Evolutionary tuning of barbed end competition allows simultaneous construction of architecturally distinct actin structures

Eukaryotic cells build multiple actin-based structures with distinct architecture and dynamics, all within a shared cytosol, allowing actin to be assembled into diverse structures with wide-ranging functions. For example, at the cortex, Arp2/3 complex assembles arborized networks comprised of relatively short filaments (∼30–100 nm long), used to drive endocytosis and lamellipodial extension (Bear et al., 2002; Rodal et al., 2005; Collins et al., 2011), while formins and Ena/VASP (Enabled/vasodilator-stimulated phosphoprotein) assemble networks consisting of longer (∼300–1,000 nm) unbranched filaments, such as filopodia, stress fibers, cytokinetic rings, microvilli, stereocilia, and sarcomeres (Burkholder and Lieber, 2001; Rigort et al., 2012; Swulius et al., 2018). These observations raise a fundamental question that has remained largely unanswered: What are the molecular mechanisms that permit cells to simultaneously build and maintain diverse structures from a common pool of building blocks? A crucial first step in answering this question has been the decades-long effort of defining the activities of individual actin-binding proteins (Chesarone et al., 2010; Edwards et al., 2014; Shekhar et al., 2016). One of the next major challenges is to determine how these actin regulatory proteins function collectively to specify the construction of distinct actin structures side-by-side in the cytoplasm. Recent progress in understanding this question has come largely from in vitro experiments defining the cooperative or competitive relationships of specific pairs of filament side-binding and crosslinking proteins, and how these relationships direct the sorting of different actin-binding proteins to different actin networks in cells (Kadzik et al., 2020).

Two of the central control points in actin assembly are the nucleation and elongation of filaments (Chesarone et al., 2010; Campellone and Welch, 2010). Nucleation is seeded by machinery such as Arp2/3 complex and formins paired with nucleation-promoting factors (Chesarone et al., 2010; Campellone and Welch, 2010). However, once nucleated, both the speed and duration of filament growth can be further controlled by factors that interact with barbed ends. For instance, in the branched actin networks nucleated by Arp2/3 complex, filament elongation is attenuated by the WH2 domains of nucleation-promoting factors interacting with barbed ends and then terminated by capping protein (CP), restricting filament length to maintain a densely branched architecture (Sweeney et al., 2015; Bieling et al., 2018; Tang et al., 2020; Funk et al., 2021). In other settings, elongation-promoting factors such as formins and Ena/VASP interact processively with filament ends to accelerate elongation and antagonize CP, thus increasing the duration of growth (Breitsprecher and Goode, 2013; Faix and Rottner, 2022). Despite our understanding of their individual effects, little is known about how these barbed end–associated factors work together to govern actin elongation, or how their relationships with each other drive the assembly of distinct networks.

Saccharomyces cerevisiae is an optimal model for investigating these questions of how barbed end regulators cooperate and/or compete to build distinct types of actin networks because it contains only two major actin structures throughout most stages of the cell cycle: cortical patches nucleated by Arp2/3 complex and polarized cables nucleated by formins. Patches and cables are both essential for cell viability, but perform distinct functions (endocytosis and polarized secretion, respectively). They are also strikingly different in other aspects, including: (1) Their size and shape; patches are only ∼0.15 µm in diameter whereas cables are ∼6 µm long. (2) Their filamentous architectures; patches are densely arborized networks consisting of short, branched filaments, whereas cables are long bundles of parallel filaments. (3) Their growth rates; patches polymerize actin at ∼0.05 µm/s (Kaksonen et al., 2003), whereas cables polymerize almost 10 times faster, at ∼0.3–0.5 µm/s (Yang and Pon, 2002; McInally et al., 2021). (4) Their decoration by distinct subsets of actin-binding proteins (Moseley and Goode, 2006; Goode et al., 2015). These contrasting features of patches and cables, combined with the fact that there are relatively few barbed end regulators in yeast (CP and formins having the primary roles), make it an ideal system for addressing the problem of how cells simultaneously construct different actin structures.

CP is a conserved heterodimer composed of α and β subunits (Cooper and Sept, 2008). Both subunits have C-terminal helical extensions, or “tentacles,” which bind to the two actin subunits exposed at the barbed end (Yamashita et al., 2003; Wear et al., 2003; Narita et al., 2006; Funk et al., 2021). S. cerevisiae has a single gene encoding each subunit (Cap1 and Cap2), whereas vertebrates express multiple isoforms. However, vertebrate CP isoforms have similar affinities for barbed ends and often are collectively referred to as “CapZ.” Although the dwell time of vertebrate CapZ at barbed ends is 20–30 min in vitro, each of its individual tentacles associates and dissociates more rapidly (Cooper and Sept, 2008), a property that may be crucial in enabling other barbed end interactors (e.g., formin Dia1 and twinfilin) to promote CapZ dissociation from filament ends (Bombardier et al., 2015; Shekhar et al., 2015; Hakala et al., 2021). In S. cerevisiae, capping protein (Cap1/2) localizes to cortical actin patches, where it limits patch growth and maintains normal patch lifetimes and movements (Kim et al., 2004, 2006; Kaksonen et al., 2005). However, it is not clear if or how yeast Cap1/2 might be displaced from barbed ends or how it cooperates and/or competes with other barbed end–associated proteins in a shared cytoplasm.

S. cerevisiae express two formins, Bni1 and Bnr1, which have distinct localization patterns and perform genetically redundant roles in assembling polarized actin cables that direct transport of vesicles and organelles to the bud (Pruyne et al. 2002). Like other formins, Bni1 and Bnr1 use their conserved formin homology 2 (FH2) domains to directly nucleate actin assembly and remain processively attached to the growing barbed end (Otomo et al., 2005; Baker et al., 2015), while their adjacent FH1 domains recruit profilin-bound actin monomers to accelerate elongation (Castrillon and Wasserman, 1994; Higgs and Peterson, 2005; Higgs, 2005; Pruyne, 2016; Courtemanche and Pollard, 2012). Early studies using bulk assays showed that yeast formins continue to elongate filaments in the presence of capping protein (Zigmond et al., 2003; Kovar et al., 2005; Moseley and Goode, 2005), and similar effects were seen for mammalian formins and CP (Harris et al., 2004). These observations led to the oversimplified conclusion that formin elongation is unaffected by CP. This view was grounded in the assumption that formin and CP exchange places only through dissociative competition, i.e., one does not associate with the barbed end until the other has dissociated. However, more recently our understanding of the formin–CP relationship has evolved. Using in vitro total internal reflection fluorescence (TIRF) microscopy, vertebrate CapZ and formins (mDia1 and FMNL2) were observed to join each other at the barbed end to form “decision complexes,” and accelerate each other’s dissociation by 10–50-fold in each direction (Bombardier et al., 2015; Shekhar et al., 2015). The associative competition mechanism could enable cells to promote rapid transitions, possibly back-and-forth, between states of actin filament growth (formin-bound) and arrest (CP-bound) to facilitate finer control of network growth. However, in vivo evidence for this mechanism has been lacking. Further, it has been unclear whether the formin–CP associative competition mechanism extends to other organisms.

To address these questions, we investigated the interplay between S. cerevisiae Cap1/2 and formins (Bni1 and Bnr1) in vitro and in vivo. Our data show that yeast formins and Cap1/2 compete for barbed ends in vitro via an associative competition mechanism, but that yeast Cap1/2 has a much slower on-rate for formin-capped barbed ends compared with vertebrate CapZ. These biochemical differences have important consequences for Cap1/2 in vivo function. They allow high concentrations of Cap1/2 to be maintained in the cytosol, to limit the growth of actin patches while not interfering with formin-mediated cable assembly. In addition, we find that Cap1/2 is required for preventing actin patches from growing abnormally large and inappropriately recruiting formins and tropomyosin, thus taking on cable-like features. Overall, our findings suggest that the competitive relationship between Cap1/2 and formins has been evolutionarily tuned to allow yeast cells to simultaneously construct two architecturally distinct actin networks optimized for different cellular functions.

To investigate the relationship between yeast Cap1/2 and formins, we started by asking whether Cap1/2 antagonizes yeast-formin-mediated actin assembly in vitro. In bulk assays, we tested the ability of Cap1/2 to attenuate the growth of actin filaments nucleated and elongated by the FH1-FH2-C fragments of Bni1 (Fig. 1 A) and Bnr1 (Fig. 1 B). Variable concentrations of Cap1/2 were introduced into the reactions shortly after the initiation of formin-mediated actin assembly, resulting in concentration-dependent attenuation of filament growth (Fig. 1, C and D). These results suggest that yeast Cap1/2 directly antagonizes Bni1 and Bnr1, consistent with an associative competition mechanism. We also found that vertebrate CapZ attenuates yeast formins, and that yeast Cap1/2 attenuates vertebrate formin (mDia1; Fig. S1). Thus, it appears that at least some of the key features of the CP–formin competitive relationship are conserved across distant species. On the other hand, we note that CapZ was more effective than yeast Cap1/2 in attenuating yeast formins, e.g., compare effects at 100 nM of CapZ versus Cap1/2 (Fig. 1). Therefore, there are also key differences in the properties of CapZ and Cap1/2, which alter their abilities to antagonize formins at the barbed end.

To gain further insights into the Cap1/2–formin mechanism at barbed ends, we used microfluidics-assisted TIRF (mf-TIRF) microscopy to directly observe the effects of Cap1/2 on formin-mediated elongation. The advantages of this technique are that it allows rapid changes in reaction components and keeps filaments aligned by flow, allowing more accurate quantification of elongation rates (Carlier et al., 2014; Shekhar, 2017; Romet-Lemonne et al., 2018). In our experimental setup, actin filaments are polymerized from spectrin–actin seeds anchored to the coverslip with their growing barbed ends free in solution. Proteins are then flowed in and their effects are monitored over time (Fig. 2, A and B). To best recapitulate the in vivo environment in these experiments, we used 100% S. cerevisiae actin, formins, profilin, and Cap1/2. To visualize filaments, we included a fluorogenic probe, SiR-actin, and found that SiR-labeled filaments elongate at only a slightly slower rate compared with control filaments (Fig. S2, A–C).

To measure the effects of Bni1 and Bnr1 (FH1-FH2-C fragments) alone on barbed end elongation (no Cap1/2), we briefly introduced a low concentration of formin to associate with most of the barbed ends, leaving a small fraction of barbed ends free as internal controls for the elongation rate at free barbed ends. Elongation rates were monitored in the constant presence of 1 µM monomeric yeast actin and 2 µM profilin. Bni1 and Bnr1 each accelerated elongation, ∼3.5- and -3.0-fold faster, respectively, compared with free barbed ends in the same reactions (Fig. 2 C). Next, we confirmed that flowing in yeast Cap1/2 (100 nM) rapidly halts elongation, and measured the percentage of filaments that were capped over time. From these data, we derived an on-rate (k⁺) for Cap1/2 of 1.6 ± 0.3 × 10⁶ M⁻¹s⁻¹ (Fig. 2 D), which agrees well with the on-rate previously determined in bulk assays (Kim et al., 2004; Fig. 2 E). Interestingly, the on-rates of S. cerevisiae and Mus musculus CP are similar to each other, whereas the on-rate of Schizosaccharomyces pombe CP is much slower. However, the off-rates (k⁻) of S. cerevisiae and S. pombe CP are similar to each other, while the off-rate of M. musculus is much slower (Fig. 2 E). Thus, the binding kinetics of CP vary considerably across species.

We also characterized the processivity of Bni1 and Bnr1 at barbed ends, again using all S. cerevisiae proteins. The processivity of Bnr1 has never been measured, and the processivity of Bni1 has only been measured using vertebrate muscle actin (Paul and Pollard, 2008). In mf-TIRF assays, we briefly flowed in Bni1 or Bnr1 to associate with barbed ends, which was verified by their increase in elongation rate. We then washed out free formins and monitored the duration of accelerated elongation. From these data, we calculated the rate of formin dissociation (k⁻) from barbed ends. This analysis showed that Bni1 and Bnr1 are both highly processive (k⁻ = 2.1 ± 0.2 × 10⁻⁴ and 4.4 ± 0.3 × 10⁻⁴ s⁻¹, respectively), and, on average, elongate barbed ends for 4,740 and 2,220 s, respectively, before spontaneously dissociating (Fig. 2 F).

We next asked how the presence of Cap1/2 (variable concentrations) influences the duration of Bni1- and Bnr1-mediated elongation. At 1,000 nM Cap1/2, similar to the concentration of Cap1/2 in yeast cells (Kim et al., 2004), the mean duration of formin-mediated elongation was reduced to 624 and 200 s for Bni1 and Bnr1, respectively (Fig. 2, G and H). From these data, we calculated the on-rate (k⁺) of Cap1/2 at barbed ends already occupied by yeast formins (Fig. 2 I). This revealed that Cap1/2 binds very slowly to formin-capped ends, ∼1,000- and 300-fold slower (for Bni1- and Bnr1-bound ends, respectively) compared with free barbed ends (Fig. 2 D). By this same approach, we also calculated the on-rate (k⁺) of vertebrate CapZ to Bni1- or Bnr1-bound barbed ends and found that CapZ binds 27- and 130-fold faster to Bni1- and Bnr1-bound barbed ends, respectively, compared with Cap1/2 (Fig. S1, E–G). These observations are consistent with our bulk assays showing vertebrate CapZ more effectively inhibits formin-mediated actin assembly compared to yeast Cap1/2 (Fig. 1).

Finally, we asked the reciprocal question, do yeast formins promote the dissociation of yeast Cap1/2 from filament ends? Previous studies have shown that vertebrate formins catalyze dissociation of vertebrate CP (CapZ) from barbed ends (Bombardier et al., 2015; Shekhar et al., 2015); however, it has remained unclear whether these effects are evolutionarily conserved. To address this question, we used two separate mf-TIRF approaches. First, we assembled spectrin-anchored filaments with free barbed ends and capped them with Cap1/2. Capped filaments were then challenged for 50 s with Bni1 or Bnr1 (1,000 nM), or control buffer, and returned to a flow containing 1 µM actin and 2 µM profilin (no formins) and monitored for growth. Bni1 and Bnr1 had little effect compared with control buffer on Cap1/2 dissociation, and in all conditions, the off-rate of Cap1/2 was ∼1 × 10⁻² s⁻¹. (Fig. S2 D). Second, we assembled capped filaments as above, then exposed them for 10 min to a flow containing Bni1 (2,500 nM) or Bnr1 (2,000 nM), or control buffer (without actin), and monitored transitions from capping to depolymerization as a read-out for Cap1/2 dissociation. From these data, we calculated an off-rate for Cap1/2 in control buffer that was similar to the off-rate in the first assay, and once again, Bni1 and Bnr1 had minimal effects on Cap1/2 displacement (Fig. S2 E). Thus, yeast formins are surprisingly ineffective in competitively displacing Cap1/2.

Together, our in vitro observations demonstrate that S. cerevisiae Cap1/2 and formins compete for barbed ends via an associative competition mechanism, but with strikingly different kinetics from vertebrate CP and formin (Bombardier et al., 2015; Shekhar et al., 2015). For instance, whereas 1 µM vertebrate CapZ reduces the average time of mDia1 processive elongation by >1,000-fold (from ∼5,000 to ∼3 s), 1 µM yeast Cap1/2 had only modest effects on Bni1 and Bnr1 processive elongation (from ∼4,740 to ∼624 s and from ∼2,220 to ∼200 s, respectively). Moreover, while vertebrate CapZ and formins mutually catalyze each other’s dissociation from barbed ends, yeast formins have minimal effects in displacing Cap1/2 from barbed ends.

Our in vitro observations suggest key differences in the properties of yeast Cap1/2 versus vertebrate CP and how they influence formin-mediated elongation at barbed ends. To test the biological significance of these differences, we compared the effects of overexpressed yeast versus vertebrate CP on actin cable assembly in vivo. To overexpress Cap1/2 and CapZ, we integrated GFP-tagged constructs (as additional copies to endogenous Cap1/2), with their expression under the control of the galactose-inducible Gal1/10 promoter. Upon induction, Cap1/2 and CapZ levels were elevated above endogenous Cap1/2 by ∼30- and 20-fold, respectively (Fig. S2, F and G). We then compared their effects on F-actin levels measured by spinning-disk confocal microscopy, which provides increased detection of fluorescence signal (Fig. 3, A and F), and on cellular F-actin organization imaged by super-resolution Airyscan microscopy (Fig. 3, B and G). Our results show that overexpression of Cap1/2 has no obvious effect on the appearance of F-actin organization (Fig. 3, A and B), nor does it significantly alter total levels of F-actin in cells, the number of actin cables, or thickness of cables (Fig. 3, C–E). Thus, yeast cable networks are highly resistant to elevated levels of Cap1/2. On the other hand, overexpression of CapZ reduced total F-actin levels in cells (Fig. 3 H) and decreased cable numbers and thickness (Fig. 3, I and J). However, the F-actin levels in individual cortical patches (visualized using fimbrin/Sac6-RFP^mScarlet) were unchanged in cells overexpressing either CapZ or Cap1/2 (Fig. S2, H and I). These in vivo observations closely mirror our in vitro results and indicate that CapZ attenuates yeast formins whereas Cap1/2 does not. The broader implication of these results is that they suggest the specific properties of yeast Cap1/2 have been tuned across evolution to allow robust formin activity in the presence of high cytosolic levels of Cap1/2 used to restrict patch growth.

We next asked whether Cap1/2 decorates cables in vivo. While it has long been known that Cap1/2 localizes to actin patches (Amatruda and Cooper, 1992), its potential decoration of cables has not yet been carefully examined using enhanced fluorescent tags and microscopy techniques optimized for detecting low signal over a high cytosolic background. Therefore, we integrated three tandem mNeonGreen tags (3xmNeon), a brighter and more photostable tag than GFP, at the C-terminus of Cap2 at its endogenous locus. By Airyscan microscopy, we observed strong Cap2–3xmNeon signal on cortical patches, and, as expected, this signal disappeared upon treatment of cells with CK666 (an Arp2/3 inhibitor; Fig. 4 A). To further improve the signal-to-noise ratio, we examined Cap2–3xmNeon localization by TIRF microscopy and observed bright signal on cortical patches in addition to more faint and highly dynamic puncta at the cortex, which did not overlap with patches (Fig. 4 B and Video 1). We were unable to track the movements of the faint dynamic puncta because of signal interference from the bright patches. However, after CK666 treatment to remove patches, we were able to track the faint Cap2–3xmNeon puncta, and we found that their velocities (∼2.5 µm/s) were similar in cells treated either with CK666 or latrunculin (which removes both patches and cables; Fig. 4, C and D; and Video 2). Thus, the puncta do not appear to be associated with patches or cables. Instead, they may represent Cap1/2 molecules rapidly diffusing near the cell cortex, possibly interacting with phospholipids (Amatruda and Cooper, 1992), or associated with the dynactin complex (Eckley et al., 1999). These results indicate that Cap1/2 is largely excluded from formin-generated actin cables.

Early immunofluorescence studies failed to detect Cap1/2 at the cytokinetic actin ring in fixed cells (Amatruda et al., 1990). However, we detected a strong Cap2–3xmNeon signal at the cytokinetic ring by live imaging (Fig. 4 A). Further, this signal was refractive to CK666 treatment, as expected, since cytokinetic rings are polymerized by formins and not by the Arp2/3 complex (Moseley and Goode, 2006). To our knowledge, this represents the first demonstration of Cap1/2 localization to the cytokinetic ring in budding yeast. Importantly, this agrees with studies in fission yeast, where capping protein localizes to the cytokinetic ring (Kovar et al., 2005). Overall, our results show that Cap1/2 is a prominent component of cortical actin patches and cytokinetic rings, but is absent from cables, and therefore is unlikely to play a major structural role in cable assembly.

Our in vitro observations showing that Cap1/2 protects barbed ends from yeast formins motivated us to next ask whether Cap1/2 normally protects patches from formin decoration. To address this, we examined the localization of Bni1 and Bnr1 (endogenously tagged with three tandem copies of mNeon) in wild-type and cap2Δ strains. Note that these 3xmNeon-tagged formins are much brighter and more photostable than GFP-tagged versions (Video 3; and Fig. S3, A and B). In wild-type cells, dynamic Bni1–3xmNeon puncta were present throughout the cytosol and enriched at the bud tip, agreeing with previous results using Bni1–3xGFP (Buttery et al., 2007; Fig. 5 A). Further, the majority of the Bnr1–3xmNeon signal was at the bud neck, similar to observations using Bnr1–GFP (Fig. 5 B; Buttery et al., 2007). However, using our enhanced fluorescent tags, we were able to detect an additional set of dynamic Bnr1–3xmNeon puncta in both the mother and bud compartments. To address whether the formin puncta might colocalize with cortical patches, we coimaged the 3xmNeon-tagged formins with Arc15-RFP^mScarlet (a patch marker) in wild-type and cap2Δ cells. TIRF microscopy was used to minimize photobleaching and to increase the signal-to-noise ratio at the cell cortex. In wild-type cells, Bni1 and Bnr1 puncta did not colocalize with patches (Fig. 5, C–H, Fig. S3, C–F, and Videos 4 and 5). On the other hand, in cap2Δ cells, most of the actin patches (75.5%) showed colocalization with Bni1–3xmNeon puncta (Fig. 5, C, E, and G; and Fig. S3, C and E). A smaller fraction of the patches (9.6%) showed colocalization with Bnr1–3xmNeon (Fig. 5, D, F, and H; and Fig. S3, D and F), possibly explained by the much lower levels of Bnr1–3xmNeon puncta compared with Bni1–3xmNeon puncta in the cytosol (compare Fig. 5, A and B). These striking differences between wild-type and cap2∆ cells suggest that Cap1/2 plays an instrumental role in protecting cortical patches from formins. Related observations have been made in fission yeast, where loss of CP leads to recruitment of the formin For3 to patches and weaker recruitment of Cdc12 (Billault-Chaumartin and Martin, 2019). Thus, the role of CP in protecting patches from formins appears to be conserved between budding and fission yeast, despite these two species being as distant evolutionarily from each other as they each are from animals (Sipiczki, 2000).

Importantly, we also asked whether the cap2∆ phenotype of overgrown/enlarged cortical actin patches (Amatruda et al., 1990; Amatruda and Cooper, 1992; Shin et al., 2018; Antkowiak et al., 2019) results from ectopic formin activity. To address this question, we measured F-actin levels at cortical patches in cap2Δ cells in a background in which formin activity can be conditionally shut off (bnr1∆ bni1-1). F-actin overgrowth at patches was still observed even at the non-permissive temperature (Fig. S4, A–C), indicating formin-independent growth of actin patches in cap2∆ cells. Thus, Cap1/2 protects patches from excessive barbed end assembly (which is formin-independent) and additionally shields patches from formin decoration, which may contribute to the cap2∆ phenotype.

Because formins are recruited to patches in cap2Δ cells, we next asked whether patches recruit other cable-associated proteins. In wild-type cells, the tropomyosin Tpm1 decorates and stabilizes formin-polymerized cables, but is excluded from patches (Pruyne et al., 1998). We designed an mNeon-tagged Tpm1 based on the strategy recently described by Balasubramanian and colleagues (Hatano et al., 2022 Preprint), integrated this construct in wild-type and cap2∆ cells (at a different locus from endogenous TPM1), and expressed it under the control of the native TPM1 promoter. In wild-type cells, mNeon–Tpm1 localized to dynamic cables that are distinct from patches marked with Arc15-RFP^mScarlet (Fig. 6 A and Video 6). However, in cap2Δ cells, most of the mNeon–Tpm1 signal was on patches rather than cables (Fig. 6 A and Video 3). We asked whether recruitment of mNeon–Tpm1 to patches in cap2Δ cells results from ectopic formin activity at patches (Fig. 5) by imaging mNeon–Tpm1 and the patch marker Arc15-RFP^mScarlet in cap2∆ cells in a background where formin activity can be conditionally shut off (bnr1∆ bni1-1). After formin inactivation by temperature shift, mNeon–Tpm1 signal was still found at Arc15-RFP^mScarlet-labeled patches (Fig. S4 H). Thus, mNeon–Tpm1 can be recruited to patches in cap2∆ cells independent of formin activity. In cap2Δ cells, we also noticed that some of the patches were more heavily decorated by mNeon–Tpm1 than others (Fig. 6 A). To address whether this heterogeneity might arise due to temporal recruitment of mNeon–Tpm1 at specific stages of patch development, we performed live imaging on patches in cap2∆ cells, monitoring the time of arrival of mNeon–Tpm1 in relation to Arp2/3 complex (Arc15-RFP^mScarlet). We observed that mNeon–Tpm1 arrives at patches late in their lifetimes and persists there even as Arp2/3 levels decline; eventually many of these mNeon–Tpm1-marked patches move out of view (Fig. 6 B, Fig. S4 D, and Video 6). The result is an ∼13-s delay between the peak of Arc15-RFP^mScarlet signal and the peak of mNeon–Tpm1 signal in cap2Δ cells (Fig. 6 C), i.e., in cap2Δ cells patches remain decorated with mNeon–Tpm1 even after Arp2/3 has left (presumably by debranching mechanisms).

We also examined the effects of cap2∆ on recruitment of Abp140-3xmNeon, which decorates patches and cables, and Sac6-GFP^Envy, which exclusively decorates patches (Doyle and Botstein, 1996; Yang and Pon, 2002). In wild type cells, Abp140 and Sac6 arrive and depart from patches with similar kinetics to Arp2/3 (Arc15). However, in cap2∆ cells, we observed that Abp140 and Sac6 persist on patches even as Arp2/3 (Arc15) is departing, albeit not as long as Tpm1 (Fig. 6, B–G and Fig. S4, D–G). Our interpretation of this observation is that the composition of aging patches may be different in cap2∆ cells compared with wild-type cells. In cap2∆ cells, the actin network persists (marked by Tpm1) even after Arp2/3 (Arc15) leaves, suggesting little or no branching remains at that point.

These observations are consistent with filaments growing abnormally long in cap2∆ cells, and the patches accumulating more F-actin (as indicated by Sac6 and Abp140 levels continuing to increase) while changing in architecture (as indicated by Arp2/3 levels declining). Overall, our results show that in cap2∆ cells, as patches age they undergo a patch-to-cable-like transition, where cable markers (formins and Tpm1) appear as patch markers (Arc15 first, then Sac6 and Abp140) decline (Fig. 6 H). We propose that patches in wild-type and cap2Δ cells start off similarly, with Arp2/3-mediated nucleation of branched filaments. However, as the patches mature in cap2∆ cells, they form abnormally long filaments (which recruit Tpm1 to their sides) with free barbed ends (which recruit formins), and thus patches become more cable-like (Fig. 6 H).

Besides patch overgrowth, cap2∆ cells show diminished cable staining (Fig. 7 A), which has been reported in previous studies (Amatruda et al., 1990; Amatruda and Cooper, 1992; Shin et al., 2018; Antkowiak et al., 2019). To our knowledge, the mechanism underlying diminished cable staining in cap2∆ cells has not been addressed. We considered whether this phenotype might result either from altered Cap1/2–formin competition at barbed ends or Cap1/2 protecting barbed ends in cables from disassembly. However, our in vitro TIRF analysis (Fig. 2), in vivo Cap1/2 overexpression analysis (Fig. 3), and absence of Cap1/2 decoration on cables (Fig. 4) argue against these models. An alternative model to explain the diminished cables is that excessive patch growth in cap2∆ cells depletes the actin monomer pool, impairing cable assembly. To test this possibility, we asked whether CK666 treatment of cap2∆ cells to remove patches could rescue cable defects (Burke et al. 2014). In the absence of CK666, cap2Δ cells had fewer and thinner (dimmer) cables despite having ∼40% more total cellular F-actin compared with wild-type cells (Fig. 7, A–D). This excess F-actin is incorporated into patch networks resulting in a shift in the distribution of F-actin between patches and cables (Fig. 7 E). However, after CK666 treatment, wild-type and cap2Δ cells had similar levels of cables (number and thickness; Fig. 7, F and G). Further, cable network function was restored, as indicated by polarized distribution of a secretory vesicle marker (mCherry-Sec4) to the bud (Fig. 7, I–L). These observations support the alternative model, that in cap2∆ cells excessive actin incorporation into patches leads to diminished cables.

Interestingly, cap2∆ cells still showed higher total F-actin levels compared with wild-type cells after CK666 treatment (Fig. 7 H) and contained aberrant, variably shaped F-actin clumps and aster-like structures (Fig. 7 A and Fig. S5 C). We verified the absence of Arp2/3 complex (Arc15-RFP^mScarlet) from these structures after CK666 treatment (Fig. 7 A and Fig. S5, A and B). It is possible that these abnormal, CK666-resistant F-actin structures in cap2∆ cells represent the above-mentioned aberrant patches that take on cable-like features (Fig. 6 H).

In this study, we set out to determine whether the associative competition mechanism described in vitro between vertebrate formins and capping protein (CapZ) extends evolutionarily to S. cerevisiae and what physiological purpose it serves. Our results reveal surprising differences in the activities of vertebrate CapZ versus S. cerevisiae Cap1/2, which appear to have important in vivo consequences. Although S. cerevisiae Cap1/2 is capable of halting actively elongating formin-bound barbed ends, indicating that the associative competition mechanism is conserved at some level, Cap1/2 is unexpectedly weak in its ability to displace formins. While Cap1/2 and CapZ both rapidly bind to free barbed ends, Cap1/2 is relatively slow to associate with formin-bound barbed ends compared with CapZ. For example, Cap1/2 associates with Bni1-capped barbed ends ∼130-fold slower than CapZ (k⁺ = 1.6 × 10³ M⁻¹ s⁻¹ and 2.1 × 10⁵ M⁻¹ s⁻¹, respectively). Our data along with previous studies (Bombardier et al., 2015; Shekhar et al., 2015) suggest that at its cellular concentration of ∼1–2 μM (Dinubile et al., 1995; Nachmias et al., 1996), CapZ is predicted to join a formin at the barbed end approximately two to four times per second, and thus be capable of rapidly pausing formin-mediated elongation and dissociating formins from barbed ends. In contrast, we have shown that S. cerevisiae Cap1/2 has a much slower on-rate for formin-bound barbed ends, and at its cellular concentration of ∼1 μM (Kim et al., 2004), Cap1/2 is predicted to join Bni1 or Bnr1 at the barbed end only once every 4 min. Interestingly, despite this greater than two orders of magnitude difference in the on-rates of CapZ versus Cap1/2 for formin-bound barbed ends, there is only a sixfold difference in their on-rates for free barbed ends (CapZ k⁺ = 6.3 μM^-1 s^-1; Cap1/2 k⁺ = 1.2 μM^-1 s^-1; Wear et al., 2003; Kim et al., 2004). As discussed below, these unique aspects of their binding kinetics have critical implications for yeast actin cable formation and lend new insights into how different organisms tune the CP–formin competitive relationship to support cell type–specific actin assembly requirements.

To address the physiological importance of yeast Cap1/2–formin competition, we compared how overexpressing versus deleting Cap1/2 affects formin-mediated cable assembly. We found that 30-fold overexpression of Cap1/2 (from ∼1.3 to 39 µM) caused no obvious change in cable networks. This observation agrees well with our in vitro results showing that Cap1/2 is slow to associate with Bni1- or Bnr1-bound barbed ends. Cables are highly dynamic structures that extend at ∼0.3–0.4 µm/s (∼150 subunits/sec), and each cable is composed of bundled, overlapping filaments ∼0.5 µm in length (Kamasaki et al., 2005; Yu et al., 2011; Swulius et al., 2018; McInally et al., 2021). Based on these properties, a formin molecule might only be active for <2 s to elongate one of the filaments in a cable. In cells overexpressing Cap1/2 (39 µM), Cap1/2 would join a formin at the barbed end only once per 5–15 s, which would rarely interfere with the <2-s active periods of formin polymerization. In contrast, overexpression of vertebrate CapZ resulted in fewer and thinner cables. Again, this is explained by the distinct association kinetics of CapZ, which has a much faster on-rate (compared with yeast Cap1/2) for barbed ends. These major differences between vertebrate and yeast CP (rather than vertebrate and yeast formins) result in important changes to the formin–CP relationship. They suggest that the specific properties of S. cerevisiae Cap1/2 allow robust formin activity (to build cables) in the presence of a high cellular concentration of Cap1/2 used to restrict actin patch growth. Thus, the formin–CP relationship can be tuned in different organisms to allow simultaneous assembly of actin structures with distinct architectures.

These observations raise another important question. If CapZ can rapidly join formin-bound barbed ends and displace formins, are there mechanism(s) in vertebrate cells that protect formins from excessive inhibition by CapZ? One possibility is that the CP inhibitor V1 (myotrophin) helps in this capacity. V1 directly binds CP and blocks its interactions with barbed ends (Bhattacharya et al., 2006; Takeda et al., 2010; Zwolak et al., 2010; Fujiwara et al., 2014). V1 is abundant in cells and based on its binding affinity, it is estimated to be inactivating >90% of CapZ in the cytosol (Taoka et al., 2003; Bhattacharya et al., 2006; Fujiwara et al., 2014). The large pool of inactive CapZ can be rapidly accessed to replenish the active pool of CapZ levels after local depletion, but it might also serve to temper the inhibitory effects CapZ has on formins. Indeed, knocking down V1 leads to phenotypes similar to overexpressing CP, some of which may be explained by exaggerated inhibition of formins (Jung et al., 2016). Interestingly, CP is present in most eukaryotic cells, but V1 is absent from the plant and fungal kingdoms. In budding yeast, high levels of active Cap1/2 are maintained to rapidly cap newly generated free barbed ends at cortical actin patches, restricting filament length and patch growth. However, yeast do not have a V1 homolog. We propose that the properties of Cap1/2 have evolved to allow formins to continuously assemble a robust cable network without interference from the high cellular concentration of Cap1/2.

What is the molecular basis for the differences we have described between yeast and vertebrate CP in competing with formins at the barbed end? CP and formins are both dimers with their two halves making separate contacts with the ultimate and penultimate actin subunits at the barbed end (Otomo et al., 2005; Narita et al., 2006; Funk et al., 2021). Thus, at any given moment, a CP or formin dimer can be bound to one or both actin subunits and still remain attached to the barbed end. A high-resolution structure of CP at the barbed end shows that it uses the C-terminal tentacle of its β-subunit to bind the ultimate actin subunit and its α-tentacle to bind the penultimate subunit (Funk et al., 2021). There is a steric clash between the β-tentacle of CP and the “knob” portion of the formin FH2 domain on the ultimate actin subunit (Otomo et al., 2005), but no clashes between the α-tentacle of CP and the formin on the penultimate subunit (Shekhar et al., 2015). Thus, when CP and formin join each other at the barbed end, they are likely both bound to the penultimate subunit (non-competitively) and dynamically compete for binding on the ultimate subunit. An alignment of the sequences of vertebrate CapZ and S. cerevisiae Cap1/2 in the β-tentacle region (Fig. S5 D) reveals divergence in some of the key hydrophobic residues used to bind actin (Kim et al., 2010), which may explain the decreased effectiveness of Cap1/2 in competing with formins. Interestingly, a different competitive relationship at the barbed end has been reported between the β-tentacle of CP and the WH2 (or “V”) domain of Arp2/3 complex activators, which leads to release of WH2-containing activators from the filament barbed end to promote new rounds of nucleation (Funk et al., 2021). Together with our observations, this suggests that the β-tentacle of CP may play a central role in mediating competitive relationships between CP and other barbed end–associated proteins.

Our data suggest that the formin–CP competition mechanism may be tuned in different biological settings to not only promote the formation of distinct cellular actin structures but also to sort different actin-binding proteins to these structures. In budding yeast, Cap1/2 localizes to cortical actin patches and limits patch growth by capping the Arp2/3-nucleated free barbed ends. Consistent with this role, patches in cap2∆ cells grow abnormally large (Amatruda et al., 1990; Amatruda and Cooper, 1992; Shin et al., 2018; Antkowiak et al., 2019). However, cap2∆ also leads to diminished actin cables, for reasons that have not been well understood. One possible explanation we considered was that Cap1/2 might be on the cables, capping free barbed ends and stabilizing the filaments. However, even using an exceptionally bright (triple-mNeon) tag on Cap1/2 and sensitive imaging techniques that detected single molecules in vivo, we could find no evidence for Cap1/2 decoration of cables. Moreover, when we overexpressed Cap1/2 by ∼30-fold, it had no effect on cable networks. These observations strongly suggest that Cap1/2 is not a component of cables and that it fails to effectively compete with formins at barbed ends or interfere with cable assembly. Instead, our data favor the view that cap2∆ leads to diminished cables as an indirect consequence of the changes it causes at patches. We show that F-actin levels increase by ∼40% in cap2∆ cells compared with wild-type cells, which may decrease the actin monomer pool available for cable assembly, leading to reduced cable formation (Burke et al., 2014; Shin et al., 2018; Antkowiak et al., 2019).

Cap1/2 also plays a key role in protecting patches from decoration by actin-binding proteins normally found only on cables. Using live imaging, we investigated how patches in cap2∆ cells change with time and discovered that as the mutant patches age, they become more cable-like, ectopically recruiting tropomyosin (Tpm1) and formins (Bni1 and Bnr1). Kovar and colleagues have shown that fimbrin and tropomyosin directly compete for F-actin binding in vitro and that fimbrin normally excludes tropomyosin from patches in vivo (Skau and Kovar, 2010). We found that early in their development patches are decorated by fimbrin/Sac6, but as they age Tpm1 begins to arrive and fimbrin/Sac6 levels decline. This may be due to the filaments in patches becoming abnormally long in cap2∆ cells. Indeed, computational models have predicted that loss of CP causes the filaments in patches to increase in length and extend radially, becoming too far apart to bind fimbrin/Sac6, a short-distance crosslinker (Volkmann et al., 2001; Berro and Pollard, 2014). We propose that this progressive defect in patch architecture clears the way for tropomyosin decoration as patches age, explaining how patches in cap2∆ cells recruit two competing proteins Sac6/Fimbrin and tropomyosin (see model, Fig. 6 H). Further, we find that the absence of Cap1/2 at barbed ends allows formins to inappropriately decorate patches, although formin activity was not required for Tpm1 recruitment. This role for Cap1/2 in shielding Arp2/3-nucleated actin networks from formins and tropomyosin appears to extend to diverse species. Related observations have been made in fission yeast, where loss of CP led to the decoration of patches with formins and tropomyosin (Billault-Chaumartin and Martin, 2019). However, this study left open the question of how patches could be decorated by both fimbrin and tropomyosin since they compete for binding to F-actin. We resolved this by performing live imaging and found that tropomyosin arrives late in the lifetime of the cap2∆ patches, as levels of branching are diminishing (reduction of Arp2/3 signal) and fimbrin is diminishing. In addition, a study in mammalian cells showed that CP knockdown leads to inappropriate decoration of Arp2/3-nucleated actin networks by the barbed end elongation factor Ena/VASP (Mejillano et al., 2004). Collectively, our work and the work of other groups highlight the physiological importance of competition at the barbed end (among capping proteins, elongation factors, and depolymerases) in producing actin networks with distinct filamentous architectures and functions.

All yeast strains used in this study are in the W303 background unless otherwise indicated. A complete list of strains used is available in Table S1. Strains expressing C-terminal tags at endogenous loci were generated by PCR-based homology repair (Longtine et al., 1998) followed by lithium acetate/polyethylene glycol transformation (Gietz and Woods, 2002). Transformants that grew on selective medium were screened for expression of the tagged protein, and then correct integration into the genome was confirmed by PCR analysis. PCR products for yeast transformation were amplified from pFA6A vectors using primers with 40-bp homology arms as described (Longtine et al., 1998). New pFA6A vectors encoding improved fluorescent proteins (mScarlet, 3xmScarlet, or 3xmNeon) were cloned using homology-based cloning with NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs) following the manufacturer’s instructions. The coding sequence for mScarlet was amplified from vector pmScarlet_C1 (#85042; Addgene) and the coding sequence for mNeon from ITPKA_3xmNeon (#129606; Addgene). The pFA6A backbone was amplified from vectors pFA6A-GFP^Envy::HIS3 (#60782; Addgene), pFA6A-yomApple::Kan (#44957; Addgene), or pFA6A-yomApple::URA3 (#44879; Addgene). Complete lists of primers and plasmids used are available in Tables S2 and S3, respectively.

Homology-based cloning was used to build integration vectors for overexpression of Cap1/2-GFP^Envy or CapZ-GFP^Envy. Sequences for CAP1, CAP2, and the GAL1/10 promoter were PCR amplified from the yeast genome, GFP^Envy was amplified from pFA6A-GFP^Envy::HIS3 (#60782; Addgene), and the plasmid backbone was amplified from the yeast integration vector pRS305. PCR products were assembled using NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs) generating pGAL-Cap1/2-GFP^Envy::LEU2 (pBG2638). Plasmid pGAL-CapZ-GFP^Envy::LEU2 (pBG2639) was built by a similar approach except that sequences for CapZα1 and β1 were amplified from Escherichia coli expression vector SNAP-CapZ (#69948; Addgene; Bombardier et al., 2015), and the GFP^Envy tag was introduced at the N-terminus of the β1 subunit in the same position as the SNAP tag in the original vector. Primers used in cloning pBG2638 and pBG2639 are described in Table S2. Transformants were selected on -Leu plates and confirmed both by PCR and visually for GFP^Envy localization to actin patches after galactose induction. The centromeric (CEN) vector for expression of mCherry-Sec4::LEU2 has been described (Rands and Goode, 2021).

For visualizing Tpm1, homology-based cloning was used to build integration vectors for expression of mNeonGreen-Tpm1 from the LEU2 locus. Sequences for the Tpm1 promoter, coding sequence, and 3′UTR were amplified from the yeast genome, the mNeonGreen sequence was amplified from plasmid ITPKA_3xmNeon (#129606; Addgene), and the vector backbone was amplified from the yeast integration vector pRS305. A linker sequence between mNeonGreen and Tpm1 was included as described previously by introducing the sequence into the primers (Hatano et al., 2022 Preprint). The primers used are described in Table S2. PCR products were assembled using NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs) generating plasmid mNeonGreen-tpm1::LEU2 (pBG2642). Transformants were selected on -Leu plates and confirmed both by PCR and visually for mNeonGreen localization to actin cables.

For purification of S. cerevisiae capping protein (Cap1/2) from E. coli, we constructed a Cap1/2 expression vector (pBG2643). To generate this plasmid, CAP1 and CAP2 genes were PCR amplified from the yeast genome using primers that introduce a Strep-tagII (WSHPQFEK) and PreScission protease cleavage sequence (LEVLFQ/GP) at the N-terminus of Cap2, and these DNA fragments were cloned into separate sites in the E. coli vector pRSFDuet-1 using homology-based cloning with NEBuilder HiFi DNA Assembly Master Mix.

To grow yeast for most experiments, a single colony was used to inoculate a 5 ml overnight culture of synthetic complete media (2% glucose), then diluted and grown at 25°C to OD₆₀₀ 0.3–0.6. However, for galactose-induced overexpression experiments, yeast were first grown overnight in synthetic complete media (2% glucose), then back-diluted by adding 50 μl of the culture to 5 ml synthetic complete media (2% raffinose), and grown at 25°C for 7–8 h until reaching OD₆₀₀ 0.4–0.6. Cells were then diluted 500-fold to a final volume of 5 ml in synthetic complete media (2% galactose) and grown overnight at 25°C to OD₆₀₀ 0.3–0.6.

For phalloidin staining of F-actin, log phase cells were fixed by adding formaldehyde to a final concentration of 3.7% and incubating for 40 min at 25°C with agitation. Following fixation, cells were rinsed three times in PBS, resuspended in PBS with 0.1% Triton X-100, and stained with Alexa Fluor 488 phalloidin (Invitrogen) overnight at 4°C. In all experiments, to control against tube-to-tube variability in phalloidin staining, wild-type cells (without any fluorescent tags) were included as an internal control and stained and imaged in the same tube as experimental cells; experimental cells could be easily distinguished from wild-type cells by expression of Arc15-mScarlet, Cap2-GFP, or CapZ-GFP. After phalloidin staining, cells were washed three times in PBS and mounted in VECTASHIELD mounting medium with DAPI (Vector Laboratories).

To obtain high-resolution detail of F-actin organization, cells were imaged on a super-resolution Airyscan laser scanning confocal. However, for measurements of fluorescence levels (e.g., total F-actin fluorescence in cells or distribution of Sec4 fluorescence between mother and bud compartments), cells were imaged on a spinning-disk confocal (details below). Unless indicated otherwise, all imaging experiments were carried out at room temperature (∼25°C). To assess F-actin organization, cells were imaged (immediately after the PBS washes described above) on a Zeiss AxioObserver LSM 880 equipped with an Airyscan super-resolution module and Gallenium Arsenide Phosphide detectors using a Plan-Apochromat 63×/1.40 oil DIC M27 objective. Alexa Flour 488 phalloidin was excited using an Argon laser (488 nm), and mScarlet-tagged proteins were excited using a DPSS 561-10 laser (561 nm), each at 0.5% laser power. Z-stacks were acquired at a step size of 200 nm through the entire volume of the cell. Image acquisition and Airyscan processing were done in Zen 2 (black) software. To measure fluorescence levels, cells were imaged on an i-E upright confocal microscope (Nikon Instruments) with a CSU-W1 spinning-disk head (Yokogawa), 100× oil objective (NA 1.4; Nikon Instruments), and an Ixon 897 Ultra-CCD camera (Andor Technology) controlled by NIS-Elements software (Nikon Instruments). Exposure times of 200 ms at 50% laser power (excitation 405, 488, and 561 nm) were used to acquire 37 Z-slices (at 0.2 μm steps) spanning the entire volume of the cell.

All image analysis was done in ImageJ (Schindelin et al., 2012). Super-resolution Airyscan images were used to quantify the number of actin cables growing from the mother-bud neck into the mother compartment, and the lateral intensity (width) of each cable. Cable number was counted manually by scrolling through each slice in a Z-stack to track cables growing from (and attached to) the mother-bud neck. To quantify the intensity of individual cables, Airyscan Z-slices were averaged into a single slice, and a 5-pixel-wide line was drawn across the middle of each mother cell taking care to avoid drawing the line in the same location as an actin patch. Peaks in phalloidin signal (from actin cables) along this line were identified using the BAR Find Peaks plugin. Cable peak values were background subtracted by subtracting the average signal from the image area outside of the cells. To determine whether cables were different between experimental cells and internal control (wild-type) cells stained in the same reaction, cable peak values from experimental cells were normalized to internal (wild-type) control cell cable peaks. This was achieved by calculating average cable peak intensity across all control peaks and then dividing each individual peak measured in both the experimental and control cells by this control average. Values reported in graphs are normalized values.

Spinning-disk confocal microscope images were used to quantify the total F-actin per cell and the percent of F-actin associated with patches. Total cell area was selected by taking advantage of the fact that the DAPI signal has a high cytosolic signal distinguishable from the slide background. Z-slices of the DAPI channel were maximum intensity projected and used to generate a binary mask outlining each cell (total cell mask). Cell masks were manually inspected to ensure neighboring cells were appropriately separated. Actin patches were detected using the Arc15-RFP^mScarlet patch marker. Z-slices of the Arc15-RFP^mScarlet channel were maximum intensity projected, contrast-enhanced, and used to generate a binary mask outlining actin patch networks (actin patch mask). These masks were used to generate three different regions of interest (ROI) in each cell: (1) the total cell ROI, (2) the actin patch ROI, and (3) the “patch-excluded” total ROI. The sum of the actin patch ROI and the patch-excluded ROI equals the total cell ROI. Measurements were taken from average intensity projections of Z-slices of the Alexa Flour 488 phalloidin channel. The background was determined by measuring the average fluorescence signal of the area outside of the cells. After background subtraction, the percent of phalloidin signal in patches was calculated by measuring the integrated density (area × mean signal) of the total cell ROI, the patch ROI, and patch-excluded ROI. The integrated density of the total cell ROI = 100% of the phalloidin signal. We then calculated the percent of this total signal in the patch ROI and the patch-excluded ROI.

For live imaging experiments, cells were grown as above for fixed cell imaging. To investigate Cap1/2 localization (using a 3xmNeon green tag on Cap2), single colonies of strains coexpressing Cap2–3xmNeon with Abp140-3xRFP^mScarlet were inoculated into 5 ml synthetic complete media (2% glucose) and grown at 25°C overnight with agitation to OD₆₀₀ = 0.3–0.5. Cells were incubated with 100 μM CK666 (to remove actin patches) or with 100 μM latrunculin B (to remove all F-actin) for 15 min at 25°C with agitation, and then immediately mounted and imaged. Live cells were mounted on 5% agarose pads made with synthetic complete media and imaged by Airyscan microscopy as described above for fixed cells. Where CK666 was used, the agarose pads were soaked in synthetic complete media plus 100 μM CK666 for at least 30 min before adding cells to prevent dilution of the drug and recovery of patches during imaging. To visualize Cap2–3xmNeon at the cell cortex while minimizing signal from the cytosol, cells were imaged at room temperature on an inverted Ti200 TIRF microscope (Nikon Instruments) equipped with 100-mW solid-state lasers (Agilent Technologies), a CFI Apo 100× TIRF objective (NA 1.49; Nikon Instruments), and an iXon EMCCD camera (Andor Technology). For dual color imaging of Cap2–3xmNeon and Arc15-RFP^mScarlet marked actin patches, images were captured every 100 ms for 2 min using triggered acquisition with the FITC–TRITC dual band filter using a 50-ms exposure time and 10 and 5% laser powers for 488 and 561 nm lasers, respectively. In experiments where Cap2–3xmNeon was imaged alone, the FITC filter set was used with a 50 ms exposure time for the 488 nm laser, and cells were imaged continuously for 2 min. Images were background subtracted in ImageJ and Cap2–3xmNeon particles were tracked using ImageJ plugin TrackMate (Tinevez et al., 2017).

Actin cable organization and secretory vesicle traffic were quantified in live cells using strains expressing the cable marker Abp140-GFP^Envy and the post-Golgi secretory vesicle marker RFP^mCherry-Sec4. Strains were grown and mounted on agarose pads with and without CK666 as above, except that cells were incubated with CK666 for 30 min prior to imaging. Z-stacks were acquired on the same spinning-disk confocal microscopy setup described above for fixed cells, except that images were acquired with a 100 ms exposure time. Z-stacks were collapsed into a single average intensity projection and background was subtracted in ImageJ (Schindelin et al., 2012). The percent of RFP^mCherry-Sec4 signal in the bud was calculated by measuring the total signal in the entire cell (integrated density) and calculating what fraction of this value is from the bud compartment (integrated density bud). The ImageJ plugin FibrilTool (Boudaoud et al., 2014) was used to quantify anisotropy (alignment) of Abp140-GFP^Envy labeled F-actin structures in the mother cell compartment.

To assess Bni1 and Bnr1 localization in cap2∆ cells, single colonies of strains Bni1–3xmNeon or Bnr1–3xmNeon were grown in synthetic complete medium and mounted on agarose pads as described above. Z-stacks were acquired using the same spinning-disk confocal microscopy described above at 100 ms exposure times. Image processing was done in ImageJ. Z-stacks were collapsed into a single maximum intensity projection and the background was subtracted. To compare formin recruitment to patches in wild-type and cap2Δ cells, cells expressing Bni1–3xmNeon or Bnr1–3xmNeon and a patch marker (Arc15-RFP^mScarlet) were imaged using the same microscope and settings as above, except that images in the 488 channel (mNeon) were acquired for five frames (0.9 sec/frame), then images in the 561 channel (Arc15-RFP^mScarlet) for one frame. Because the Bni1 and Bnr1 signals are relatively low, we took extra precautions to avoid signal bleed-through from the 561 channel by using separate FITC/Cy2 and TRITC/Cy3 filter cubes instead of using a dual bandpass filter. To quantify the % of patches that showed Bni1–3xmNeon or Bnr1–3xmNeon colocalization during the 2-min observation window, the mNeon signal was plotted as a function of the RFP^mScarlet signal and a linear regression model was used to find the line of best fit for the data and to calculate the coefficient of correlation (R; examples in Fig. S3, C and D; distribution of R values for individual patches in Fig. S3, E and F). As an additional control for nonspecific colocalization, we rotated the signal in the 488 channel by 90° with respect to the 561 channel and determined the degree of colocalization (R² < 0.1), then used this as a threshold for scoring specific colocalization, i.e., only mNeon signals of R² >0.1 were counted as correlating with RFP^mScarlet signal.

In wild-type and cap2Δ cells, we compared recruitment of F-actin binding proteins (Abp140-3xmNeon, Sac6-GFP^Envy, and mNeon–Tpm1) to actin patches (marked by Arc15-RFP^mScarlet) throughout patch lifetime on an inverted Ti200 TIRF microscope (Nikon Instruments) equipped with 100-mW solid-state lasers (Agilent Technologies), a CFI Apo 100× TIRF objective (NA 1.49; Nikon Instruments), and an iXon EMCCD camera (Andor Technology). Images were captured every 100 ms for 2 min using triggered acquisition with the FITC-TRITC dual-band filter using a 50 ms exposure time and 10 and 5% laser powers for 488 and 561 nm lasers, respectively. TIRF time-lapse images were background subtracted in ImageJ, and the F-actin binding protein signal at patches was measured by drawing a 5 by 5 pixel (0.8 by 0.8 μm) wide box around a single patch (marked with Arc15-RFP^mScarlet) and recording the signal in both the 488 and 561 channels over the entire 2-min observation window. This analysis only included patches whose entire lifetimes could be observed (from initial appearance at the cortex to disappearance), and which did not spatially overlap with other patches. Raw intensity measurements were smoothed in GraphPad (Prism) using the 20-nearest neighboring points and a second-order smoothing polynomial. The baseline signal was determined by taking the average of the smoothed signal in the first 10 time frames and the last 10 time frames for each channel. Then, the baseline was subtracted from all time points. From these data, we identified the time and intensity of the peak signal in both the 488 and 561 channels. Raw intensity measurements were scaled by subtracting the calculated baseline signal, setting the baseline to zero, and then setting the peak signal measured during the 2-min observation window to 100% for each channel. Lastly, the time of the peak Arc15-mScarlet signal was set to 0 s. This allowed us to align and average scaled (but not smoothed) intensity measurements made across multiple patches regardless of when the patch occurred during the 2-min observation window. This facilitated comparison of F-actin binding protein recruitment to patches across multiple patches in wild-type and cap2Δ cells.

To determine if the altered actin patch morphology observed in cap2Δ cells results from ectopic formin recruitment to the patches, we generated a strain that enabled temperature-sensitive inactivation of formin activity, cap2Δ bni1-1 bnr1Δ. This strain was generated by crossing BGY24 (cap2Δ) to BGY715 (bni1-1 bnr1Δ; Sagot et al., 2002). Two independent isolates of cap2Δ bni1-1 bnr1Δ were used in all experiments. We performed statistical tests (one-way ANOVA), which revealed that the observed changes in F-actin signal (at 25 and 37°C) are consistent between independent isolates and that any differences between the isolates are extremely small and not contributing to the statistical differences we observe between genotypes (as graphed in Fig. S5, A–C). The data presented are pooled from the two isolates. To visualize F-actin in wild-type and cap2Δ cells before and after formin inactivation, single colonies of wild-type, cap2Δ, bni1-1 bnr1Δ, and cap2Δ bni1-1 bnr1Δ strains were used to inoculate 5 ml yeast extract peptone dextrose cultures and grown at 25°C to OD₆₀₀ 0.3–0.6. Cultures were then split, such that half was shifted to the nonpermissive temperature of 37°C for 30 min before fixing in 3.7% formaldehyde for 40 min (at 37°C). The other half of the culture was maintained at 25°C and then fixed as above (at 25°C). Cells were stained with Alex488-phalloidin as above, mounted, and samples were imaged separately on a spinning-disk confocal and an Airyscan microscope, as described above for fixed cell imaging. As above, in all experiments, a wild-type control strain (marked with RFP^mScarlet) was grown at 25°C in parallel, then mixed with experimental cells during fixation and phalloidin stained to control for tube-to-tube variability in staining. The total F-actin signal per cell was measured as described above for fixed-cell imaging. The signal intensity of individual patches was measured in ImageJ from a single Airyscan Z-section by drawing a 5 × 5-pixel box around the central Z-plane of each patch. Signal background (outside of the cell) was subtracted from these measurements and values are reported relative to patch intensities of the internal control strain.

To determine whether formin activity is required for tropomyosin recruitment to patches, we generated a cap2Δ bni1-1 bnr1Δ strain expressing mNeon–Tpm1 and the patch marker Arc15-RFP^mScarlet. This strain was constructed by crossing a cap2Δ bni1-1 bnr1Δ strain to a strain expressing mNeon–Tpm1 and Arc15-RFP^mScarlet. These strains were then grown in synthetic complete media (2% glucose) and imaged on a spinning-disk confocal microscope as above for live cell imaging. To inactivate formins, cells were shifted to 37°C for 10 min, then imaged as above except at 37°C using a CherryTemp (Cherry Biotech) stage heater.

S. cerevisiae actin was purified from commercial baker’s yeast (Red Star Yeast) using DNase I affinity and Mono Q (GE Healthcare) chromatography as described (Goode, 2002; Pollard et al., 2020). Briefly, wet yeast bricks (packed cells) were washed with water, flash-frozen in pellets in liquid nitrogen, and then mechanically lysed by grinding into a fine powder while submerged in liquid nitrogen using a Waring blender. Lysed cell powder was stored at −80°C until use. To initiate a preparation, yeast powder was mixed 1:1 (wt:vol) with room temperature G-buffer (5 mM Tris-HCl, pH 8.0, 0.2 mM ATP, 0.1 mM CaCl₂, and 0.5 mM DTT) supplemented with protease inhibitors (0.5 μg/ml each of leupeptin, aprotinin, antipain, chymostatin, and pepstatin A, and 1 mM phenyl methylsulfonyl fluoride [PMSF]) and then centrifuged at 215,000 × g for 1 h in a Ti60 rotor (Beckman). Supernatants were loaded onto a 20 ml DNase I column. The DNase I column was washed with 10 column volumes of G-buffer, then 2 column volumes of G-buffer plus 10% formamide, then 2 column volumes of G-buffer plus 0.2 M ammonium chloride, and finally 2 column volumes of G-buffer alone. Actin was eluted from the column with 2 column volumes of G-buffer plus 50% formamide and immediately loaded onto a 1-ml Mono Q column on an AKTA FPLC (GE Healthcare). The Mono Q column was washed with 5 ml of G-buffer and then actin was eluted using a 20-ml linear salt gradient (100–300 mM KCl) in G-buffer. Peak fractions containing actin (which eluted at ∼250 mM KCl) were concentrated to 40–45 μM, then dialyzed three times over a 48-h period, each time against 1 liter of G-buffer. Actin was clarified by centrifugation at 373,000 × g for 20 min, aliquoted, flash-frozen, and stored at −80°C.

Rabbit skeletal muscle actin was purified from acetone powder as described (Spudich and Watt, 1971; Kuhn and Pollard, 2005). Actin was partially extracted from skeletal muscle tissue and stored as acetone powder at −80°C. Then, acetone powder was resuspended in G-buffer and cleared by centrifugation at 50,000 × g. Actin was polymerized overnight at 4°C by addition of 50 mM NaCl and 2 mM MgCl₂. Then, 0.6 M NaCl was added to F-actin to dissociate actin-binding proteins and the F-actin was pelleted by centrifugation at 361,000 × g for 2.5 h. F-actin pellets were resuspended in G-buffer, dounce homogenized, and dialyzed three times against 1 liter G-buffer as above.

For TIRF microscopy assays, actin was labeled with on surface-exposed lysines with Alexa-488 as described (Shekhar and Carlier, 2017). Briefly, G-actin was polymerized by dialyzing overnight against modified F-buffer (20 mM Pipes, pH 6.9, 0.2 mM CaCl_2, 0.2 mM ATP, and 100 mM KCl). F-actin was incubated for 2 h at room temperature with a fivefold molar excess of Alexa-488 NHS ester dye (Life Technologies). The F-actin was then pelleted by centrifugation at 450,000 × g for 40 min at room temperature, and the pellet was resuspended in G-buffer, homogenized with a dounce, and incubated on ice for 2 h to depolymerize the filaments. The monomeric actin was then repolymerized on ice for 1 h by the addition of 100 mM KCl and 1 mM MgCl₂, and the F-actin was pelleted by centrifugation for 40 min at 450,000 × g at 4°C. The pellet was homogenized with a dounce and dialyzed overnight at 4°C against 1 liter of G-buffer. The solution was precleared by centrifugation at 450,000 × g for 40 min at 4°C. The supernatant was collected, and the concentration of actin and labeling efficiency were determined by measuring absorbance at 280 and 495 nm, respectively. Molar extinction coefficients used were as follows: ε₂₈₀ actin = 45,840 M⁻¹ cm⁻¹, ε₄₉₅ Alexa-488 = 71,000 M⁻¹ cm⁻¹, and ε₂₈₀ AF488 = 7,810 M⁻¹ cm⁻¹.

Untagged S. cerevisiae profilin (Pfy1) was purified by gravity flow column chromatography using Poly-L-proline Sepharose affinity as described previously (Kaiser et al., 1989). To generate a column, Poly-L-proline (Sigma P3886) was conjugated to CnBr-Sepharose (Sigma C9210) by suspending 15 g CnBr-Sepharose in 200 ml 1 mM HCl and rocking for 15 min. Swelled CnBr-Sepharose was collected on a sintered glass filter and washed with 500 ml of 1 mM HCl followed by 500 ml of coupling buffer (0.1 M NaHCO₃, 0.5 M NaCl). After rinsing, CnBr-Sepharose was resuspended in 100 ml coupling buffer and divided into 25 ml aliquots, and 125 mg Poly-L-proline dissolved in coupling buffer was added to each aliquot. Poly-L-proline and CnBr-Sepharose were incubated together overnight at 4°C with rocking. Following incubation, Poly-L-proline Sepharose beads were collected by centrifugation and washed with coupling buffer followed by 100 mM Tris-HCl (pH 8.0). Rinsed aliquots were resuspended in ∼40 ml 100 mM Tris-HCl, pH 8.0, and incubated with rocking at 4°C for 2 h. Poly-L-proline Sepharose beads were then washed three times with 25 ml storage buffer (20 mM Tris-HCl, pH 8.0, 150 mM KCl, 0.2 mM DTT), pooled, and stored at 4°C in storage buffer with 0.01% sodium azide. Untagged Pfy1 was expressed from plasmid pBG285 (pAW1) as described previously (Wolven et al., 2000). E. coli strain Rosetta(DE3) with the pRARE rare codon plasmid was grown to log phase in Terrific Broth at 37°C and induced with 1 mM IPTG for 4 h at 37°C. Cells were harvested by centrifugation and pellets were stored at −80°C overnight. Cell pellets were resuspended in lysis buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 2 mM DTT) supplemented with protease inhibitors lacking PMSF and mechanically lysed by passing through an LM20 Microfluidizer (Microfluidics) two times at 18,000 PSI. Following lysis, PMSF (1 mM final) was added and then the lysate was clarified by centrifugation at 27,000 × g for 30 min at 4°C. The Poly-L-proline Sepharose column was preequilibrated with five column volumes of lysis buffer, and then the clarified cell lysate was passed over the column two times. Next, the column was washed two times with 10 column volumes of wash buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 2 mM DTT, 100 mM NaCl) and then Pfy1 was eluted using two column volumes of elution buffer (wash buffer plus 8 M urea). Pfy1 was dialyzed four times sequentially over a 24-h period, each time against 1 liter of storage buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 2 mM DTT), then concentrated using an Amicon Ultra centrifugal filter device (Milipore) to a final concentration of 100–200 μM, flash frozen in 20-μl aliquots, and stored at −80°C.

Active C-terminal FH1-FH2-C fragments of S. cerevisiae formins Bni1 (1227-1953) and Bnr1 (757-1375) were subcloned into an E. coli expression vector with N-terminal hexahistidine (6xHIS) followed by a small ubiquitin-related modifier (SUMO) tag in the pET-23b plasmid backbone, generating plasmids 6xHIS-SUMO-Bni1-(1227-1953; pBG1936) and 6xHIS-SUMO-Bnr1-(757-1375). Plasmids were transformed into E. coli strain Rosetta(DE3) with the pRARE rare codon plasmid. Cells were grown to log phase in Terrific Broth at 37°C and induced with 1 mM IPTG for 16 h at 18°C. Cells were harvested by centrifugation and pellets were stored at −80°C. Pellets were resuspended in lysis buffer (30 mM imidazole, pH 8.0, 0.5 mM DTT, and 2× PBS [40 mM sodium phosphate buffer and 200 mM NaCl, pH 7.4]) supplemented with 150 mM NaCl, 1% NP-40, and protease inhibitors. Next, cells were lysed using a probe sonicator on ice, clarified by centrifugation at 27,000 × g for 30 min at 4°C, and mixed with 1 ml nickel-nitrilotriacetic acid (Ni-NTA) agarose beads (Qiagen) for 1 h at 4°C. The slurry was transferred to an empty column and the agarose was allowed to settle by gravity. The column was washed twice with 10 column volumes of lysis buffer plus 350 mM NaCl and then twice with 10 column volumes of wash buffer (20 mM Hepes, pH 7.5, 100 mM NaCl, 5% glycerol). Formins were then eluted using wash buffer plus 250 mM imidazole. The 6xHIS-SUMO tag was removed from the formins by overnight digestion at 4°C with Ulp1 protease while dialyzing the reaction against 500 ml HEKG₅ buffer (20 mM Hepes, pH 7.5, 1 mM EDTA, 50 mM KCl, 10% glycerol) with 2 mM DTT. Following dialysis, formins were further purified by gel filtration on a Sup6 column (GE Healthcare) equilibrated in buffer HEKG₁₀ with 1 mM DTT using an AKTA FPLC (GE Healthcare). Peak fractions were pooled, concentrated, aliquoted, snap-frozen, and stored at −80°C.

The active C-terminal FH1-FH2-C fragment of mouse formin mDia1 (residues 553–1255) was purified from S. cerevisiae as described (Gould et al., 2011). Briefly, 2 liters of cells were grown at 25°C to OD₆₀₀ = 0.8 in selective medium containing 2% raffinose. Galactose (2% final) was added to induce expression for 12–16 h. Yeast cells were harvested by centrifugation and resuspended in a 1:3 (vol/wt) ratio of water, mechanically lysed in a coffee grinder with liquid N₂, and stored at −80°C. Frozen yeast powder was thawed in a 1:4 (vol/wt) ratio of 20 mM imidazole (pH 8.0), 1.5× PBS (40 mM sodium phosphate buffer, 200 mM NaCl, pH 7.4), 0.5% NP-40, 0.2% thesit, 1 mM DTT, and protease inhibitors. Next, the lysate was clarified by centrifugation at 60,000 rpm for 80 min at 4°C in a Ti70 rotor (Beckman/Coulter) and then incubated with Ni-NTA agarose beads (Qiagen) for 1.5 h at 4°C. The beads were washed three times with 20 mM imidazole (pH 8.0), 1× PBS, 1 mM DTT, 200 mM NaCl, and eluted with 0.5 ml of 300 mM imidazole, pH 8.0, 50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM DTT, 5% glycerol, and then purified by gel filtration as above for yeast formins. Peak fractions were pooled, concentrated, aliquoted, snap-frozen, and stored at −80°C.

To purify S. cerevisiae capping protein (Cap1/2), plasmid pBG2643 was transformed into E. coli strain Rosetta(DE3) with the pRARE rare codon plasmid. Cells were grown to log phase in Terrific Broth at 37°C and induced with 1 mM IPTG for 4 h at 37°C. Cells were harvested by centrifugation and pellets were stored at −80°C. Pellets were resuspended in buffer W (100 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 1 mM EDTA) supplemented with 1 mM DTT and protease inhibitors, then chemically lysed by incubation with 0.1 mg/ml Lysozyme on ice for 30 min with intermittent agitation. The cell lysate was clarified by centrifugation at 27,000 × g for 30 min at 4°C and mixed with 5 ml Strep-Tactin Sepharose resin (iba-LifeSciences) equilibrated with buffer W. The resin was transferred to an empty column and allowed to settle by gravity. The lysate was passed over the column three times, and then the column was washed five times with 10 column volumes of buffer W. Finally, the resin was transferred to a tube with 5 ml of buffer W plus 1 mM DTT. The StrepTactin tag was removed by digestion with PreScission protease (GST tagged) overnight at 4°C, releasing untagged Cap1/2. Following cleavage, the GST-PreScission protease was removed by incubation with 500 μl Pierce Glutathione Agarose (16101; Thermo Fisher Scientific) at 4°C for 1 h with rocking. Cap1/2 was concentrated to 60–80 μM, dialyzed against 1 liter storage buffer (20 mM Hepes, pH 7.5, 1 mM EDTA, 50 mM KCl, 1 mM DTT), aliquoted, flash frozen, and stored at −80°C.

SNAP-tagged chicken (Gallus gallus) capping protein (CapZ α1β1) was expressed using a plasmid (#69948; Addgene) described previously (Bombardier et al., 2015). Briefly, SNAP-CapZ was expressed in BL21(DE3) pLysS cells grown at 37°C to early log phase and induced with 0.4 mM IPTG at 37 °C for 8 h. Cells were harvested by centrifugation and stored at −80°C. The SNAP-CapZ was purified as above for mDia1, except that the cells were lysed by sonication in the presence of 0.5 mg/ml of lysozyme, then purified on Ni-NTA agarose followed by gel filtration (Sup 6 column).

Gel-filtered monomeric muscle actin (5% pyrene-labeled) in G-buffer (5 mM Tris-HCl, pH 8.0, 0.2 mM ATP, 0.2 mM CaCl₂, and 0.2 mM DTT) was converted to Mg-ATP-actin immediately before each reaction (Moseley and Goode, 2005). Final reactions were 60 μl containing 1.5 μM G-actin with 7.5 μM yeast profilin (Pfy1). To initiate a reaction, ATP-G-actin with profilin was mixed with yeast formins (either Bni1 or Bnr1) and/or control HEKG₅ buffer, then mixed with 3 μl of 20× initiation mix (40 mM MgCl₂, 10 mM ATP, and 1 M KCl). Each reaction was performed in triplicate using a 96-well plate, and fluorescence was monitored for 60 min at excitation 365 nm and emission 407 nm at 25°C in a Tecan Infinite M200-PRO microplate reader. Reactions were monitored for 3 min, then paused while Cap1/2, CapZ, or control HEK buffer was spiked in, then monitored for another 57 min. The slope of each actin assembly curve, starting 130 s after addition of Cap1/2, CapZ, or buffer, was measured for a 230-s window. Formin-mediated actin assembly slopes are reported relative to the buffer control condition without capping protein.

TIRF coverslips were cleaned in a bath sonicator at 60°C in 2% HELLMANEX III detergent for 60 min, followed by successive sonications in 1 M NaOH for 30 min, and in 100% ethanol for 60 min with rinses in milli-Q water between each step. Coverslips were then dried in an N₂ stream and coated with an 80% ethanol solution adjusted to pH 1.0 with HCl containing 10 mg/ml methoxy-PEG-silane MW 2,000 and incubated overnight at room temperature. The mf-TIRF chamber was assembled using a 40-μm-high Polydimethylsiloxane mold with three inlets and one outlet mechanically clamped onto a PEG-silane-coated coverslip. The chamber was then connected to a Maesflow microfluidic flow-control system (Fluigent) and rinsed with mf-TIRF buffer (10 mM imidazole, pH 7.4, 50 mM KCl, 1 mM MgCl₂, 1 mM EGTA, 0.2 mM ATP, 10 mM DTT, and 15 mM glucose). Next, spectrin-actin seeds were flowed in, which passively absorbed to the glass coverslip. Then nonspecific binding sites were blocked by rinsing the chamber with 1% BSA in HEK buffer (20 mM Hepes, pH 7.5, 1 mM EDTA, 50 mM KCl, and 1 mM DTT), and a final rinse with TIRF buffer. Actin filaments with free barbed ends were assembled by exposing spectrin-actin seeds to a flow containing profilin-actin (1 μM yeast G-actin and 2 μM Pfy1) in mf-TIRF buffer at room temperature. To improve photostability, oxygen scavengers, 20 μg/ml catalase, and 100 μg/ml glucose oxidase were included in all reactions. F-actin was visualized by including 100 nM SiR-actin (Cat. No. CY-SC001; Cytoskeleton) from a 20 μM stock in anhydrous DMSO. Next, yeast formins, Bni1 (160 nM), or Bnr1 (130 nM) were flowed into the chamber and allowed to bind barbed ends for 50 s before returning to a flow in of profilin-actin for another 50 s, and then flow in of profilin-actin with Cap1/2 (100–2,000 nM) or control buffer. Next, filament elongation was monitored for 15 min, with images acquired at 5-s intervals, using an inverted Ti200 TIRF microscope system (Nikon Instruments) equipped with 100-mW solid-state lasers (Agilent Technologies), a CFI Apo 100× TIRF objective (NA 1.49; Nikon Instruments), and an iXon EMCCD camera (Andor Technology) with a 50-ms exposure time and 15% laser power for the 649 nm laser.

Images were analyzed in ImageJ. Background fluorescence was subtracted using the rolling-ball background subtraction algorithm (rolling-ball radius 20 pixels), and kymographs were generated with the Kymograph plugin. Typically, 100–200 filaments were imaged across two to three fields of view, and all of the filaments were monitored for change in growth behavior (arrest of growth or a switch from fast to slow growth). In control reactions (lacking Cap1/2), the percentage of filaments growing at the formin (faster) rate over time was plotted and fit with a one-phase exponential decay in GraphPad Prism. This was used to calculate the dissociation rate constant (k⁻) of yeast formins from barbed ends. In reactions containing Cap1/2, the percentage of filaments capped over time was similarly plotted and fit with an exponential plateau in GraphPad Prism. This was used to calculate the association constant (K_obs) of Cap1/2 to formin-bound barbed ends at different Cap1/2 concentrations. These values (K_obs) were plotted as a function of Cap1/2 concentration, and the association rate constant (k⁺) was derived from the slope of the line.

All experiments were repeated at least three times and yielded similar results. Means and errors (SD or SEM) were calculated using GraphPad Prism. Data distributions were assumed to be normal, but this was not formally tested. Statistical comparison between indicated conditions was conducted using the two-sided Student’s t test or one-way or two-way ANOVA, as indicated in the figure legends. Differences were considered significant if the P value was <0.05 (*).

Fig. S1 compares the in vitro effects of yeast Cap1/2 and vertebrate CapZ on formin-mediated (Bni1, Bnr1, and mDia1) actin assembly in bulk assays, and the inhibitory effects of vertebrate CapZ on yeast formin (Bni1 and Bnr1) stimulated elongation of actin filaments in mf-TIRF assays. Fig. S2 shows the in vitro elongation rates of yeast actin filaments visualized with a SiR-actin probe in mf-TIRF assays, and the minimal effects of formins (Bni1 or Bnr1) on off rate of Cap1/2 from the barbed end (related to Fig. 2 in the main text). Also shown is quantification of the high-level (galactose-induced) expression of Cap1/2-GFP^Envy and GFP^Envy-CapZ in cells and evidence that it has no significant effect on Sac6-RFP^mScarlet marked actin patches (related to Fig. 3 in the main text). Fig. S3 shows the improved detection of formins Bni1 and Bnr1 in cells when tagged with a 3xmNeon tag and the quantification of formin (Bni1 and Bnr1) colocalization with Arc15-RFP^mScarlet labeled patches (related to Fig. 5 in the main text). Fig. S4 shows that even in the absence of formin activity in vivo, there is excessive accumulation of F-actin (indicated by phalloidin staining) at cortical actin patches in cap2∆ cells (related to Fig. 6 in the main text). Further, it shows that actin-binding proteins have altered dynamics at cortical patches in cap2∆ cells, and that aberrant decoration of older patches by tropomyosin (Tpm1) in cap2∆ cells occurs independent of formin activity. Fig. S5 shows that upon CK666 treatment the Arp2/3 complex signal at cortical patches is lost in both wild-type and cap2∆ cells, and provides additional examples of F-actin organization in phalloidin stained wild-type and cap2∆ cells (related to Fig. 7 in the main text). It also shows an alignment of the capping protein β-tentacle primary sequences from budding yeast (S. cerevisiae), fission yeast (S. pombe), human (Homo sapiens), and chicken (G. gallus). Video 1 shows dynamic puncta of 3xmNeon-tagged Cap2 visible at the cell cortex that is distinct from actin patches. Video 2 is an example time series of cells expressing Cap2–3xmNeon treated with control carrier, CK666 or LatB (analyzed in Fig. 5). Video 3 shows continuous imaging of example cells analyzed in Fig. S3 expressing formins (Bni1 or Bnr1) tagged with GFP or 3xmNeon, revealing that the 3xmNeon tag is brighter and more photostable. Videos 4 and 5 (related to Fig. 5) show that 3xmNeon-tagged Bni1 (Video 4) and Bnr1 (Video 5) colocalize with actin patches (marked by Arc15-RFP^mScarlet) in cap2Δ but not WT cells. Video 6 shows that Arc15-RFP^mScarlet marked actin patches in cap2∆ cells acquire cable-like characteristics as they age and inappropriately recruit tropomyosin (mNeon–Tpm1; related to Fig. 6). Table S1 contains a list of yeast strains used in this study. Table S2 contains a list of primers used and a description of each primer. Table S3 contains a list of plasmids used and a description of each plasmid.

Data supporting the findings of this manuscript are available from the corresponding author upon request.

We are grateful to Colby Fees, Luther Pollard, and Bengi Turegun for technical assistance with protein purification and in vitro microfluidics-assisted TIRF microscopy, and Jeff Gelles for intellectual input.

This research was supported by grants from the National Institutes of Health to A.C.E. Wirshing (F32 GM135967) and B.L. Goode (R35 GM134895).

Author contributions: A.C.E. Wirshing and B.L. Goode designed the experiments and wrote the manuscript. A.C.E. Wirshing and S.G. Rodriguez performed the experiments and analyzed the data.