The open to closed D-loop conformational switch determines length in filopodia-like actin bundles

Filopodia are composed of tightly bundled parallel F-actin filaments that form the core of thin finger-like protrusions from the plasma membrane. They have multiple roles in cell differentiation, migration, pathogen infection and environment sensing [1,2]. As exploratory organelles, the stochastic protrusion and retraction of filopodia results in structures of varying lengths that can reach from the cell body to the extracellular matrix, to other cells and across tissues in the case of cytonemes [3,4]. Filopodia lengths are typically up to 10 microns, while cytonemes can reach 10 times further. Growth or shrinkage rates are typically 1–2 µm/min and reaching ∼25 µm/min [5]. The underlying mechanisms determining filopodial length will involve balancing the rates of actin incorporation, bundling, severing and disassembly with lifetime of each phase [6]. What makes any given filopodium long or short remains an open question, yet is crucial for understanding the molecular basis of filopodial formation and how filopodia respond to intracellular and extracellular signals that guide cell sensing and movement.

Candidate determinants of filopodia length include F-actin conformational changes, as well as the influence of actin binding proteins formins [7–9], myosin-X [10,11], fascin [12,13], capping protein [14,15], and cofilin [16,17]. Reconstitutions of fascin-mediated bundles emerging from Arp2/3 complex-nucleated actin with either VASP or mDia2 showed that mDia2 gives longer filopodia than VASP [8]. While myosin-X overexpression results in elongated filopodia, this depends on the tail domain rather than the motor domain [11]. Phosphorylation site mutants of fascin affect filopodial lengths [12]. In native neuronal filopodia cofilin-bound actin filaments are present in longer filopodia and are more common towards the cell body, consistent with increased affinity of F-actin for cofilin that occurs as the actin filament ages [16–18]. When actin polymerizes, G-actin monomers are initially incorporated into filaments with ATP occupying the actin nucleotide binding pocket. Filament formation causes rapid ATP hydrolysis. In the ADP-Pi state, hydrolysis has occurred but inorganic phosphate remains within the nucleotide binding pocket before it is slowly released, eventually leaving older filaments in the less stable ADP-bound state [18–20]. Along a filament there is a nucleotide gradient which locally marks its age, and this shifts the conformation of actin subunits and mechanical properties of the filament [21–23], including at a flexible site on the filament surface, the DNase 1-binding loop (D-loop). Cofilin binds to ADP-actin filaments causing a twist in the filament that increases the depolymerization rate from the pointed end, severs and diassembles actin filaments [24–27]. Cofilin localises as cofilactin filaments at the base of long filopodia in neurons where it may or may not be the trigger for actin depolymerization [17].

To explore the molecular determinants of filopodial length regulation in an unbiased way we conducted a phenotypic antibody screen on a cell-free system of filopodia-like structures (FLS). FLS are nucleated on addition of Xenopus egg high-speed supernatant extracts (HSS) to PI(4,5)P₂-containing supported lipid bilayers [28]. FLS are comprised of fascin-bundled parallel actin filaments and recapitulate the growth and shrinkage dynamics seen in cellular filopodia [5]. We previously used FLS in a phenotypic screen as the antigen for phage display selections with a library of human single chain variable fragments (scFvs) to identify antibodies that specifically bound FLS. After incubation of the Xenopus egg extracts with the identified scFvs, FLS were grown and their morphological properties analysed, leading to the isolation of antibodies that cause longer and shorter FLS [29]. Nanobody reagents have been successfully isolated against fascin previously [30,31] and the difference with our FLS approach is the ability to identify novel antigens.

Here we have identified the antigen and epitope of several antibodies that make FLS longer and determined their mechanism of action. Using immunostaining, enzyme-linked immunosorbent assays and displacement of actin-binding proteins, we show that four independent antibody clones bind jasplakinolide-stabilized actin preferentially and displace LifeAct, suggesting that they recognise the open DNase1-binding loop (D-loop) conformation of actin filaments. By measuring FLS length and actin assembly/disassembly rates using photobleaching, we show that the disassembly rate is diminished by antibody binding while the assembly rate is unchanged. The antibodies’ effect on lengthening FLS can be overcome by increasing cofilin and actin concentrations in the extracts. We propose that a switch in actin filament D-loop conformation from the open to closed state within the actin bundle is an underlying molecular feature of length control in filopodia and related actin networks.

Incubation of HSS extracts with a supported lipid bilayer enriched in PI(4,5)P₂ yields the assembly of FLS that are readily followed by spinning disk confocal microscopy z-stacks followed by 3D reconstruction and image-based quantification using our pipeline FLSAce [5,28]. We converted the scFvs selected in our previous phage display phenotypic screen on FLS [29] into monomeric human IgG1s, expressing the bivalent full antibodies (HuIgG1s) which we preincubated with HSS before addition to the supported lipid bilayer. The FLS were allowed to grow and imaged after 25–35 min when the balance of growth and shrinkage results in a steady-state. Incubation with four antibodies: mAb2, mAb6, mAb14 and mAb17 resulted in FLS of significantly longer lengths, similar to their scFv counterparts, whereas the isotype control antibody R347 was similar to the untreated control and had no effect on FLS (Figure 1A,B).

As we previously observed, the lengths of unperturbed FLS in a stationary state of growth can be well-described by an exponential distribution, which is intrinsically linked to their growth dynamics [5]. Plotting these distributions on a log-scaled y-axis allowed the nature of the length change driven by the antibodies to be discerned (Figure 1C): in the unperturbed control condition, the number of FLS decreases sharply with increasing length, with no FLS longer than roughly 32 µm detectable; in contrast, the lengthening antibodies (mAb2, 6, 14, 17) instead increase the number of FLS at longer lengths, such that lengths are no longer exponentially distributed (Figure 1C). In reconstitutions of filopodia-like networks from purified proteins, high concentrations of formins also lengthened the resulting actin bundles changing the length distribution in a similar way [8].

The change in FLS length arose in a dose-responsive manner for each antibody, while there was no change with either isotype control antibody R347 and NIP228 (Figure 1C, Supplementary Figure S1). To begin to identify the antigen of the antibodies, we used immunostaining and Western blotting. While by immunostaining the four lengthening mAbs were capable of staining the full length of the actin shaft of untreated FLS (Figure 1D), no band was obtained using the antibodies for Western blotting of HSS (unpublished observations). No staining of the FLS shaft was observed with either the isotype control antibody NIP228 or mAb4 from the same phenotypic screen (Figure 1D, Supplementary Figure S2A), which we had previously identified as an anti-SNX9 antibody [29]. To test whether a similar staining pattern was observed in cells, we immunostained HEK293 cells together with Alexa-488 conjugated phalloidin for filamentous actin, seeing close correspondence, which was confirmed by controls (Figure 1E, Supplementary Figure S2B). Microinjection of mAb14 into one cell of Xenopus embryos at the two-cell stage led to delayed, absent or partial formation of the cytokinetic furrow in the injected cell compared with R347 control antibody, consistent with a disruption in F-actin (Figure 1F).

Next, we performed immunoprecipitation experiments with mAbs 2, 6, 14 and 17 in the presence and absence of actin monomer sequestering drug latrunculin B [32]. A distinct band of ∼45 kDa was isolated from HSS by each of mAb 2, 6, 14 and 17 and not the R347 isotype control, which was abolished by the latrunculin B treatment (Figure 2Ai, full gels are shown in Supplementary Figure S3). A subsequent Western blot using a commercial pan-actin antibody identified the immunoprecipitated protein as actin (Figure 2Aii). To determine whether purified actin was sufficient to bind the antibodies we performed sedimentation experiments with purified rabbit muscle actin under conditions of ∼50% G/F-actin, with a high speed sedimentation to separate F-actin filaments and low speed sedimentation to separate actin bundles (Figure 2B,C). Over 90% of the R347 control antibody remains in the supernatant with G-actin. In contrast, approximately half or more of mAb 2, 6, 14 and 17 sedimented in the pellet with F-actin filaments and F-actin bundles (Figure 2C). There was some increase in actin sedimenting in the pellet (e.g. mAbs 6 and 17), our conclusion from trial experiments is that it originates from variable disruption of fragile pellets (Figure 2B,C).

To confirm the antibody specificity for F-actin we designed an actin-down ELISA. Reconstituted rabbit skeletal muscle actin was split into two equal volumes: one half of which was prepared as monomeric G-actin, with the other half polymerized into F-actin filaments (Figure 2D). The two samples were prepared into a dilution series on an ELISA plate, and probed with mAbs 2, 6, 14 and 17, the R347 isotype control, and a pan-actin primary antibody that is able to detect both G and F-actin. As expected, R347 failed to bind, and we verified that the pan-actin control bound to G and F-actin similarly. This shows that the locally increased concentration achieved by clustering of actin monomers into filaments does not necessarily increase antibody binding by avidity effects. Each of the lengthening antibodies bound F-actin in preference to G-actin (Figure 2E, Supplementary Figure S3E). Quantification of the EC50 values from the binding curves suggested an ∼10-fold increase in affinity of mAb 2, 6, 14 and 17 towards F-actin relative to G-actin (Supplementary Figure S3F).

To test whether the antibodies were capable of displacing known actin binding proteins we used GFP-utrophin tandem CH1/CH2 domains (UtrCH) [33], which normally labels the full length of FLS, and observed that the probe was displaced from FLS in the presence of each of mAb 2, 6, 14 and 17 (Figure 3A, Supplementary Figure S4A). At the 1 µM antibody concentration, the GFP-UtrCH was restricted to a thin strip nearest the supported lipid bilayer (Figure 3A). The displacement of GFP-UtrCH and the increase in length of FLS, were similarly dose-responsive for each mAb (Figure 3A,B, Supplementary Figure S4B,C). To quantify the displacement, we measured the mean fluorescence contributed by all FLS over the field of view at each z-slice for both GFP-UtrCH and Alexa-568 actin channels, showing that the extent of GFP-UtrCH is reduced by higher concentrations of antibody and that only fluorescent actin allows the visualization of the longer FLS length in the presence of mAb14 (Figure 3C,D, Supplementary Figure S4A–D). Looking at individual FLS, the transition between UtrCH labelled and unlabelled regions was sharper than the averaged quantification, which was evocative of a switch-like transition within the actin bundle between the newly incorporated F-actin that could bind UtrCH and the older, extra portion of actin which could not (Figure 3E).

We noted that GFP-UtrCH binds to an area of F-actin including the D-loop [22,34,35]. To verify that the lengthening antibodies interact with the D-loop we tested a second probe, LifeAct, a 17 residue peptide from the yeast protein Abp140 [36], that has a smaller footprint on the F-actin filament more closely localized to the D-loop region (Figure 3F). Atto-655-LifeAct is displaced in the presence of either 1 µM mAb14, or mAb6 while the Atto-488-actin was similar to LifeAct labelling in FLS treated with the control mAb (Figure 3F,G, Supplementary Figure S4F). The degree of LifeAct displacement was exacerbated compared with GFP-UtrCH, in particular for mAb14, suggesting that the D-loop is the predominant site of antibody binding (Figure 3F,G).

We wondered how common it is for anti-actin antibodies to cause a lengthening of actin bundles and displacement of actin-binding probes. We tested a panel of monoclonal anti-actin antibodies for these properties finding that most do not lengthen FLS or displace GFP-UtrCH (Supplementary Figure S5). Antibody B, which was raised against residues 29–39 of human actin, adjacent to the D-loop, caused a slight lengthening effect at the low concentration attainable (Supplementary Figure S5). This suggests that the D-loop is an epitope specific to the FLS lengthening antibodies, and that D-loop binding causes increased FLS length.

The actin D-loop is a flexible region [20,37] and changes conformation as it is incorporated into filaments, where it is present at the surface of the filament and inserts into a hydrophobic cavity between the SD1 and SD3 domains of the neighbouring N⁺² subunit. There is evidence that the D-loop conformation changes during the nucleotide cycle, and that it can be stabilized in specific conformations by binding proteins (including LifeAct [38]) and small molecules, jasplakinolide and phalloidin [39,40], Figure 4A–C. The D-loop is a site of actin regulatory protein binding, with different effectors proposed to recognise different nucleotide and conformational states; for example, the depolymerization factor cofilin recognises ADP-actin in the closed conformation, whereas coronin strongly prefers the ADP-Pi state [21,22,27,41].

Phalloidin and jasplakinolide are two natural actin stabilizing compounds that compete for binding at the same site (away from the D-loop) within the actin filament [42,43]. They can be used to stabilize polymerized actin in distinct D-loop conformations allosterically (Figure 4A,B). Aged ADP F-actin is stabilized in the closed conformation by phalloidin, whereas jasplakinolide is able to switch the D-loop back to an open conformation (despite the absence of the released inorganic phosphate) [21,39,40,43]. We therefore modified our F-actin binding ELISA, using polymerized filaments stabilized with either jasplakinolide or phalloidin to determine the binding affinity of the antibodies to specifically open vs closed actin D-loop conformations. mAb14 bound to phalloidin-stabilized actin with 10-fold higher affinity than untreated F-actin and a 100-fold higher affinity for jasplakinolide-stablized filaments (Figure 4D). In contrast, the pan-actin control antibody which recognises F- and G-actin with the same specificity showed no such preference, and in fact treatment with either inhibitor compromised binding (Figure 4B). mAbs 2, 6 and 17 all showed a similar result to mAb14, with in each case a ∼10-fold higher EC50 towards jasplakinolide over phalloidin-treated F-actin and 100-fold over untreated F-actin (Supplementary Figure S6). We wondered whether the jasplakinolide or phalloidin stabilization of filaments would affect FLS length similar to the the antibodies. Treatment with either compound resulted in a modest decrease in FLS length and fuzzy morphology consistent with native actin filament conformations being an important aspect of correct FLS morphology (Supplementary Figure S7).

The lengthening effect of the antibodies on FLS implies that their interaction with the open D-loop conformation will act to either increase actin assembly or decrease disassembly. To test this, FLS were initiated and allowed to grow for either 5 min (to capture structures in the initial growing phase), or 25 min (when FLS have typically reached steady-state with growth balanced by shrinkage). The fluorescently-labelled actin in the entire field of view was then bleached and imaged every minute for the next 10 min to observe the reincorporation of bright freshly-added actin assembling at FLS tips and disassembly of faintly-fluorescent bleached FLS (Figure 5A).

The initial rate of FLS length increase in the first 60 s after bleaching was comparable between R347 and mAbs 14, after which the rates diverged, with mAb14 treatment maintaining a rate similar to the initial period (Figure 5B). This suggests there was no change in the rate of actin polymerization on antibody treatment. The lengthening antibodies instead displayed strongly reduced disassembly, with the bleached regions remaining evident over the course of the imaging period for each of mAbs 14, 2 and 6 (Figure 5C, Supplementary Figure S8A, Supplementary Movies S1 and S2), explaining the observed increasing divergence in FLS length over time (Figure 5B). The inhibition of disassembly was even more apparent when imaging at steady state (25–35 min), where the bleached actin disassembled with R347 and persisted with mAbs 14, 2 and 6. At this time point, the rate of freshly incorporated actin remained comparable to addition during the growth phase in the R347 treated FLS and could be slower with the lengthening antibodies (Figure 5C, Supplementary Figure S8B, Supplementary Movies S1 and S2), likely due to the reduced availability of fresh monomers. Thus, a dynamic analysis of actin incorporation rates demonstrates that the lengthening effect of the anti-D-loop antibodies is due to the protection of FLS from disassembly.

We last tested whether either actin (as antigen) or cofilin (which can disassemble F-actin) could suppress either the lengthening or the GFP-UtrCH displacement effects of mAb14. We prepared FLS with supplemental actin without pre-spinning the resuspended aliquots to provide some F-actin as well as extra G-actin and added purified cofilin, and compared the effect in the presence and absence of the lengthening antibodies. With the R347 control antibody, adding supplemental actin or cofilin did not displace GFP-UtrCH from FLS (Figure 6A) and increased actin concentrations caused lengthening of FLS (Figure 6A,B). In contrast, the addition of actin alone in the presence of mAb14 reduced the length of FLS and increased the degree of UtrCH-actin overlap, consistent with actin binding to mAb14 leading to a lowered effective antibody dose. Adding supplemental cofilin decreased FLS lengths while the displacement of GFP-UtrCH was maintained (Figure 6A,C). Supplemental actin in combination with cofilin had no additional effect on the length phenotype to cofilin alone, but as with the other instances of actin addition it restored ∼50% of the fluorescent actin — GFP-UtrCH overlap (Figure 6D,E). Overall our data reveal that FLS are robustly lengthened when a specific conformation of F-actin is stabilized (the best candidate being an open D-loop) or the D-loop site is physically occupied by antibodies, which acts to prevent prevent disassembly of the actin bundle. The stabilization can be countered by supplementing the extracts with additional cofilin.

We have identified a unique set of antibody reagents that provide insight into how F-actin conformational changes impact disassembly mechanisms of filopodia-like actin bundles. Structural transitions within the D-loop of actin filaments are implicated in the differential interactions of actin binding proteins, which in turn leads to the assembly of the large variety of architecturally distinct actin networks required for specific cellular functions. The slow release of the inorganic phosphate group generates an age gradient along the filament linked to the nucleotide state, which can be used as signals for actin regulatory factor binding [21]. It is recognised that the proximal ends of parallel actin bundles, such as found in microvilli or stereocilia are enriched in ADP-actin interacting proteins [44].

Actin subunits undergo conformational changes linked to filament formation and ATP hydrolysis, including a flattening and change in angle between the subdomains [20,21,45]. The D-loop is a flexible region implicated in the conformational differences between actin in different nucleotides states and bound to different proteins. In ‘young’ ATP-bound filaments mixtures of open and closed D-loop conformations are visible in 2.2 Å structures whereas in ‘aged’ ADP-actin filaments the D-loop seems predominantly closed [20,21]. In lower resolution structures, phalloidin-bound F-actin is seen in a closed state, as is LifeAct-bound F-actin [38–40], (although at this resolution these could represent an averaged state of multiple ‘closed’ conformations) whereas, in contrast, jasplakinolide-stabilized actin filaments have the D-loop in an open state [21,46]. Engineered utrophin CH1 domain probes whose binding sites overlap with the D-loop of the actin filament have different localization properties to actin in different parts of the cell and they also display differential binding to jasplakinolide and phalloidin-stabilized actin where the D-loop conformation is implicated [47]. These structural aspects have also been linked to functional consequences. For example, the structure of actin filaments from maize that form very stable, long tracks important for vesicle and organelle transport, are found in the open D-loop conformation [48] whereas malaria actin that is short and dynamic is found in the closed D-loop conformation [46]. Recently, structures of INF2-bound actin show that movement of the D-loop is part of its severing mechanism [37] and in molecular dynamics simulations of actin filaments under strain, the D-loop interface with the neighbouring subunit is opened [49].

The set of antibodies we identified have a 100-fold higher affinity for jasplakinolide-stabilized F-actin and disrupted the steady-state length distribution of FLS, which implies an alteration in the molecular mechanisms of length regulation. Actin-stabilizing compounds phalloidin and jasplakinolide can decouple D-loop conformational changes from the nucleotide state. Using phalloidin and jasplakinolide to stabilize these specific conformations, in combination with their displacement of closed-D-loop binding LifeAct, suggested that the antibodies specifically recognise the open D-loop conformation. In doing so the antibodies prevented actin disassembly, which we could reveal by separating actin assembly and disassembly processes using photobleaching and timelapse imaging. It is possible that the antibodies bind elsewhere on the actin filament and prevent the probes binding by stabilizing actin conformation allosterically rather than by directly binding and occluding the D-loop. Molecular modeling, structural biology and mutagenesis could be used to further elucidate the epitopes of the antibodies. Initial molecular dynamics simulations have been inconclusive due to the size of the actin filament.

Addition of supplemental cofilin reduced the excess bundle length caused by the antibodies. Disassembly factors cofilin and coronin respond to filament nucleotide status, where D-loop flexibility is one structural component. Cofilin and coronin both bind at the D-loop region and have opposite effects on its susceptibility to cross-linking [50]. Cofilin preferentially binds ADP-actin, inserting between the C-terminus and D-loop to alter the helical twist of the filament and ultimately induce severing. It requires the release of the inorganic phosphate group from the filament to function [20,27]. In contrast, coronin-1B recognises the ADP-Pi state ∼50 times as strongly as it does the ADP state [21,41]. We have not distinguished whether cofilin directly competes with antibodies for D-loop binding or if it induces D-loop closure or disassembly regardless of antibody binding at high concentrations.

Distinct from previous work on actin depolymerization studying small numbers of filaments, here we have been examining large fascin-bundled actin superstructures within a cytosolic-like environment of Xenopus egg extracts. Thus, this work accounts for the complex multiprotein interactions and effects of multiple actin filaments working together as a dynamic unit, and we have identified that the fundamental structural characteristics of the actin filament represent a major functional contribution. Our antibodies inhibit actin bundle disassembly, displace D-loop interactions and preferentially bind jasplakinolide-stabilized filaments, all of which suggest that the antibodies bind and stabilize an open D-loop state. In summary, our work therefore implies that the transition to the closed state of the actin filament D-loop is a critical conformational switch for disassembly and resulting length regulation of complex filopodia-like F-actin bundles, which may also be relevant in cells.

This research has been regulated under the Animals (Scientific Procedures) Act 1986 Amendment Regulations 2012 after ethical review by the University of Cambridge Animal Welfare and Ethical Review Body and covered by Home Office Project Licence PP1038769 (licence holder: J.L. Gallop) and Home Office Personal licences held by H. Fisher, T. Jones-Green, and J. Mason.

The screen that isolated mAbs 2, 4, 6, 14 and 17 was described in Jarsch et al. [29]. For each of these antibodies, as well as the R347 and NIP228 isotype controls, plasmids for expression of the HuIgG1 heavy chain and the IgG light chain were contransfected into CHO-transient cells and the IgGs purified from the cell culture medium via Protein A chromatography. DSHB antibodies A–E were obtained as hybridoma conditioned culture media supernatants from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology, Iowa City, IA 52242, U.S.A. Antibody A is JLA20, deposited to the DSHB by Jim Jung-Ching Lin, Cold Spring Harbor Laboratory [51], Antibody B is CPTC-ACT Group 1, deposited by Clinical Proteomics Technologies for Cancer, National Cancer Institute, Antibody C is 224-236-1, deposited by Guenther Gerisch, Max-Planck Institute of Biochemistry [52], Antibody D is BB8/20.1 1, deposited by Belinda Bullard, University of York, and Antibody E is 8-7A5, deposited by Angel L. De Blas, University of Connecticut [53]. The mouse anti-pan actin antibody was from Cytoskeleton (#AAN02-S). The anti-human Alexa-647 conjugated secondary antibody (#109-605-008) and anti-mouse HRP conjugated secondary antibody (#715-035-150) were from Jackson ImmunoResearch. The anti-human HRP conjugated secondary antibody was from Southern Biotech (#9040-05). The anti-mouse IRDye CW800 conjugated secondary antibody was from LI-COR Biosciences (#926-32210) Atto-488 conjugated actin was from Hypermol (#8153-02). Atto-655-LifeAct was from Cambridge Research Biochemicals (#crb1101303h). Alexa-568 conjugated actin (#A12374), Alexa-488 coupled phalloidin (#A12379), unlabelled phalloidin (#P3457) and jasplakinolide (#J7473) were from Thermo Fisher Scientific. Purified rabbit skeletal muscle actin (#AKL99) and human recombinant cofilin (#CF01) were from Cytoskeleton. The GFP-UtrCH binding domain construct was originally a kind gift from Sarah Woolner, University of Manchester, and cloned into a pGEX vector for bacterial expression. This construct was transformed into Rosetta DE3 pLysS Escherichia coli (Merck, #70956) and expression induced with IPTG overnight at 19°C. Cells were harvested by centrifugation and resuspended in 150 mM NaCl, 20 mM HEPES, 2 mM EDTA, 2 mM DTT and EDTA-free cOmplete protease inhibitor (Roche #11873580001), then lysed by probe sonication. The lysate underwent ultracentrifugation at 40 000 rpm for 45 min in a 70Ti rotor (Beckman-Coulter), with the supernatant applied to 5 ml equilibrated Glutathione Sepharase 4B beads (Merck #17-0756) at 4°C for 1 h. Following five washes in 150 mM NaCl, 20 mM HEPES, 2 mM EDTA, 2 mM DTT, three washes in 300 mM NaCl, 20 mM HEPES, 2 mM EDTA, 2 mM DTT and two washes in 150 mM NaCl, 20 mM HEPES, 2 mM EDTA, 2 mM DTT, 25 mM CaCl₂, the beads were resuspended in 2 ml of the latter buffer with 100 units Thrombin (GE Healthcare #27-0846) and cleaved at 4°C overnight. The beads and supernatant were separate via a gravity flow column, and the purified GFP-UtrCH separated from the thrombin by passing over a HiTrap Q FF column (#GE17-5156) on an ÄKTA FPLC system (both Cytiva) in a buffer containing 150 mM NaCl, 20 mM HEPES, 2 mM EDTA, 2 mM DTT. The elutions were pooled and concentrated in an Amicon Ultra 10 kDa MWCO spin concentrator (Merck #UFC801024), then diluted in 50% glycerol and stored at −20°C.

Xenopus laevis egg extracts, supported lipid bilayers composed of 60% phosphatidylcholine (#840053C), 30% phosphatidylserine (#840032C) and 10% PI(4,5)P₂ (#840046X, all Avanti Polar Lipids), and FLS assays were performed as previously described [54]. For all experiments, a final 1:6 dilution of 25 mg/ml extracts was used in a 50 µl volume reaction mix also containing 2 mM DTT, energy mix (50 mM phosphocreatine, 20 mM Mg-ATP) and made up in extract buffer (XB) (100 mM KCl, 100 nM CaCl₂, 1 mM MgCl₂, 10 mM K-HEPES pH 7.4, 50 mM sucrose). When required (and as indicated in Figs), GFP-UtrCH and 655-LifeAct were added to the reaction mix at 250 nM, fluorescent actins were added at 210 nM, supplemental unlabelled actin was added at 1 µM, and cofilin was added at 550 nM. Antibodies were exchanged into a buffer containing 100 mM KCl, 1 mM MgCl₂, and 10 mM K-HEPES pH 7.4 via a 10 kDa MWCO spin concentrator, and added at the indicated concentrations to the FLS mix alongside an equal volume of double-sucrose XB (XB as above containing 100 mM sucrose rather than 50 mM). DSHB antibodies were buffer exchanged as above and 8 µl added to the reaction mix, alongside an equal volume of double-sucrose XB. Phalloidin and jasplakinolide were prepared as 2.5 mM stocks in methanol, and added to the reaction mix at the indicated concentrations. The vehicle control was methanol added at a 1:500 final dilution, corresponding to the volume required for the highest phalloidin and jasplakinolide concentration. Each intervention was allowed to pre-incubate for 10 min before being added to the supported lipid bilayer to initiate the assay. Steady state snap-shots were imaged over the time period 25–35 min after assay initiation. Timelapse experiments were imaged over the time periods 5–15 and 25–35 min after initiation.

For FLS immunolabelling experiments, assay mixes in XB containing frog egg extracts, DTT, energy mix, Atto-488 actin were prepared and initiated as above. At T = 20 min, 0.5 µl phalloidin from a 10 mg/ml methanol stock was added for a further 5 min. The structures were gently washed twice with 50 µl XB containing 1× energy mix and 2 mM DTT, then fixed in 50 µl 4% formaldehyde (#28908, Thermo Fisher Scientific) in XB for 75 min. 50 µl 50 µM glycine in PBS was added for 15 min to quench the fixative, then the structures were blocked for 60 min in 10% goat serum (Merck #G6767) in PBS. Primary antibodies were prepared at a 1:100 dilution (corresponding to final concentrations between 40 and 65 nM) in 1% goat serum in PBS, and 50 µl added to each well, then incubated at 4°C overnight. Assays were carefully washed in 1% goat serum-PBS three times for 5 min. The anti-human Alexa-647 secondary antibody was prepared at a 1:200 dilution in 1% goat serum-PBS, added, and incubated for 1 h. Assays were carefully washed in 1% goat serum-PBS three times for 5 min and immediately imaged.

FLS were imaged at room temperature on a combined spinning disk confocal/TIRF system provided by CAIRN research equipped with an Eclipse-Ti inverted microscope (Nikon), X-light Nipkow spinning disk unit (CREST), Spectra X LED illuminator (Lumencor) and NanoScanZ 250 µm piezo stage (Prior) via a 100x 1.40 NA Plan Apo VC oil objective (Nikon) on an Evolve Delta EMCCD camera (Photometrics) in 16-bit depth using Metamorph software. Alexa/Atto 488, 568 and 647/655 samples were visualised using 470/40, 560/25 and 628/50 excitation and 525/50, 585/50 and 700/75 emission filters respectively. FLS were imaged at 1 micron Z-stack intervals, taking 30–60 micron stacks, with the exception of the immunolabelling and live imaging experiments, which were both imaged at 0.5 micron intervals. For bleaching experiments, an iLas2 illuminator (Roper Scientific) via a 100x 1.49 NA oil immersion objective was used to bleach the actin channel in an entire field of view at the level of the supported lipid bilayer via the TIRF laser (100% power, three repetitions). Images were taken via the standard confocal set up used in all other experiments. At the correct timepoint after initiation, a reference 60 micron z stack at 0.5 micron intervals was taken before bleaching, followed by bleaching at the base slice, as described above. An immediate time-lapse was initiated, capturing a 26 micron z stack at 0.5 micron intervals every 20 s for 10 min. A second 60 micron z stack at 0.5 micron intervals was then taken at the conclusion of the timelapse. For each assay, a region of interest was captured between 5 and 15 min after assay initiation, then a second region of interest was used to assay the 25–35 min timepoint.

HEK293 cells (ATCC) were grown as adherent cultures at 37°C in a 5% CO₂ atmosphere. They were maintained in Dulbecco's modified Eagle's medium (Merck #D6546) supplemented with 10% fetal bovine serum (Merck #F9664) and 1X penicillin-streptomycin-glutamine (Thermo Scientific #10378016), and passaged twice weekly. They were passaged onto glass coverslips, and allowed to grow to ∼50% confluency before processing. Coverslips were fixed in 4% formaldehyde for 10 min, then washed three times in PBS. They were permeabilised and blocked in a mix containing 0.1% saponin, 10% goat serum and 50 mM NH₄Cl in PBS for 45 min. Primary antibody mixes containing 4 µg/ml (∼25 nM) of each relevant antibody were prepared in 1% goat serum-PBS, and incubated on the coverslips overnight at 4°C. They were then washed three times at 4°C in 1% goat serum-PBS and incubated with a mix containing Alexa-488 phalloidin (1:250) and the anti-human alexa-647 secondary antibody (1:500) in PBS for 45 min at 4°C. They were then washed four times at 4°C in 1% goat serum-PBS, and mounted using Hydromount (National Diagnostics #HS-106). Images were acquired on an Airyscan Zeiss LSM 880 via a Plan-Apochromat 63x NA 1.4 objective using the Airyscan Fast imaging mode as a confocal z-stack taken at 0.2 µm intervals. Images were processed using Airyscan processing, strength 6, and are presented as maximal z-projections.

Freshly laid X. laevis eggs were fertilized and dejellied using 2% cysteine in MMR, then transferred into 4% Ficoll-0.1% MMR for microinjection. Once embryos reached the two cell stage, one cell in each embryo was injected with 10 nl of 9 µM mAb14 or R347 using a Parker Picospritzer III microinjector. Embryos were monitored on a Zeiss Stemi 305 stereo microscope, with images captured at each division stage. Embryos for each antibody at the four and eight cell stage were scored for the presence of partial or absent cleavage furrows in one half of the embryo, or a delay in cleavage events by two independent researchers, and the mean of those values reported.

Samples were prepared in 4x Laemmli sample buffer (final 1x composition: 2% SDS, 50 mM Tris pH 6.8, 10% glycerol, 5% beta-mercaptoethanol) and run on 4–20% mini-PROTEAN TGX precast polyacrylamide gels (Bio-Rad #4561096) in Tris-glycine-SDS running buffer, using the PageRuler Plus Prestained Protein Ladder (Thermo Fisher #26619) as a molecular mass marker. For coomassie staining, InstantBlue stain (abcam #ab119211) was added to gels overnight, then destained in ddH₂O prior to scanning. For the anti-actin Western blot, gels were transferred onto nitrocellulose membrane using the iblot 2 dry blotting system (Thermo Fisher #IB23002). Membranes were blocked in 10% milk (Merck #70166) in TBST, then incubated with the anti pan actin primary antibody overnight at 4°C (dilution 1:500 in 1% milk-TBST). Following three washes in 1% milk-TBST, the membranes were incubated with the anti-mouse CW800 secondary antibody (1:5000 in 1% milk-TBST), followed by three further washes in 1% milk-TBST before scanning. Both coomassie gels and Western blots were scanned on an Odyssey Sa reader (Li-COR biosciences). Densitometry analysis was performed on scanned coomassie gel images using FIJI (ImageJ) software [55]. The intensity of each band representing actin or the antibody heavy chain in the pellet and supernatant for each treatment was measured, as well as two strips above and below these bands to measure the background. The average of these strips was taken and subtracted from each individual band measurement. For each condition the percentage of protein in the pellet was calculated from these background subtracted measurements as: Pellet Intensity / (Pellet Intensity + Supernatant Intensity) × 100.

Antibodies were pre-cleared by centrifugation at 5000 × g for 5 min at 4°C. For each condition, 24 µg of relevant antibody was coupled to 100 µl protein A dynabeads in PBS for 2 h at 4°C on a rotator, after which beads were washed twice in 1 ml 0.02% Tween20-PBS (PBST), once in 1 ml PBST + 250 mM NaCl, and twice in 1 ml XB. Meanwhile, for each condition, 100 µl X. laevis egg extracts were preincubated with either 20 µM Latrunculin-B (Merck, #428020), or the equivalent volume of DMSO (vehicle control) for 10 min on ice. The washed antibody-coupled beads were incubated with the relevant pretreated frog egg extracts for 4 h at 4°C on a rotator. They were then washed in 1 ml XB followed by 3 × 1 ml washes in PBST. The beads were then resuspended in 30 µl Laemmli sample buffer and heated at 95°C in a heat block for 5 min. The supernatant was separated from the beads and run on gels for coomassie staining and Western blotting.

Rabbit skeletal muscle actin was resuspended to 1 mg/ml in general actin buffer (5 mM Tris–HCl pH 8.0, 0.2 mM CaCl₂, 0.2 mM ATP, 0.5 mM DTT) overnight. Antibodies were diluted to 0.6 µg/ml in PBS. Both underwent a clearing spin at 150 000 × g for 1 h at 4°C. For each assay, 52.5 µl of cushion buffer (2.5 µl 10x actin polymerization buffer polymerization buffer [500 mM KCl, 20 mM MgCl₂, 10 mM ATP] (Cytoskeleton #BSA02), 7 µl glycerol, 2 µl ddH₂O, 1 µl 10x PBS, 40 µl general actin buffer) was prepared and added to the bottom of the ultracentrifuge tube. On top of this the reaction mix was carefully overlaid as follows: 40 µl cleared actin, 2.5 µl 10x polymerization buffer, 10 µl (µg) cleared antibody. Assays were incubated at 4°C for 1 h, then centrifuged at either 150 000 × g for 90 min (for the F-actin assay), or 14 000 × g for 60 min (for the bundled filament assay). The supernatant was carefully removed and 30 µl laemmli sample buffer added. The pellet was then resuspended in 105 µl PBS and 30 µl laemmli sample buffer added. Both supernatant and pellet samples were heated in a 95°C heatblock for 5 min, then 10 µl of each sample run an an SDS–PAGE gel for coomassie staining.

Rabbit skeletal muscle actin was resuspended to 1 mg/ml (21 µM) in 5 mM Tris–HCl pH 8.0 + 0.2 mM CaCl₂. Equal volumes were split to prepare F-actin or G-actin. G-actin was prepared by adding 0.2 mM ATP and leaving on ice for 30 min, before adding a 1:10 volume of 5 mM Tris–HCl pH 8.0 + 0.2 mM CaCl₂ and leaving on ice for a further 30 min. F-actin was prepared by adding a 1:10 volume of 10x actin polymerization buffer and leaving at room temperature for 1 h. For jasplakinolide and phalloidin treatments, F-actin was prepared as above and left at room temperature overnight, before jasplakinolide or phalloidin was added to a final concentration of 40 µM. A dilution series of actin concentrations was prepared with G-actin diluted in 5 mM Tris–HCl pH 8.0 + 0.2 mM CaCl₂ + 0.2 mM ATP, and F-actin (and F-actin treated with jasplakinolide or phallodin) diluted in 1x actin polymerization buffer diluted in 5 mM Tris–HCl pH 8.0 + 0.2 mM CaCl₂. 50 µl volumes of each dilution were added to wells of a MaxiSorp 96-well Immuno plate (Thermo Scientific #437111) in triplicate and incubated overnight, with F-actin plates at room temperature and G-actin plates at 4°C. The plates were then washed in 200 µl PBS-0.1% Tween-20 three times, then blocked in 200 µl 3% BSA in PBS for 1 h. Antibodies were prepared in 1% BSA-PBS. R347 and mAbs 2, 6, 14 and 17 were used at 10 nM final concentration, the pan actin antibody was used at 1:1000. Wells were washed in 200 µl PBS-0.1% Tween-20 three times, then 100 µl of the primary antibody mixes added and incubated for 1 h. Anti-human or anti-mouse HRP antibodies were prepared in 1% BSA-PBS at 1:10 000 dilution. Wells were washed in 200 µl PBS-0.1% Tween-20 three times, then 100 µl secondary antibodies mixes added and incubated for 1 h, followed by three final 200 µl PBS-0.1% Tween-20 washes. HRP was detected using the QuantaBlu Fluorogenic peroxidase kit (Thermo Scientific #15169). The substrate and peroxidase solutions were mixed according to manufacturers instructions, and 50 µl added to wells for 30 min, followed by the addition of 50 µl QuantaBlu stop solution. Plates were read on a Hidex Sense using 320/40 and 444/12 excitation and emission filters, or on a Perking Elmer Multilabel EnVision microplate reader. Response curves were calculated and plotted using the [Agonist] vs. response –Variable slope analysis in GraphPad Prism software (version 10.1.0).

FLS length data was measured using our custom FIJI (Image J) plugin FLSACE, described in [5], which maps FLS through Z-stacks. Lengths reported are the path-length of the segmented structure measured from the base slice. Structures are defined as FLS when they have a length of at least 3 µm, a base area of at most 20 µm², and a base circularity of at least half that of a circle.

To calculate UtrCH and LifeAct displacement, the total area of fluorescence from the base slice upwards was measured in each z-slice in both channels of the confocal stack. In the actin channel, the actin extent was defined as the first z-slice with an actin area of zero. The UtrCH or LifeAct extent was defined as the first slice where the UtrCH area is less than or equal to half the actin area in that slice. This ratio of UtrCH or LifeAct extent/actin extent was calculated for each field of view and plotted as the half-maximal UtrCH extent/actin extent. The area of fluorescence measurements are normalised and plotted as the averaged fluorescence Z-slice profiles. For every Z-slice in a field of view, the fluorescence area measurement was normalised relative to the area of that channel in the first slice above the base slice. The graphs plot the mean normalised area at each z-slice for both channels.

To plot the fluoresence intensity along an example individual FLS, the mean intensity of both channels in a 5 × 5 pixel box at the centre of the FLS of interest was measured at each Z plane, beginning at the base slice, using the multi measure tool of the ROI manager in FIJI. A background measurement was taken for each channel and subtracted from the raw data, with each datapoint for a channel normalised as a parcentage of the intensity at the base slice for that channel, and plotted as the normalised intensity percentage relative to the Z interval above the base slice.

To plot the timelapse mean length profiles, the lengths of individual FLS were tracked over time in FLSACE, with T = 0 for each individual FLS set as its first detection by FLSACE. To be included, FLS had to be detected for a minimum of six consecutive frames (2 min). The graph plots the mean length of all detected FLS at each time point relative to this first appearance. All graphs were prepared and all statistical analysis performed in GraphPad Prism software (version 10.1.0), as described in Fig legends.

Data is available on request to J.L.G. The antibodies are available from AstraZeneca with a Material Transfer Agreement.

Stacey Chin is a current employee of AstraZeneca and holds AstraZeneca shares. Other authors declare no competing interests.

The work was funded by: a Wellcome Trust (www.wellcome.org) Senior Research Fellowship 219482/Z/19/Z to J.L.G., a MedImmune/AstraZeneca (www.astrazeneca.co.uk) PhD studentship (RG75909/A15360) awarded to C.D., J.L.G. and P.S.I., a BBSRC (www.ukri.org/councils/bbsrc/) Flexible Talent mobility grant RG99217/22799 to J.R.G. We acknowledge core institute funding by the Wellcome Trust (092096) and Cancer Research UK (www.cancerresearchuk.org/) (C6946/A14492). MedImmune/AstraZeneca scientists were involved in study design, data analysis, decision to publish and preparation of the manuscript and are listed as authors. No other funders played any role in study design, data collection or analysis, decision to publish or preparation of the manuscript.

Open access for this article was enabled by the participation of University of Cambridge in an all-inclusive Read & Publish agreement with Portland Press and the Biochemical Society under a transformative agreement with JISC.

Jennifer Gallop: Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing — original draft, Project administration, Writing — review and editing. Jonathan Richard Gadsby: Conceptualization, Data curation, Formal analysis, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing — original draft, Writing — review and editing. Pantelis Savvas Ioannou: Data curation, Formal analysis, Validation, Investigation, Methodology, Writing — review and editing. Richard Butler: Software, Visualization, Methodology, Writing — review and editing. Julia Mason: Resources, Validation, Investigation, Methodology, Writing — review and editing. Alison Smith: Resources, Supervision, Methodology, Writing — review and editing. Ulrich Dobramysl: Data curation, Software, Methodology, Writing — original draft, Writing — review and editing. Stacey Chin: Resources, Supervision, Project administration, Writing — review and editing. Claire Dobson: Conceptualization, Resources, Supervision, Funding acquisition, Project administration.

We thank Thomas Blake, Laura Machesky and Daniel St Johnston for valuable comments on the manuscript. We thank the Gurdon Institute Imaging facility for microscopy support and image analysis.

FLS

filopodia-like structure

HSS

high-speed supernatant