in filter set (Excitation: 340/30 nm, Emission: 400/30 nm) at 25 °C. For polymerization experiments, 20 µL of a 10 µM solution of pyrene–labeled Mg²⁺–ATP–G–actin (WT or mutant) was placed in a black flat bottom 96–well plate (BrandTech Scientific, USA). 80 µL of 1.25 × polymerization buffer (10 mM TRIS, pH 7.8, 100 mM KCl, 5 mM MgCl₂, 0.5 mM EGTA, 0.1 mM DTT and 0.1 mM ATP) was added to the wells using the built–in pipetting function of the plate reader and the polymerization of actin was followed as an increase in pyrene–fluorescence. For depolymerization experiments, 3 µL of a 20 µM F–actin solution was placed in one well of the plate. The solution was rapidly diluted by the addition of 297 µL G–buffer. The depolymerization of the filaments was monitored as a decrease in pyrene–fluorescence.

The effects of Arp2/3 (+ GST–N–WASP–VCA) and cofilin–1 on filament assembly or disassembly were analyzed by dispensing the actin solution and the ABP solution into the well by hand without mixing.

Polymerization buffer or G–buffer was added, the reaction components were thereby mixed and recording of the ensuing changes in fluorescence intensity was started.

The bulk–polymerization rate was determined according to Doolittle et al. (Doolittle et al, 2013). A linear regression was fitted to the linear region around the time–point of half–maximal fluorescence. The depolymerization experiments were evaluated by fitting a single exponential function to the data, including a linear component that accounts for bleaching.

Co–sedimentation experiments

Actin was polymerized at 20 μM for 3 hours at room temperature before being moved to 4 °C overnight. For co–sedimentation experiments at pH 7.8, 5 μM of “aged” F–actin and varying concentrations of human cofilin–1 were mixed in assay buffer–C (10 mM TRIS, pH 7.8, 100 mM KCl, 5 mM MgCl₂, 0.5 mM EGTA, 0.1 mM DTT and 0.1 mM ATP) to a final volume of 50 μL in 1.5 mL polypropylene centrifuge tubes (Beckman Coulter, Brea, CA, USA) suited for high–speed centrifugation. For experiments performed at pH 6.5, assay buffer–C was supplemented with piperazine–N,Nʹ–bis(2– ethanesulfonic acid) (PIPES, pH 6.5) to a final concentration of 100 mM to induce a pH shift from 7.8 to 6.5. Actin and cofilin–1 were added after the pH shift. Samples were incubated for 10 minutes at 23°C and then centrifuged at 136,000 × g (15 min, 4°C). The pellet and supernatant fractions were subjected to SDS–PAGE and the gel was imaged at a ChemiDoc–MP–gel documentation system (Bio– Rad Laboratories, Hercule, CA, USA). The amount of actin and cofilin–1 in the pellet and supernatant was determined via densitometry using the Image Lab software (Bio–Rad Laboratories, Hercule, CA, USA).

Transient kinetic analysis of the actomyosin complex

Myosin binding to pyrene–labeled actin leads to strong pyrene–fluorescence quenching. Mixing of the rigor–pyrene–acto–NM2A–2R/NM2C–2R complex with an excess of unlabeled actin allows the determination of k_-A (dissociation rate of myosin from F–actin in the absence of nucleotide). Prior to the experiment, pyrene–labeled phalloidin–stabilized WT or mutant F–actin and the myosin solutions were treated with apyrase (Merck KGaA, Darmstadt, Germany) to remove residual nucleotide. Pyrene– labeled actin was mixed with a 1.5–fold excess of NM2A–2R/NM2C–2R in assay buffer–P (20 mM MOPS pH 7.0, 100 mM KCl, 5 mM MgCl₂, 5 mM DTT) and incubated for 5 minutes on ice. The solution was mixed with an excess of chicken α–skeletal actin in assay buffer–P to final concentrations of 0.5 μM pyrene–labeled actin, 0.75 μM NM2A–2R and 5 μM α–skeletal actin. The solution was immediately transferred into the well of a black 96–well microplate with a transparent bottom and a lid (Corning, Corning, NY, USA). Fluorescence was excited at 365 nm and measured at 407 nm in a CLARIOstar Plus microplate reader (BMG LABTECH, Offenburg, Germany) using bottom optics over 9 hours. The dissociation rate (k_-A) of myosin from actin was determined by applying a single–exponential fit to the data.

The apparent affinity of NM2A–2R to WT or mutant F–actin in the absence (K_A) or presence (K_DA) of nucleotide was determined using the method developed by Kurzawa and Geeves (Kurzawa & Geeves, 1996). To determine K_A, all protein samples were treated with apyrase before the experiment. 100 nM pyrene–labeled phalloidin–stabilized F–actin was incubated with increasing concentrations of NM2A– 2R (10–1,300 nM) for 5 minutes on ice in assay buffer–P without DTT. The solutions were shot against 20 µM ATP at a HiTech Scientific SF61 stopped–flow system (TgK Scientific Limited, Bradford on Avon, UK) in a 1:1 ratio at 20 °C. Fluorescence was excited at 365 nm and recorded using a KV389 cut–off filter. At least five transients were averaged for each concentration of NM2A–2R and fitted using a single–exponential function. The relative fluorescence amplitudes were plotted against the pre–mix myosin concentrations. The affinity was determined by applying a quadratic equation to the dataset, where A is the amplitude, A_max the amplitude under saturation, [F–actin] the actin concentration prior mix, [NM2A–2R] the variable NM2A–2R concentration and K the dissociation constant in the absence or presence of nucleotide.

Slight modifications of the experimental protocol were necessary to determine K_DA. Incubation with apyrase was omitted, instead the pyrene–actomyosin complex was incubated with 30 µM ADP (Merck KGaA, Darmstadt, Germany) for 20 minutes on ice prior to measurement. To account for the lower affinity in the presence of ADP, the range of used myosin concentrations was extended up to 3,000 nM. Additionally, the concentration of ATP in the mixing experiment was increased to 300 µM pre– mix. Data was recorded and evaluated as done for experiments without nucleotide.

The apparent ATP–affinity of the actomyosin complex was determined by mixing the apyrase–treated, pyrene–labeled, phalloidin–stabilized actomyosin complex with increasing concentrations of ATP at the stopped–flow system. Fitting an exponential function to the individual transients yields the ATP– dependent dissociation rate of NM2A–2R from F–actin (k_obs). The k_obs–values were plotted against the corresponding ATP concentrations and a hyperbola was fitted to the data. The fit yields the apparent ATP affinity (1/K₁) of the actomyosin complex and the maximal dissociation rate.

Molecular modelling

Homology models of human cytoskeletal γ–actin, human cofilin–1 and the human NM2A motor domain were generated with MODELLER (Sali & Blundell, 1993). Models of filamentous human cytoskeletal γ–actin with bound cofilin–1 or NM2A motor domain were generated based on cryo–EM structures obtained from the Protein Data Bank of the barbed end side of a cofilactin cluster (pdb: 6UC4, 6UBY, 6VAO) (Huehn et al, 2020) and a human cytoplasmic actomyosin complex (pdb: 5JLH) (Von Der Ecken et al, 2016). The structures were accessed on May 25, 2023. Mutation E334Q was introduced using the mutagenesis tool in PyMOL (Schrödinger, LLC, 2015). PyMOL and ChimeraX (Goddard et al, 2018; Pettersen et al, 2021) were used for distance measurements and visualization of Coulomb surfaces.

Identification and alignment of the CM–loop region of cytoskeletal myosin isoforms were performed using the match function of ChimeraX, by superimposing and aligning cytoskeletal myosin isoform structures on NM2C (pdb-NM2C: 5JLH, pdb-Myo5A: 7PM5, pdb-Myo5C: 5HMP, pdb-Myo1D: AF-O94832-F1, pdb-Myo15: 7UDT, pdb-Myo19: AF-Q96H55-F1). Where appropriate, the corresponding human sequences were retrieved from UniProt and structurally aligned.

Normal mode analyses (NMA) were performed using the Elastic Network Model (ENM) elNémo (Suhre & Sanejouand, 2004) and an nrbl value of 5.

Data Analysis

Data analysis and graph plotting were performed with Origin 2023 (OriginLab Corporation. Massachusetts, USA). Errors are given as standard error of the mean (SEM) based on three independent experiments if not otherwise specified. The significance of the data was evaluated in Origin 2023 using a two–sample t–test (p > 0.05 ≙ ns, p ≤ 0.05 ≙ *, p ≤ 0.01 ≙ **, p ≤ 0.001 ≙ ***, p ≤ 0.0001 ≙ ****).

Author Contributions

All authors discussed the results and contributed to the final manuscript; J.N.G., T.R., C.T. purified proteins; J.N.G. performed experiments; A.M. and R.H. performed in silico studies; J.N.G. and D.J.M. analyzed the data; J.N.G., M.H.T., N.D.D. and D.J.M. contributed to the experimental design; J.N.G., A.M., R.H. and D.J.M. designed figures; J.N.G. and D.J.M. conceived and coordinated the study, and wrote the manuscript; D.J.M. was responsible for funding acquisition, supervision and project administration.

Acknowledgements

The authors thank Roberto Dominguez (University of Pennsylvania, Philadelphia, USA) for providing vector pBIGBac–Arp2/3 and Jan Faix (Medizinische Hochschule Hannover, Hannover, Germany) for providing vectors pGEX–eGFP–cofilin–1 and pETDuet–CP.

D.J.M was supported by grants from Deutsche Forschungsgemeinschaft (DFG) (MA1081/23-1, MA1081/28-1), J.N.G., A.M., N.DD and D.J.M. are members of the European Union’s Horizon 2020 research and innovation program under the EJP RD COFUND–EJP N° 825575 with support from the German Federal Ministry of Education and Research under Grant Agreement 01GM1922B. D.J.M. is a member of the Cluster of Excellence RESIST (EXC 2155; DFG–Project ID: 39087428–B11). J.N.G. is supported by the HiLF I grant for early career researcher from Medizinische Hochschule Hannover. We gratefully acknowledge support provided by the Core Unit for Structural Biochemistry, the Core Unit for Laser Microscopy at MHH. Computing time was provided on supercomputers Lise and Emmy at NHR@ZIB and NHR@Göttingen, as part of the Alliance for National High Performance Computing (NHR) infrastructure. The calculations for this research were conducted with computing resources under the project ID nib00018. T.R. was enrolled in the PhD program Molecular Medicine of Hannover Biomedical Research School (HBRS).

Supplementary Figures

Supplementary Figure 1

Supplementary Figure 2

Supplementary Figure 3

Supplementary Figure 4