To test their relative importance in the initiation versus propagation of endocytic actin networks, we measured the influence of Dip1 and Wsp1 mutations on actin dynamics in fission yeast using the endocytic actin patch marker Fim1 labeled with green fluorescent protein (GFP) (Berro and Pollard, 2014). In wild-type cells, Fim1-marked actin patches accumulate in cortical puncta over ~5 s before moving inward and simultaneously disassembling (Figure 1A–C, Figure 1—video 1; Berro and Pollard, 2014; Sirotkin et al., 2010). To quantify actin patch initiation defects, we measured the rate at which new Fim1-marked puncta appeared in the cell (Figure 1D). As expected based on previous results, the Dip1 deletion strain showed a significant reduction in the patch initiation rate compared to the wild-type strain (0.030 versus 0.0076 patches/s/µM²) and a corresponding decrease in the number of actin patches in the cell (Figure 1D and E, Figure 1—video 2; Basu and Chang, 2011). However, once actin assembly was initiated, Fim1-GFP accumulated more rapidly than in the wild-type strain (Figure 1B and C). These observations are consistent with previously reported measurements (Basu and Chang, 2011) and suggest that Dip1 contributes to the initiation but not the propagation of branched endocytic actin networks. We note that while dip1 deletion caused increased Fim1-GFP to accumulate in individual patches (Figure 1B), the ratio of the total actin in patches versus total actin in the cell decreased in the dip1Δ strain because fewer patches were initiated. Therefore, the reduced rate of actin patch assembly cannot be explained by excess filamentous actin in the patches depleting the concentration of soluble actin monomers in the cytoplasm (Figure 1—figure supplement 1).

Figure 1 with 5 supplements see all

Download asset Open asset

Figure 1—source data 1

Analysis of spinning disk confocal videos of S. pombe source data.

https://cdn.elifesciences.org/articles/60419/elife-60419-fig1-data1-v2.xlsx

Download elife-60419-fig1-data1-v2.xlsx

To investigate the contribution of Wsp1 toward initiation and propagation of the actin networks, we deleted the sequence encoding the CA segment of Wsp1 in the endogenous wsp1 locus and measured the influence of this mutation on actin dynamics. Deletion of the CA segment prevents Wsp1 from binding or activating Arp2/3 complex (Marchand et al., 2001), but leaves intact its WASP-homology 1 (WH1) domain, proline-rich segment, and actin-binding Verprolin-homology motif (V). In the wsp1ΔCA mutant the average time between the first appearance of Fim1-GFP and when it reaches peak concentration, which we refer to here as the Fim1 assembly time, increased from 5.1 to 6.7 s, consistent with another recent study (Figure 1B and C; Sun et al., 2019). In addition, the wsp1ΔCA mutation decreased the percentage of actin patches that internalized (Figure 1F, Figure 1—video 3). These observations are consistent with a role for Wsp1 in the propagation of branched actin during endocytosis. However, to our surprise, we found that wsp1ΔCA cells also showed a 40% decrease in the rate of initiation of new actin patches compared to wild-type cells (Figure 1D). While this defect is less than observed in dip1Δ cells, it suggests – contrary to our initial prediction – that Wsp1 may play a role in initiating new endocytic actin patches. Combining wsp1∆CA and dip1∆ in the same strain (dip1∆ wsp1∆CA) did not decrease the actin patch initiation rate more than dip1∆ alone (Figure 1D, Figure 1—video 4). This suggests that Wsp1 may contribute to the Dip1-mediated actin patch initiation pathway rather than acting in a separate parallel pathway for initiation of actin assembly.

Previous biochemical and structural data suggested that Dip1 and Wsp1 might simultaneously bind Arp2/3 complex, so they could potentially cooperate to activate nucleation (Luan et al., 2018a; Luan et al., 2018b; Wagner et al., 2013). We reasoned that by directly synergizing with Dip1 to activate Arp2/3 complex, Wsp1 could contribute to actin network initiation. However, how the two NPFs together influence the activity of Arp2/3 complex is uncertain. Previous data showed Dip1 is a more potent activator of Arp2/3 complex than Wsp1, and that mixing both Wsp1 and Dip1 in a bulk actin polymerization assay increased the actin polymerization rate, but the reason for the increase was unknown (Wagner et al., 2013). Specifically, because those experiments were carried out at sub-saturating conditions, it was unclear whether Wsp1 and Dip1 activate in independent but additive pathways or alternatively, if the two NPFs synergize in activation of Arp2/3 complex. To test this, we titrated Dip1 into actin polymerization reactions containing Arp2/3 complex and the minimal Arp2/3-activating region of Wsp1, Wsp1-VCA. Dip1 approaches saturation above ~30 μM, and at this concentration the addition of Wsp1-VCA increased the maximum polymerization rate in the pyrene actin assembly assays ~1.6-fold over reactions with Dip1 alone (Figure 2A and B, Supplementary file 1). These results suggest that the increased actin polymerization rates in the presence of both NPFs cannot be explained by an additive effect in activating Arp2/3 complex, but instead the NPFs are synergistic.

Figure 2 with 2 supplements see all

Download asset Open asset

Figure 2—source data 1

Analysis of TIRF microscopy videos source data.

https://cdn.elifesciences.org/articles/60419/elife-60419-fig2-data1-v2.xlsx

Download elife-60419-fig2-data1-v2.xlsx

Our bulk actin polymerization assays demonstrate that Wsp1 and Dip1 synergize to activate Arp2/3 complex, but it is unclear whether synergetic activation requires that the complex bind a preformed actin filament, as required when Wsp1 activates Arp2/3 complex on its own. Therefore, it is unclear whether the synergistic activation mechanism could explain how Wsp1 contributes to initiation of new endocytic actin patches. To better understand how the two NPFs synergize, we used single molecule TIRF microscopy to monitor the assembly of Oregon Green 488-labeled actin in the presence of Arp2/3 complex and Wsp1, Dip1, or both NPFs. We labeled Dip1 with Alexa Fluor 568 (568-Dip1) to mark actin filaments nucleated by Arp2/3 complex and Dip1. In the presence of Arp2/3 complex and 568-Dip1, we observed assembly of linear filaments, a subset of which had Dip1 bound at one end (Figure 2C). These filaments, which largely represent Dip1-Arp2/3 nucleated filaments (Balzer et al., 2018), account for 3.4% of the total filaments in the reaction after two-and-a-half minutes, with the other 96.7% arising from spontaneous nucleation (Figure 2C and D, Figure 2—figure supplement 1). Adding Wsp1-VCA to the reaction significantly increased both the total number and the percentage of linear filaments with bound Dip1 (Figure 2C and D – Figure 2—figure supplement 1). These data demonstrate that synergistic activation of Arp2/3 complex by the two NPFs results in the nucleation of linear rather than branched actin filaments. Therefore, we conclude that like Dip1-mediated activation (Wagner et al., 2013), synergistic co-activation by both NPFs does not require a preexisting actin filament.

Because Wsp1 may function as an oligomer when clustered at endocytic sites (Padrick et al., 2008), we also tested if Wsp1-VCA dimerized with glutathione S-transferase (GST) synergizes with Dip1 to activate Arp2/3 complex. Under our initial reaction conditions (100 nM Arp2/3 complex and 1 μM GST-VCA), we did not detect synergistic co-activation of the complex by Dip1 and GST-VCA (Figure 2D). However, in reactions in which the concentration of Arp2/3 complex was increased 2.5-fold, the number and percentage of pointed ends with bound Dip1 increased significantly in the presence of GST-VCA, indicating dimeric (GST) Wsp1-VCA synergizes with Dip1 (Figure 2E and F, Figure 2—figure supplement 1). To further investigate the influence of Wsp1 dimerization on synergy, we compared the influence of monomeric and dimeric Wsp1-VCAs on the maximum polymerization rate in bulk pyrene actin polymerization assays containing Arp2/3 complex and a range of Dip1 concentrations (Figure 2G). These data showed that both dimeric and monomeric Wsp1-VCA significantly decrease the concentration of Dip1 required to saturate the reaction (Figure 2H, Supplementary file 1). However, unlike monomeric Wsp1-VCA, dimeric Wsp1-VCA did not increase the maximum polymerization rate at saturating Dip1. These data demonstrate that monomeric Wsp1 is more potent in its synergy with Dip1 than dimeric Wsp1, and point to differences in the mechanism of synergistic activation between monomeric and dimeric Wsp1-VCA (see Discussion).

To determine the mechanism of co-activation by Dip1 and Wsp1, we first examined the kinetics of activation by Dip1 alone to identify steps in the activation pathway that might be accelerated by Wsp1. We measured time courses of actin polymerization in reactions containing actin, S. pombe Arp2/3 complex, and a range of concentrations of Dip1 (0–15 μM) and asked whether various kinetic models were consistent with the polymerization time courses (Figure 3A). In the simplest model we considered (Figure 3B [model i]), Dip1 binds to Arp2/3 complex and initiates an irreversible activation step to create a filament barbed end. This step could represent an activating conformational change, such as movement of Arp2 and Arp3 into the short pitch helical arrangement or subunit flattening of Arp3 (Figure 3B; Rodnick-Smith et al., 2016b; Shaaban et al., 2020; Wagner et al., 2013). The value of the irreversible activation step was floated in the simulations, and the other rate constants were fixed or restrained as described in the supplementary materials. This simple model produced simulated time polymerization courses that fit the data poorly compared to the measured time courses (Figure 3A, Figure 3—figure supplements 1 and 2). Specifically, the simulations predicted faster polymerization than observed at time points near steady state when the concentration of free actin monomers is low. Therefore, we wondered whether collision with and binding of one or more actin monomers to the Dip1-Arp2/3 assembly might be required to complete the activation process and create a nucleus. To test this, we altered the kinetic model to include one or more actin monomer-binding steps before creation of the nucleus (Figure 3B, [models ii and iii]). These models produced simulated polymerization time courses that fit the data significantly better than the reaction pathway without actin monomer collisions (Figure 3C and D, Figure 3—figure supplements 1 and 2). The pathway with one actin monomer-binding step fits the data most closely, but the fits with two or three monomer-binding steps also improved the fit over the reaction pathway without actin monomer binding (Figure 3D). These data suggest that actin monomer binding to the Dip1-bound Arp2/3 complex stimulates activation.

Figure 3 with 2 supplements see all

Download asset Open asset

Figure 3—source data 1

Source data for pyrene actin polymerization time courses and simulations.

https://cdn.elifesciences.org/articles/60419/elife-60419-fig3-data1-v2.xlsx

Download elife-60419-fig3-data1-v2.xlsx

Our data suggest that slow binding of actin monomers to Dip1-bound Arp2/3 complex limits the nucleation rate. Importantly, unlike Dip1, Wsp1 binds both Arp2/3 complex and actin monomers, so it can directly recruit actin monomers to nascent nucleation sites (Beltzner and Pollard, 2008; Marchand et al., 2001). We wondered if Wsp1 synergizes with Dip1 by recruiting actin monomers to the Dip1-Arp2/3 complex assembly. To test this, we asked whether the actin monomer-recruiting V region of Wsp1 is required for synergistic co-activation of Arp2/3 complex by Dip1 and Wsp1. We found that while adding Wsp1-VCA to actin polymerization reactions containing saturating Dip1 increased the maximal polymerization rate ~1.6-fold compared to reactions without Wsp1, addition of Wsp1-CA had little or no effect on the maximum polymerization rate (Figure 3E and F). Therefore, we conclude that actin monomer recruitment by Wsp1 is required for potent synergy with Dip1. Wsp1-CA slightly decreased the concentration of Dip1 required for half maximal saturation (K_1/2), suggesting that it influences one or more of the activation steps (Supplementary file 1). However, this influence was small compared to the reduction in the K_1/2 of Dip1 caused by Wsp1 with a V region (Supplementary file 1).

Our data show that Wsp1 must recruit actin monomers to Arp2/3 complex to potently synergize with Dip1 in activation. To better understand how actin monomer recruitment contributes to synergy we sought to kinetically model synergistic activation of Arp2/3 complex by the two NPFs. To decrease the number of unknown rate constants inherent in an explicit model of all reactions in a mixture containing Dip1, Wsp1, and Arp2/3 complex, we used a simplified model based on the activation by Dip1 alone (Figure 4A and B). We asked if this simplified model could fit time courses of actin polymerization for reactions containing both Dip1 and Wsp1 if the rate constants of key steps were increased. We limited the fitting to reactions with Dip1 concentrations greater than 0.5 µM, as higher concentrations of Dip1 limit the contribution of branching nucleation to actin assembly (Balzer et al., 2019). This allowed us to ignore the action of Wsp1 alone on Arp2/3 complex in the simulated reactions. Given that Wsp1 directly tethers actin monomers to the complex, we first asked whether increasing the monomer affinity for the Dip1-Arp2/3 assembly could explain the increased rate of filament nucleation. To test this, we simulated polymerization using the Dip1 alone activation model and allowed the off rate (k₋₁₀) for the actin monomer bound to the Dip1-Arp2/3 assembly to float. All other rate constants were fixed at the values determined for reactions without Wsp1. These simulations fit the data poorly, indicating the synergy between Wsp1 and Dip1 cannot be explained by increased affinity of actin monomers for the Dip1-Arp2/3 assembly alone (Figure 4B and C). However, when we floated the dissociation constant for actin monomers (k₋₁₀) and either the k_off of Dip1 for Arp2/3 complex (k₋₉) or the rate constant for the activation step (k₁₁) (or all three), the simulations closely matched the measured polymerization time courses (Figure 4C–E, Supplementary file 3). These observations suggest that multiple steps in the Dip1-mediated activation pathway are accelerated when Wsp1-bound actin monomers are recruited to Arp2/3 complex. We note that a similar set of simulations in which the association rate constants were floated for the actin monomer or Dip1-binding steps (instead of the dissociation rate constants) yielded the same results (Figure 4—figure supplement 1).

Figure 4 with 2 supplements see all

Download asset Open asset

Figure 4—source data 1

Source data for pyrene actin polymerization time courses and simulations in Figure 4.

https://cdn.elifesciences.org/articles/60419/elife-60419-fig4-data1-v2.xlsx

Download elife-60419-fig4-data1-v2.xlsx

We wondered whether the activating step(s) represented by k₁₁ might be one of the steps accelerated in the presence of both Dip1 and Wsp1. This step could represent activating conformational changes that occur in the recently published cryoEM structure of activated Arp2/3 complex bound to Dip1 (Shaaban et al., 2020). During activation, approximately one-half of the complex rotates into a new position, moving Arp2 and Arp3 into a conformation (the short pitch conformation) that mimics an actin dimer across the short pitch helix of an actin filament. In addition, each actin-related protein undergoes an intrasubunit conformational change called flattening. While no available assays probe flattening, we took advantage of a previously developed engineered site-specific crosslinking assay to test the influence of Wsp1 and Dip1 on stimulation of the short pitch conformation (Hetrick et al., 2013; Rodnick-Smith et al., 2016a). In this assay, the crosslinker bis-maleimidoethane (BMOE) (8 Å spacer length) crosslinks a pair of engineered cysteine residues only when Arp2 and Arp3 are in or near a short pitch arrangement. As expected, Wsp1 and Dip1 alone each stimulate short pitch crosslinking (Figure 4—figure supplement 2). Addition of a saturating concentration of Dip1 to a saturating concentration of Wsp1 increased short pitch crosslinking over that observed for either NPF alone, suggesting the two NPFs together may accelerate this activating step during synergistic activation. Dip1 also increased short pitch crosslinking when added to a reaction containing Wsp1 and actin monomers.

Given that co-stimulation of the short pitch conformation by Wsp1 and Dip1 would require simultaneous binding of Arp2/3 complex by the two NPFs, we also asked if Wsp1 and Dip1 co-bind to Arp2/3 complex. GST-tagged Wsp1-VCA was able to pull down a significant amount of Dip1 in the presence but not the absence of Arp2/3 complex, indicating formation of a Wsp1-Arp2/3-Dip1 assembly (Figure 4—figure supplement 2). Detection of the Dip1-Arp2/3-Wsp1 assembly with this method required using a mutant Arp2/3 complex that binds with increased affinity to Wsp1, presumably because the Dip1-Arp2/3 interaction is weak (Wagner et al., 2013).

To purify S. pombe Dip1, an N-terminally GST tagged Dip1 plasmid was generated by cloning the full length Dip1 sequence into the pGV67 vector as described previously (Balzer et al., 2018). The restriction sites chosen for cloning resulted in the presence of a short N-terminal polypeptide sequence (GSMEFELRRQACGR) on the end of the coding sequence for Dip1 after cleavage with tobacco etch virus (TEV). To purify Dip1, BL21(DE3) RIL Escherichia coli cells transformed with this pGV67-Dip1 plasmid were grown to an O.D. 595 of 0.6–0.7, induced with 0.4 mM isopropyl 1-thio-β-D-galactopyranoside (IPTG), and incubated overnight at 22°C. Cells were lysed by sonication in lysis buffer (20 mM Tris pH 8.0, 140 mM NaCl, 2 mM EDTA, 1 mM dithiothreitol [DTT], 0.5 mM phenylmethylsulfonyl fluoride [PMSF], and protease inhibitor tablets [Roche]) and then clarified by centrifugation (JA-20 rotor [Beckman], 18,000 rpm, 25 min, 4°C). The supernatant was pooled and loaded onto a 10 mL glutathione sepharose 4B (GS4B) column equilibrated in GST-binding buffer (20 mM Tris pH 8.0, 140 mM NaCl, 2 mM EDTA, 1 mM DTT). The column was then washed with GST-binding buffer until no protein was detected in the flow through (~10 CV). Protein was eluted with 20 mM Tris pH 8.0, 140 mM NaCl, and 50 mM glutathione. Fractions containing GST-Dip1 were pooled and dialyzed overnight against 20 mM Tris pH 8.0, 50 mM NaCl, and 1 mM DTT at 4°C in the presence of TEV protease (25:1 ratio of GST-Dip1 to TEV protease [by mass]). The dialysate was loaded onto a 6 mL Resource Q column equilibrated in Q_A buffer (20 mM Tris pH 8.0, 50 mM NaCl, and 1 mM DTT) and eluted over a 20 column volume gradient to 100% Q_B buffer (20 mM Tris pH 8.0, 500 mM NaCl, and 1 mM DTT). The protein was concentrated in a 10 k MWCO Amicon-Ultra centrifugal filter (Millipore Sigma) and loaded onto a Superdex 200 HiLoad 16/60 gel filtration column equilibrated in 20 mM Tris pH 8.0, 100 mM NaCl, and 1 mM DTT. Fractions containing pure Dip1 were pooled and flash frozen in liquid nitrogen. The final concentration of S. pombe Dip1 was determined by the absorbance at 280 nm using an extinction coefficient of 36,330 M⁻¹cm⁻¹.

The Dip1 construct used for site-specific labeling with the cysteine-reactive Alexa Fluor 568 C5 maleimide (Thermo Fisher) had the six endogenous cysteine residues in Dip1 mutated to alanine and a single cysteine added to the short N-terminal polypeptide sequence on the end of the coding sequence as previously described (Balzer et al., 2018). Expression and purification of this Dip1 mutant were identical to the wild-type purification until the protein was loaded onto the Superdex 200 HiLoad 16/60 gel filtration column. To increase labeling efficiency, the size exclusion column was equilibrated in 20 mM HEPES pH 7.0 and 50 mM NaCl before loading and eluting the concentrated protein. Peak fractions containing Dip1 were pooled and concentrated to 40 µM for labeling. A 10 mM solution of Alexa Fluor 568 C5 maleimide dye in water was added dropwise to the protein while stirring at 4°C until the solution reached a 10−40 molar excess of dye to protein. After 12–16 hr the reaction was quenched by dialyzing against 20 mM Tris pH 8.0, 50 mM NaCl, and 1 mM DTT for 24 hr. Labeled Dip1 was loaded onto a Hi-Trap desalting column and the peak fractions were pooled and flash frozen in liquid nitrogen. The final concentration of Alexa Fluor 568 dye was determined by measuring the absorbance at 575 nm and dividing this by 92,009 M⁻¹cm⁻¹, the extinction coefficient of the dye. The final concentration of 568-Dip1 was determined by the following equation: $\frac{A_{280-(}{A}_{575}\times 0.403)}{\mathrm{36,330}{M}^{-1}{cm}^{-1}}$.

To purify S. pombe Wsp1-VCA, residues 497–574 were cloned into the pGV67 vector containing an N-terminal GST tag followed by a TEV cleavage site. A 5 mL culture of BL21(DE3)-RIL E. coli cells transformed with the pGV67-Wsp1-VCA vector in LB plus 100 μg/mL ampicillin and 35 μg/mL chloramphenicol was grown overnight at 37°C. One milliliter of this culture was used to inoculate 50 mL of LB plus ampicillin and chloramphenicol, which was allowed to grow at 37°C with shaking until turbid. Ten milliliters of this turbid culture were added to a 2.8 L flask containing 1 L of LB plus ampicillin and chloramphenicol and grown to an O.D. 600 of 0.4–0.6 before inducing by adding IPTG to 0.4 mM. Cells were allowed to express for 12–14 hr at 22°C before adding EDTA and PMSF to 2 mM and 0.5 mM, respectively. Cells were then harvested and lysed by sonication in lysis buffer (20 mM Tris pH 8.0, 140 mM NaCl, 2 mM EDTA, 1 mM DTT, 0.5 mM PMSF, and protease inhibitor tablets [Roche]) and then clarified by centrifugation (JA-20 rotor [Beckman], 18,000 rpm, 25 min, 4°C). The clarified lysate was then loaded onto a 10 mL GS4B column equilibrated in GST-binding buffer (20 mM Tris pH 8.0, 140 mM NaCl, 2 mM EDTA, and 1 mM DTT) and washed with ~10 column volumes of the same buffer. Protein was eluted with approximately three column volumes of elution buffer (20 mM Tris pH 8.0, 100 mM NaCl, 1 mM DTT, and 50 mM reduced L-glutathione) and fractions containing GST-SpWsp1-VCA were pooled and dialyzed overnight in 3500 MWCO tubing against 2 L of 20 mM Tris pH 8.0, 50 mM NaCl, and 1 mM DTT at 4°C in the presence of a 1:25 ratio (by mass) of TEV protease to recombinant protein. To purify the GST tagged Wsp1-VCA, the addition of TEV to the dialysis was omitted. The dialysate was loaded onto a 6 mL Source30Q column equilibrated in Q_A buffer (20 mM Tris pH 8.0, 50 mM NaCl, and 1 mM DTT) and eluted over a 20 column volume gradient to 100% Q_B buffer (20 mM Tris pH 8.0, 500 mM NaCl, and 1 mM DTT). Fractions containing GST-Wsp1-VCA were concentrated to 1.5 mL and loaded onto a Superdex 75 gel filtration column equilibrated in 20 mM Tris pH 8.0, 150 mM NaCl, and 1 mM DTT. Fractions containing pure protein were pooled and concentrated in a 3500 MWCO Amicon-Ultra centrifugal filter (Millipore Sigma) at 4°C. The concentrated, pure protein was flash frozen in liquid nitrogen. The final concentration of S. pombe Wsp1-VCA was determined by measuring the absorbance at 280 nm and dividing this by the Wsp1 extinction coefficient of 5500 M⁻¹cm⁻¹.

To construct an expression plasmid for S. pombe Wsp1-CA, residues 519–574 were cloned into the pGV67 vector containing an N-terminal GST tag followed by a TEV cleavage site. The purification was carried out as described for S. pombe Wsp1-VCA above.

To purify S. pombe Arp2/3 complex, 10 mL of a turbid culture of S. pombe (strain TP150) cells was added to a 2.8 L flask containing 1 L of YE5S. Cultures were grown for ~12 hr at 30°C with shaking followed by the addition of solid YE5S media, equivalent to the mass required for 1 L of liquid culture, and growth for ~4 more hours. Before harvesting, EDTA was added to a final concentration of 2 mM. All subsequent steps were carried out at 4°C. The cultures were centrifuged, and the pellet was resuspended in 2 mL of lysis buffer (20 mM Tris pH 8.0, 50 mM NaCl, 1 mM EDTA, and 1 mM DTT) per gram of wet cell pellet, plus six protease inhibitor tablets (Roche) per liter of lysis buffer. The resuspended cells were lysed in a microfluidizer (Microfluidics Model M-110EH-30 Microfluidizer Processor) at 23 kPSI over five to six passes. After lysis, PMSF was added to 0.5 mM and the lysate was spun down in a JA-10 (Beckman) rotor at 9000 rpm for 25 min. The supernatant was transferred to prechilled 70 mL polycarbonate centrifuge tubes and spun at 34,000 rpm for 75 min in a Fiberlite F37L rotor (Thermo-Scientific). The pellet was discarded and the supernatant was filtered through cheesecloth into a prechilled graduated cylinder to determine the volume. Under heavy stirring, 0.243 g of ammonium sulfate per milliliter of supernatant was added over approximately 30 min. The solution stirred for an additional 30 min, and then was pelleted in a Fiberlite F37L rotor at 34,000 rpm for 90 min. The pellet was resuspended in 50 mL of PKME (25 mM PIPES, 50 mM KCl, 1 mM EGTA, 3 mM MgCl2, 1 mM DTT, and 0.1 mM ATP) and dialyzed overnight in 50,000 MWCO dialysis tubing against 8 L PKME. The dialysate was clarified by centrifugation in the Fiberlite F37L rotor at 34,000 rpm for 90 min. A 10 mL column of GS4B beads was equilibrated in GST-binding buffer (20 mM Tris pH 8.0, 140 mM NaCl, 1 mM EDTA, and 1 mM DTT) before it was charged with 15 mg of GST-N-WASP-VCA to make a GST-VCA affinity column. The charged column was washed with additional binding buffer until no protein was detectable in the flow through by Bradford assay. The column was then equilibrated in PKME pH 7.0, the supernatant was loaded at 1 mL/min, and the column was washed with additional PKME (~45 mL). A second wash with PKME + 150 mM KCl was done until no protein was detected in the flow through by Bradford assay (~30 mL). Protein was eluted with PKME + 1 M NaCl into ~2 mL fractions until no protein was detected by a Bradford assay (~30 mL). Fractions containing Arp2/3 complex were pooled and dialyzed overnight in 50,000 MWCO dialysis tubing against 2 L of Q_A buffer (10 mM PIPES, 25 mM NaCl, 0.25 mM EGTA, 0.25 mM MgCl2, and pH 6.8 with KOH). Arp2/3 complex was further purified by ion exchange chromatography on an FPLC using a 1 mL MonoQ column with a linear gradient of Q_A buffer to 100% Q_B buffer (10 mM PIPES, 500 mM NaCl, 0.25 mM EGTA, 0.25 mM MgCl2, and pH 6.8 with KOH) over 40 column volumes. Fractions containing Arp2/3 complex were pooled and dialyzed overnight in 50,000 MWCO dialysis tubing against Tris pH 8.0, 50 mM NaCl, and 1 mM DTT. The dialysate was concentrated to 1.5 mL in a 30,000 MWCO concentrator tube (Sartorius Vivaspin Turbo 15 #VS15T21) using the Fiberlite F13B rotor at 2500 rpm over several 5–10 min cycles. Between each cycle the solution was mixed by gentle pipetting. The concentrated sample was loaded on a Superdex 200 HiLoad 16/60 gel filtration column equilibrated in Tris pH 8.0, 50 mM NaCl, and 1 mM DTT. Fractions containing pure Arp2/3 complex were concentrated as described above and the final concentration was determined by measuring the absorbance at 290 nm and dividing by 139,030 M⁻¹cm⁻¹, the extinction coefficient (ε₂₉₀) of Arp2/3 complex, before flash freezing.

Biotin-inactivated myosin was prepared by reacting 2 mg of rabbit skeletal muscle myosin (Cytoskeleton Cat # MYO2) with 5 μL of 250 mM EZ-Link-Maleimide-PEG11-Biotin dissolved in DMSO. The labeling reaction was carried out in 500 μL reaction buffer (20 mM HEPES pH 8.0, 500 mM KCl, 5 mM EDTA, 1 μM ATP, and 1 mM MgCl2) on ice for 6 hr. The biotin–myosin was then dialyzed against 0.5 L of storage buffer (20 mM imidazole pH 7.0, 500 mM KCl, 5 mM EDTA, 1 mM DTT, and 50% glycerol) using a 3500 MWCO dialysis thimble (Thermofisher Slide-A-Lyzer MINI dialysis unit 0069550). The final volume of biotin–myosin was measured, and the concentration was determined based on 1.85 mg/mL equaling 3.86 µM. Biotin–myosin was stored at −20°C.

Actin was purified using a modified method based on the one described by Spudich and Watt, 1971. Five grams of rabbit muscle acetone powder was resuspended in 100 mL G buffer (2 mM Tris pH 8.0, 0.2 mM ATP, 0.5 mM DTT, 0.1 mM CaCl₂, and 1 mM sodium azide [NaN₃]) and stirred at 4°C for 30 min. All additional steps are carried out at 4°C. Resuspended rabbit muscle was centrifuged for 25 min at 16,000 rpm in a Beckman JA-20 rotor and the supernatant was filtered through glass wool into a prechilled graduated cylinder. The pellet was resuspended in an additional 100 mL G buffer, pelleted as described above, and the supernatant was filtered through glass wool and pooled with the original supernatant. The total volume was measured and the solution was brought to 50 mM KCl and 2 mM MgCl₂ by adding 25 µL 2 M KCl and 2 µL 1 M MgCl₂ per 1 mL of supernatant with stirring. The solution was allowed to stir for 1 hr to polymerize actin. After 1 hr, 0.056 g KCl per 1 mL of original supernatant volume was added to bring the solution to 0.8 M KCl. The solution was stirred for an additional 30 min to dissociate tropomyosin before it was pelleted in a Beckman 70Ti rotor at 31,700 rpm for 2 hr. The supernatant was discarded and two-third of the pellet was homogenized in 5 mL of G buffer using a Dounce homogenizer and dialyzed in 1 L of G buffer to begin to depolymerize actin filaments. The remaining one-third of the pellet was homogenized in 4 mL of labeling buffer 25 mM Tris pH 7.5, 100 mM KCl, 0.3 mM ATP, 2 mM MgSO₄, and 3 mM sodium azide (NaN₃) and dialyzed against 1 L of labeling buffer for at least 4 hr to remove DTT. The concentration of actin was then measured ([actin] = A₂₉₀/0.0266 µM⁻¹cm⁻¹) and actin was diluted to 23.8 µM (1 mg/mL). A 4–7 molar excess of N-(1-pyrene)Iodoacetimide was added to the actin dropwise with stirring. The labeling reaction proceeded for ~14 hr, covered from light. Pyrene-labeled actin was pelleted using a Beckman 90Ti rotor at 36,200 rpm for 2 hr and then homogenized in 1.5 mL G buffer and dialyzed against 1 L G buffer while protected from light. Both unlabeled and pyrene-labeled actin were dialyzed against G buffer for 1.5–2 days with at least three exchanges of dialysis buffer. Following dialysis, depolymerized actin was centrifuged in a Beckman 90 Ti rotor at 36,200 rpm for 2 hr to pellet out any polymerized actin and the top 80% of the supernatant was the gel filtered using S-300 resin and G buffer. Fractions of 3–4 mL were collected over ~300 mL G buffer and the peak fraction was identified using a Bradford assay. The concentration of unlabeled actin was determined using the following equation: [Actin, µM] = A₂₉₀ × 38.5 µM/OD. The concentration of pyrene-labeled actin was determined using the following equation: [Pyrene-labeled actin, µM] = (A₂₉₀ – (0.127 × A₃₄₄)) × 38.5 µM/OD. The labeling percentage of pyrene-labeled actin was determined by dividing the pyrene-labeled actin concentration by the concentration of pyrene ([pyrene, µM] = A₃₄₄ × 45.5 µM/OD). Actin was stored at 4°C with pyrene-labeled actin covered in foil to protect from light.

Oregon Green-labeled actin was purified as pyrene actin until resuspension in labeling buffer. Actin was resuspended in 2 mL labeling buffer (25 mM imidazole pH 7.5, 100 mM KCl, 0.3 mM ATP, 2 mM MgCl₂, and 3 mM sodium azide (NaN₃)) and dialyzed against 1 L of labeling buffer for at least 4 hr to remove DTT. The concentration of actin was then measured ([actin] = A₂₉₀/0.0266 µM⁻¹cm⁻¹) and actin was diluted to 23.8 µM (1 mg/mL). A 10–12 molar excess of Oregon Green 488 Maleimide (Invitrogen O-6034) was added to the actin dropwise with stirring. The labeling reaction proceeded for ~14 hr, covered from light. Oregon Green-labeled actin was centrifuged, homogenized, and gel filtered as with pyrene actin. The concentration of pyrene-labeled actin was determined using the following equation: [Oregon Green-labeled actin, µM] = (A₂₉₀ – (0.16991 × A₄₉₁)) × 38.5 µM/OD. The labeling percentage of Oregon Green-labeled actin was determined by dividing the Oregon Green-labeled actin concentration by the concentration of Oregon Green dye ([Oregon Green 488 Maleimide, µM] = A₄₉₁/0.081 µM⁻¹cm⁻¹). Oregon Green-labeled actin was stored at 4°C covered in foil to protect from light.

Crosslinking assays were carried out as previously described (Rodnick-Smith et al., 2016b). Briefly, 15 μL reactions were initiated by adding BMOE to 20 μM to make a final solution that contained 1 μM dual cysteine SpArp2/3 complex in 10 mM imidazole pH 7.0, 50 mM KCl, 1 mM EGTA, 1 mM MgCl₂, 200 μM ATP plus Wsp1-VCA, Dip1, or latrunculin actin as indicated. Reactions were incubated at room temperature for 45 s before quenching with SDS-PAGE loading buffer with fresh DTT. Reactions were analyzed by western blotting using an anti-Arp3 antibody (1:1000).

TIRF flow chambers were constructed as previously described with slight modifications (Kuhn and Pollard, 2005). All following cleaning steps were carried out at room temperature. Coverslips (24 × 60 #1.5) were cleaned in Coplin jars by sonicating in acetone followed by 1 M KOH for 25 min each, with a deionized water rinse between each sonication step. Coverslips were then rinsed twice with methanol and aminosilanized by incubating in a 1% APTES (Sigma), 5% acetic acid in methanol solution for 10 min before sonicating for 5 min, and then incubating for an additional 15 min. Coverslips were then rinsed with two volumes of methanol followed by thorough flushing with deionized water. After air drying, TIRF chambers were created by pressing two pieces of double-sided tape onto a cleaned coverslip with a 0.5 cm wide gap between them. A glass microscope slide was then placed on top of the coverslip and tape perpendicularly to create a cross-shape forming a chamber in the middle with a volume of ~14 µL. Chambers were passivated by flowing in 300 mg/mL methoxy PEG succinimidyl succinate, MW5000 (JenKem) containing 1–3% biotin-PEG NHS ester, MW5000 (JenKem) dissolved in 0.1 M NaHCO₃ pH 8.3 and incubating for 4–5 hr. Excess PEG was washed out with 0.1 M NaHCO₃ pH 8.3 before flowing deionized water into chambers for storage. Chambers were stored at 4°C for less than 1 week. Immediately prior to imaging, 1 μM NeutrAvidin (ThermoFisher) was added to chambers and incubated for 8 min followed by 8 min with 50–150 nM biotin inactivated myosin (Cytoskeleton, Inc), both prepared in 50 mM Tris pH 7.5, 600 mM NaCl. Chambers were washed two times with 20 mg/mL BSA in 50 mM Tris pH 7.5, 600 mM NaCl followed by two washes with 20 mg/mL BSA in 50 mM Tris pH 7.5, 150 mM NaCl. Chambers were finally preincubated with TIRF buffer (10 mM imidazole pH 7.0, 1 mM MgCl₂, 1 mM EGTA, 50 mM KCl, 100 mM DTT, 0.2 mM ATP, 25 mM glucose, 0.5% methylcellulose [400 cP at 2%], 0.02 mg/mL catalase [Sigma], and 0.1 mg/mL glucose oxidase [MP Biomedicals]) after which point they were ready to add reaction mixture.

In a typical reaction, 1 μL of 2.5 mM MgCl₂ and 10 mM EGTA were mixed with 5 μL of 9 μM 33% Oregon Green-labeled actin and incubated for 2 min. Four microliters of the actin solution were then added to 16 μL of a solution containing 1.25× TIRF buffer and any other proteins. Reactions were imaged on a Nikon TE2000 inverted microscope equipped with a 100 × 1.49 numerical aperture TIRF objective, 50 mW 488 nm and 561 nm Sapphire continuous wave solid state laser lines (Coherent), a dual band TIRF (zt488/561rpc) filter cube (Chroma C143315), and a 1× to 1.5× intermediate magnification module. Images were collected using an 512 × 512 pixel EM-CCD camera (iXon3, Andor). For two color reactions, typical imaging conditions were 50 ms exposures with the 488 nm laser (set to 5 mW) and 100 ms exposures with the 561 nm laser (set to 35 mW) for 1 s intervals. The camera EM gain was set to 200. The concentration of 568-Dip1 was kept in the low nanomolar range in all assays to prevent high backgrounds of nonspecifically adsorbed 568-Dip1 from obscuring Dip-Arp2/3 filament nucleation events.

In a typical reaction, 2 μL of 10× ME buffer (5 mM MgCl₂ and 20 mM EGTA) was added to 20 μL of 15% pyrene-labeled actin and incubated for 2 min in 96-well flat bottom black polystyrene assay plates (Corning 3686). To initiate the reaction, 78 μL of buffer containing all other proteins was added to the actin wells using a multichannel pipette. This brought the final buffer concentration in the reaction to 10 mM imidazole pH 7.0, 50 mM KCl, 1 mM EGTA, 1 mM MgCl₂, 200 μM ATP, and 1 mM DTT. Polymerization of actin was measured by exciting pyrene actin at 365 nm and monitoring the emission at 407 nm using a TECAN Safire two plate reader. The maximum polymerization rate of pyrene actin polymerization assays was determined by measuring the slope of each curve at each time point and converting from RFU/s to actin (nM)/s by assuming that the actin filament concentration was zero at the minimum fluorescence value and 0.1 µM actin was unpolymerized at the maximum fluorescence. The maximum polymerization rate was plotted for a series of reactions with increasing concentrations of Dip1 and fixed concentrations of all other proteins. For these plots, data points were fit to the following equation: Max poly rate = (max poly rate_max× [Dip1])/(K_1/2 + [Dip1]) + y-intercept, where K_1/2 represents the [Dip1] needed to get half-maximum max polymerization rate and the y-intercept represents the maximum polymerization rate in the absence of Dip1. Note that Figure 2A shows only a subset of the assays while the maximum polymerization rates of the entire dataset are shown in Figure 2B. See Supplementary file 1 for details on fits.

The percentage of Dip1-Arp2/3 complex nucleated actin filaments was determined by counting the number of actin filament pointed ends bound by 568-Dip1 in Oregon Green-labeled actin polymerization assays imaged using TIRF microscopy (Balzer et al., 2018). The quantification was performed on a region of interest from the movie frame that corresponded to 2 min and 30 s from the initiation of the reaction. All pointed ends present in the quantified frame were tracked from their initial appearance to ensure accuracy in 568-Dip1 pointed end-binding determination. The number of pointed ends bound by 568-Dip1 was divided by the total number of pointed ends in the region of interest. At least four replicate actin polymerization assays were quantified for each condition. The statistical significance of the data for datasets of only two conditions was tested by Student’s t-test in GraphPad Prism. The statistical significance for datasets of more than two conditions was tested by one-way ANOVA with Tukey’s post-hoc test in GraphPad Prism. Both tests were two tailed and the significance values are reported in the figure legends.

All modeling was carried out using the open source software application COmplex PAthway SImulator (COPASI) (Hoops et al., 2006). Fluorescence values from time courses of polymerization of 3 µM 15% pyrene-labeled actin in the presence of indicated proteins were converted to .txt files using a custom MatLab script and loaded into COPASI software. The actin filament concentrations were determined by assuming 0.1 µM actin was unpolymerized at equilibrium (Pollard, 1986). Optimization of reaction parameters was carried out by simultaneously fitting all traces from a reaction set, using the genetic algorithm method in the parameter estimation module.

Models were built by identifying interactions between the components in polymerization assays to build up a set of reactions to describe the polymerization. For many parameters included in our set of reactions, we were able to use previously measured rate constants (see Supplementary file 2). Rate constants that had not been previously measured were allowed to float. We assumed pointed end elongation was negligible. To limit the number of floated parameters in a given simulation, we first conducted polymerization assays with actin alone at a range of concentrations (2–6 μM), and then determined a reaction pathway and rate constants that could accurately describe spontaneous nucleation and polymerization of actin alone (see Supplementary Figure 2). The on rates for actin dimerization, trimerization, and tetramerization were fixed at 1.16 × 10⁷ M⁻¹s⁻¹, the observed on rate for actin monomers binding to filament barbed ends. To simplify the models, steps that created a nucleus were considered irreversible and nuclei were modeled as catalysts that convert monomeric actin to filamentous actin, as previously described (Helgeson and Nolen, 2013; Beltzner and Pollard, 2008). We note that the best fits for spontaneous nucleation and elongation of actin filaments were obtained using a model in which either a dimer, trimer, or tetramer could serve as the nucleus. This pathway is distinct from models we previously used to simulate spontaneous nucleation and elongation in actin alone reactions (Helgeson and Nolen, 2013). To model reactions containing Dip1 and Arp2/3 complex (Figure 3), we fixed the rate constants for the spontaneous nucleation of actin at the values determined in reactions containing actin alone, except for k₋₁, which was re-evaluated based on an ‘actin alone’ polymerization time course measured at the same time (in the same set) as the Dip1 and Arp2/3-containing reactions. We used the objective value, or the normalized mean square weighted sum of squares, as a measure of how well the model fits the experimental data. The mean square weighted sum of squares (MSS) was determined by the following equation: $MSS=\Sigma \omega *(x-y)²$, where ${\displaystyle \omega =\frac{1}{<x^2>}}$ and x and y correspond to the experimental point in the dataset and simulated value, respectively.

S. pombe strains were constructed by PCR-based gene tagging and genetic crossing (see Supplementary file 4). Deletion cassettes with selectable markers were introduced into endogenous chromosomal loci by homologous recombination of PCR-amplified gene tagging modules (Bähler et al., 1998; Wach et al., 1994). Amplified modules were introduced into cells by lithium acetate transformation (Keeney and Boeke, 1994). Successful integrants were isolated based on selectable markers and confirmed by PCR and sequencing.

Fim1-mEGFP expressing cells were constructed in a previous study (Wu and Pollard, 2005). The dip1∆ strain was constructed by replacing the gene open reading frame with a ura4⁺ cassette amplified from KS-ura4 (Bähler et al., 1998). The C-terminal truncation of Wsp1 was constructed by replacing the sequence encoding the CA domain (aa 541–574) in the endogenous wsp1 locus with the stop codon-Tadh1-natMX6 cassette amplified from a custom-made pFA6a-GFP-natMX6 plasmid.

Strains combining two or more deletion or tagged alleles were constructed by standard genetic crosses on Malt Extract (ME) plates and tetrad dissection on YES plates followed by screening for wanted gene combinations by replica plating onto appropriate selective plates, microscopy, and PCR diagnostics.

Imaging was performed on an UltraView VoX Spinning Disc Confocal System (PerkinElmer) mounted on a Nikon Eclipse Ti-E microscope with a 100×/1.4 NA PlanApo objective, equipped with a Hamamatsu C9100-50 EMCCD camera, and controlled by Volocity (PerkinElmer) software. Stably integrated S. pombe strains were grown to OD₅₉₅ = 0.2–1.0 in liquid YES medium (Sunrise) while shaking in the dark at 25°C over 2 days. For microscopy, cells from 0.5 to 1 mL of culture were collected by a brief centrifugation in a microfuge at 2000 g and 5 μL of partly resuspended pellet was placed onto a pad of 25% gelatin in EMM on a glass slide, covered by a coverslip and sealed with VALAP (1:1:1 vaseline:lanolin:paraffin mix). Samples were imaged after a 5 min incubation to allow for partial depolarization of actin patches. For time-lapse imaging, single-color images in a cell medial plane were taken every second for 60 s. Z-series of images spanning the entire cell width were captured at 0.4 μm intervals.

Image analysis was performed in ImageJ (National Institutes of Health). To make sure that different mutant backgrounds did not alter the expression levels of tagged proteins and that imaging conditions remained stable throughout an imaging session, for each dataset, we measured average background-subtracted whole-cell intensities, which correspond to the tagged protein expression levels. For each time series, we measured whole cell intensities of five cells, subtracted these values for either extracellular background or the intensities of untagged wild-type cells, and averaged the background-subtracted values at each time point. All Fim1-GFP strains within each dataset had similar whole-cell intensities.

The intensities and positions of fluorescent protein-tagged markers in individual endocytic actin patches were manually tracked throughout lifetime of each patch using a circular region of interest (ROI) with a 10-pixel diameter. Mean fluorescence intensities of patches were subtracted for mean cytoplasmic intensities measured in cell areas away from patches, and distances traveled by patches from the original positions were calculated. Time courses of background subtracted intensities and distances from origin for individual patches were aligned to the time of peak intensity (time = 0) and averaged at each time point.

Percent of internalized patches was measured by following fates of all patches present in frame 25 of time-lapse movies in a single medial plane. A patch was counted as internalized if it shifted from its original position by more than two pixels and, for each cell, the number of internalized patches was divided by the total number of analyzed patches. Percent internalization was measured and averaged from 10 wild-type, 10 wsp1∆ cells, 22 dip1∆, and 20 wsp1∆CA dip1∆ cells.

To measure patch density, all patches in 10 cells were counted in a z-series taken at a single time point. The area of the cell was measured at the medial focal plane and the patch density was calculated by dividing the number of patches by the cell area. Patch initiation rates (patches/μm²/s) were measured by counting patches that newly appeared during the first 20 s of time-lapse movies taken in a single medial cell plane and dividing the number of new patches by cell area and 20 s. The patches that were already present in frame 1 were excluded from the count.

All fluorescence intensity, distance, patch initiation, and patch density values are displayed as mean ± SEM, with the exception of Figure 1B, which shows the mean ± standard deviation. The data are from single imaging experiment where all strains were imaged under identical conditions. The number of cells and patches analyzed are indicated in figure legends. The statistical significance for datasets of more than two conditions was tested by one-way ANOVA with Tukey’s post-hoc test in GraphPad Prism. Both tests were two tailed and the significance values are reported in figures and figure legends.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Research reported in this publication was supported by the National Institute of General Medical Sciences of the NIH under award number R35GM136319 (BJN) and T32 GM007759 (to CJB and LAH) and by the American Heart Association, grant no. 18PRE33960110 (CJB). We thank Andrew Wagner for generating the dip1∆ S. pombe strain and for help with the two-color TIRF microscopy experiments.

1. Received: June 26, 2020
2. Accepted: November 10, 2020
3. Accepted Manuscript published: November 12, 2020 (version 1)
4. Version of Record published: December 1, 2020 (version 2)

© 2020, Balzer et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.