Rhodamine-labeled actin from rabbit skeletal muscle and human recombinant plasma gelsolin were purchased from Cytoskeleton Inc (Denver, CO). Unlabeled actin was purified from rabbit skeletal muscle acetone powder (Pel-Freeze Biologicals Inc., Rogers, AR), then gel filtered over a Sephacryl S-300 size exclusion column equilibrated in buffer A [2 mM Tris-HCl pH 8, 0.2 mM CaCl₂, 1 mM NaN₃, 0.2 mM, adenosine 5-triphosphate (ATP), and 0.5 mM dithiothreitol (DTT)] as previously described (Pardee and Spudich, 1982). Purified actin was fluorescently labeled with Alexa-488 succimidyl ester dye as previously described (Kang et al., 2012; Park et al., 2022), yielding a labeling efficiency of ∼0.18–0.5 fluorophores per actin monomer. Ca²⁺-G-actin (actin monomer) underwent cation exchange to form Mg²⁺-G-actin by the addition of MgCl₂ at a concentration equal to the initial G-actin concentration plus 10 μM and 0.2 mM ethylene glycolbis (β-aminoethyl ether)-N,N,N’N-tetraacetic acid (EGTA).

Actin polymerization was initiated at 1-2 µM actin (50% Rhodamine labeled or 20% Alexa 488 labeled) using 1/10th total volume of 10X KMCI polymerization buffer (100 mM imidazole, 500 mM KCl, 20 mM MgCl₂, 3 mM CaCl₂, 10 mM ATP, and 10 mM DTT, pH 7.5 or 6.0) at room temperature for 1-2 h. Actin was diluted to 0.74–1 μM, following the addition of 5.5–10 nM gelsolin to obtain a gelsolin to actin molar ratio of 1:134 or 1:100. Samples were diluted using optical imaging buffer containing 0.2 mg/mL glucose oxidase, 1 mg/mL catalase and 15 mM glucose to minimize photobleaching then fixed to 0.1% v/v poly-L-lysine treated coverslips. Time-dependent TIRF imaging was conducted over various time (0, 1, 5, 10, 20, and 30 min) intervals. Images were acquired using a Nikon Eclipse Ti TIRF microscope equipped with a Hamamatsu Image EM X2 CCD Camera, a ×100 oil immersion objective, Nikon LU-N4 laser and Nikon Imaging Software (NIS) Elements (version 4.50) (pixel size = 0.16 µm) Average filament lengths were measured using ImageJ (NIH) (Schindelin et al., 2012) and the Persistence Software (Graham et al., 2014). Statistical analysis was performed using ANOVA one-way and post hoc Tukey tests. Length distributions were plotted using the Gaussian (Equation 1) and log-normal (Equation 2) functions to fit the data: y = y 0 + A w π / 2 e − 2 x − x c 2 w 2 (1) y = y 0 + A 2 π w x e − ln x x c 2 2 w 2 (2) where w represents log standard deviation or width in Equations 1, 2, x c represents the center of distribution, and y 0 and A represents the base and area of the distribution. For time lapse imaging analysis, change in average filament length over time at each pH condition was fitted using an exponential decay function according to Equation 3 y = A 1 e − x t 1 + y 0 (3) where t 1 is the time constant, y 0 is the offset and A 1 is the amplitude.

Functionalized flow cells were modified from previously described protocols (Heidings et al., 2020; Winterhoff et al., 2016). Coverslips were cleaned by sonicating at 60°C in 1M KOH, 1M HCl, and 70% ethanol for 45 min, then thoroughly rinsed with warm ddH₂O. 200 nM of N-ethylmalemide (NEM)-inactivated myosin and 1 mg/mL bovine serum albumin (BSA) solutions were separately loaded into a flow cell and left to incubate for 5 min before loading filaments into the flow cell. Images were taken every 1 s to visualize filament severing before and after the addition of gelsolin. To quantify the change in average filament length over time, the data points were fitted with exponential decay functions. The fit yielded the decay time constant (t ₁ ) which is inversely proportional to the rate of gelsolin severing (Heidings et al., 2020). Negative control experiments were done with actin filaments and gelsolin in the absence of Ca²⁺ using 1X KMI polymerization buffer (10 mM imidazole, 50 mM KCl, 2 mM MgCl₂, 1 mM ATP, and 1 mM DTT, pH 7.5 or 6.0).

Bulk actin depolymerization was measured using pyrene-labeled G-actin (1–2 μM, 10% pyrene labeled) as previously described (Vemula et al., 2021; Zhang et al., 2006). F-actin samples were diluted below the actin critical concentration (∼100 nM) in 1X KMCI buffer without or with gelsolin [(gelsolin]: [actin]),1:125]. The time course of pyrene fluorescence intensities was monitored using the SpectraMax Gemini XPS microplate reader equipped with a Molecular Devices SoftMax Pro software (Version 7.0) (λ_ex = 365 nm and λ_em = 407 nm) for 5 min (Δt = 1 s) at room temperature. Negative control experiments were done with actin filaments and gelsolin in the absence of Ca²⁺ using 1X KMI polymerization buffer (10 mM imidazole, 50 mM KCl, 2 mM MgCl₂, 1 mM ATP, and 1 mM DTT, pH 7.5 or 6.0).

Bending persistence lengths of actin filaments with or without gelsolin were measured by analyzing thermal bending modes (Castaneda et al., 2019; Demosthene et al., 2024). A molar ratio of 1:185 [(gelsolin): (actin)] was used to prevent tumbling of short filaments. Coverslips were incubated in a buffer solution containing 1 mg/mL BSA, 50 mM Tris-HCl (pH 7.5), 1 mM NaN₃, and 150 mM NaCl for 1 h to prevent filaments from adhering to the glass. Actin samples were dispensed onto BSA-treated coverslips, then sequential time-lapse images of thermally fluctuating filaments in the absence or presence of gelsolin at pH 7.5 or 6.0 were recorded using a Hamamatsu ORCA-Flash 4.0 V3 Digital CMOS camera (pixel size = 0.07 µm). Time-lapse videos were recorded for 1 min at 7–10 frames/s using the Nikon Imaging Software (NIS) Elements (version 4.50).

Fourier decomposition of filament shape with respect to the tangent angle [θ(s)] along the chosen segment length was processed as indicated by Equation 4 (Castaneda et al., 2019; Gittes et al., 1993): θ s = 2 L   ∑ n = 0 ∞ a n ⁡ cos n π s L (4)

Amplitude of the cosine is denoted by a_n and is nonzero when filaments are in a relaxed state, n is the mode number and L is the segment length of the filament. The amplitude variance a n for each filament bending mode was calculated in the Origin 8.5 software. Modes 1 and 2 were averaged and used for analysis due to higher modes being affected by experimental noise (Castaneda et al., 2019). Analysis of bending L _p from thermal fluctuations was achieved by using the amplitudes and variances calculated to approximate the flexural rigidity the actin filaments, in accordance with Equation 5: L p = L 2 n 2 π 2 v a r a n ) (5) Flexural rigidity of actin filaments can also be defined using Equation 6: κ = L p k B T (6) where κ is the flexural rigidity of the filament, L _p is the persistence length, and k _B T is the thermal energy (Gittes et al., 1993).

PyMOL software was used to visualize interactions of gelsolin domains 1, 2 and 3 (G1-G3) [PDB ID: 1RGI (Burtnick et al., 2004)] and gelsolin domains 4, 5 and 6 (G4-G6) [PBD ID: 1H1V (Choe et al., 2002)] with actin monomers. Ribbon diagram of actin is shown in hot pink and pink and gelsolin is shown in green, lime green and cyan. G1-G3 amino acid residues, Ala-116, Gly-114, and Asp-109 and actin residue, Glu-167 were represented as ball and stick structures. G4-G6 residues, Pro-494, Gly-492, Asp-487 and actin residue, Glu167՛ (adjacent actin monomer) were also represented as ball and stick. Actin and gelsolin complexes were shown to form salt bridges through Ca²⁺.

A gelsolin to actin molar ratio of 1:100 or 1:134 [(actin) = 1 µM] was used for AFM experiments. These ratios are consistent with those chosen for steady-state TIRF imaging and yielded the best AFM images. Freshly cleaved mica substrates were coated 0.1% v/v (3-Aminopropyl)triethoxysilane (APTES) before putting actin. To determine the height of filaments with or without gelsolin in air, the AFM (Park NX10, Parks Systems Corp., South Korea) was operated in non-contact mode (NCM) at a scanning rate of 0.8–1 Hz, 512 × 512 or 1,024 × 1,024 points and scan size range of 0.5–1.0 µm. PPP NCHR (Nanosensors, Switzerland) AFM probe was used with a nominal spring constant of 42 N/m and nominal resonance frequency of 300–330 kHz. Height images were corrected using minimal data processing. The minimum height of the images was fixed to zero and height profiles were extracted using the Gwyddion Software (Nečas and Klapetek, 2012). Helical half-pitch conformational changes were determined by measuring the periodic variations corresponding to G-actin subunits along the filament longitudinal profile section (Ikawa et al., 2007; Gilmore et al., 2013). Statistical analysis was performed using ANOVA one-way and post hoc Tukey tests.

To evaluate the effect of pH on gelsolin-mediated actin filament severing, we calculated the steady-state average filament length (L _avg) using TIRF microscopy imaging (Figure 1A). The L _avg (Figure 1B) and length distributions (Supplementary Figures S1, S2) were used to determine filament severing efficiency in the presence of gelsolin and varying pH. 1 min after the addition of gelsolin to actin filaments (1 μM, 20% Alexa 488-labeled) at pH 7.5 (at a molar ratio of 1:100), the average filament lengths decreased from L _{avg, pH7.5} = 5.2 ± 0.2 µm to L _{avg, pH 7.5, gel} = 3.2 ± 0.1 µm (± standard error of mean, S.E.M) (Figure 1B). At pH 6.0, the control average filament length significantly decreased compared to pH 7.5 (L _{avg, pH6.0} = 4.5 ± 0.1 µm) in the absence of gelsolin. Upon addition of gelsolin at pH 6.0, L _{avg, pH6.0, gel} decreased further to 2.4 µm. At 5 min, L _{avg, pH 7.5, gel=}3.2 ± 0.1 µm and L _{avg, pH 6.0, gel} = 2.3 µm were recorded, and at 10 min, L _{avg, pH 7.5, gel=}2.9 ± 0.1 µm and L _{avg, pH 6.0, gel} = 2.3 ± 0.1 µm were recorded. Overall, the actin filaments at pH 6.0 are significantly shorter than those at pH 7.5. 15 min after gelsolin addition, average filament lengths were similar at pH 6.0 (L _{avg, pH 6.0, gel} = 2.3 ± 0.1 µm) and pH 7.5 (L _{avg, pH 7.5, gel=}2.4 ± 0.1 µm). We also assessed the effect of pH on steady-state filament length distribution in the absence and presence of gelsolin. This data was fitted using the Gaussian and Log-normal functions (Supplementary Figures S1, S2) and indicates that the length distribution in the presence of gelsolin at pH 6.0 narrows down earlier than at pH 7.5.

Next, we directly visualized real-time filament severing activities utilizing a functionalized flow cell (Supplementary Videos S1, S2). Isolated filaments in their native shapes were tracked after the addition of gelsolin to monitor the evolution of L _avg over time. To determine the relative rate of gelsolin severing, the data was fitted using Equation 3 (Heidings et al., 2020) (Figure 1C). The decay time constant t₁ at pH 7.5 was 251 ± 32 s and decreased to 135 ± 18 s at pH 6.0 (Figure 1D). A lower decay time constant indicates that gelsolin severed filaments more efficiently at pH 6.0 than at pH 7.5. Overall, we observed shorter L _avg at pH 6.0 compared to pH 7.5 in the absence of gelsolin. This agrees with a previous study that showed pH influences actin polymerization, resulting in filaments with shorter length at pH 5.8 than at pH 7.8 (Crevenna et al., 2013). This effect was attributed to modulation of the nuclei formation from G-actin by pH (Crevenna et al., 2013). Earlier studies suggested that average filament lengths or shortening were independent of pH (Wang et al., 1989), but recent studies and our data suggests this is not the case. When the effects of both pH and gelsolin were combined, a rapid decrease in average filament length and increase in severing rates were revealed using the real-time severing assays.

To understand the effect of pH on gelsolin-mediated filament disassembly, we used bulk pyrene assays to monitor variations in fluorescence intensities of pyrene-labeled actin over time (Lamb et al., 1993; Fan et al., 2017). The time-dependent fluorescence intensity (Figure 1E) was fitted using an exponential decay function which yielded decay constants (t ₁ ) for pH 7.5 (61.75 ± 3.70 s) and 6.0 (41.02 ± 3.33 s). We then derived the relative disassembly rates of actin filaments in the presence of gelsolin from the inverse of t _1, which provided a rate of 1.61 ± 0.27 × 10² a.u./s at pH 7.5 and 2.43 ± 0.30 × 10² a.u./s at pH 6.0) (Figure 1F). This corresponds to a relative disassembly rate that was 1.5-fold faster at pH 6.0 in the presence of gelsolin. In the absence of gelsolin, no distinct change in fluorescence intensity was noted at pH 7.5 and 6.0 (Supplementary Figure S3). Both of our real-time severing activity data and bulk pyrene fluorescence data are consistent with previous studies that showed enhanced severing activity and actin depolymerization in the presence of full-length gelsolin or gelsolin fragment [residues 28–161 which consists of linker; (G1+)] at pH 5.0–6.0 (Lamb et al., 1993; Fan et al., 2017).

To test whether pH alone affected gelsolin-mediated actin severing, we performed negative controls in the absence of 0.3 mM CaCl₂ using real-time severing assays. At pH 7.5 we observed no significant change in average filament length after the addition of gelsolin for negative controls (Supplementary Video S3; Supplementary Figure S4). In contrast, severing for negative control at pH 6.0 in the presence of gelsolin had a decrease in average filament length from 8.6 µm to 0.9 µm and decay time constant_pH6.0 = 164.32 s (Supplementary Video S4). Gelsolin-mediated filament severing efficiency for negative controls at pH 6.0 was reduced compared to similar conditions supplemented with Ca²⁺. Using real-time severing assays, we showed that gelsolin-induced severing is inefficient at neutral pH in the absence of Ca²⁺. To confirm this data, we performed bulk pyrene fluorescence assays for both pH conditions. In the absence of Ca²⁺ we determined severing rates decreased by more than half at pH 6.0 (1.64 × 10² a.u./s) and an even greater decrease was quantified at pH 7.5 (0.12 × 10² a.u./s) in the presence of gelsolin (Supplementary Figures S5A,B). While Ca²⁺ plays an important role in inducing gelsolin-mediated severing, pH is also an important factor in modulating gelsolin for filament severing. Previous studies have shown that acidic pH (pH 5) lowers the requirement for Ca²⁺ due to bypassing gelsolin’s Ca²⁺ sensitive C-tail latch, while neutral pH conditions (pH 8) remain Ca²⁺ dependent (Lamb et al., 1993; Lagarrigue et al., 2003; Garg et al., 2011). This suggests that acidic pH dominates in regulating gelsolin severing while in the absence or presence of Ca²⁺.

We further evaluated the effect of pH on filament bending mechanics in the absence or presence of gelsolin by using TIRF microscopy imaging. Thermal bending modes of freely fluctuating filaments were analyzed and filament bending persistence lengths were calculated (Supplementary Videos S5–S8; Figure 2). The mode amplitude variance [var (a _n )] calculated from the bending mode analysis (Castaneda et al., 2019; Gittes et al., 1993) (see Method section for details) indicates that thermal bending is affected by varying pH and gelsolin. In the absence of gelsolin, thermal bending was lower at pH 7.5 (Figure 2A) than at pH 6.0 (Figure 2B), as evidenced by reduced var(a_n) (Supplementary Videos S5, S6). “In the case of gelsolin-bound filaments, we observed reduced thermal bending and var(a_n) for both pH conditions compared to the control (Supplementary Videos S7, S8; Figures 2C, 2D). Reduced thermal bending and variance correlates to higher average bending L _p of filaments at pH 7.5 (L _{p, pH7.5} = 17.3 ± 1.8 μm) in comparison to bending L _p of filaments at pH 6.0 (L _{p, pH6.0} = 14.8 ± 1.3 μm) (Figure 2E) in the absence of gelsolin. Using the L _p values, we calculated the average flexural rigidity κ (Equation 6) to be 7.12 × 10⁻²⁶ N m² and 6.10 × 10⁻²⁶ N m² at pH 7.5 and 6.0, respectively, suggesting that actin filaments are slightly less rigid in acidic conditions. In addition, bending L _p at pH 7.5 (L _{p, +gel pH 7.5} = 20.6 ± 1.2 μm) was 19% higher than the control L _p at pH 7.5, while L _p at pH 6.0 (L _{p, +gel pH 6.0} = 18.6 ± 1.0 μm) was 26% higher than the control L _p at pH 6.0 (Figure 2F). Respective κ values were 8.55 × 10⁻²⁶ N m² and 7.67 × 10⁻²⁶ N m². Our results reveal lower L _p and κ values in acidic conditions for control and gelsolin-bound filaments at pH 6.0, with bending rigidity for control filaments at pH 6.0 being significantly lower than gelsolin-bound filaments at pH 7.5. While our study focuses on a more rigid actin species, Mg²⁺-F-actin, the trend is consistent with a previous study reporting that persistence lengths of Ca²⁺-F-actin decreases with decreasing pH (Arii and Hatori, 2008). Lowered flexural rigidity at pH 6 may be a result of the rapid depolymerization and instabilities at filament ends (Fan et al., 2017; Wioland et al., 2018), that may also impact the structure and helical pitch.

Comparing bending stiffness between control and gelsolin-bound filaments demonstrates that gelsolin binding renders filaments stiffer, indicating gelsolin’s ability to modulate filament bending mechanics. This agrees with a previous study showing that gelsolin binding regulates actin mechanics, where torsional rigidity had a three-fold decrease due to conformational changes (Prochniewicz et al., 1996). Yet, the mechanism contributing to gelsolin’s modulation of bending stiffness of actin filaments remains unclear. Another essential ABP, cofilin, was found to reduce both torsional (Prochniewicz et al., 2005) and flexural rigidity (McCullough et al., 2008) when binding to actin. However, we found that gelsolin binding increases actin filament flexural rigidity rather than decreasing it as seen with cofilin. We speculate different severing mechanisms are at play with cofilin and gelsolin. Cofilin was found to affect filament twist (Huehn et al., 2018), making filaments more flexible in twisting and bending (Prochniewicz et al., 2005; McCullough et al., 2008; Kang et al., 2014). Cofilin severing occurs at or near boundaries between bare actin and cofilin-bound segments, at sites of local mechanical and structural discontinuities (Elam et al., 2013; Tanaka et al., 2018). In contrast, gelsolin initiates severing by first attaching gelsolin domain 2 (G2) at the side of actin subdomain (SD) 1 and 2 of longitudinally adjacent monomers. Simultaneous repositioning of the gelsolin domain 1 and 2 (G1-G2) and gelsolin domain 3 and 4 (G3-G4) linkers allow gelsolin domain 1 (G1) and gelsolin domain 4 (G4) to bind between actin SD 1 and 3 of laterally adjacent monomers to induce steric clashing of inter subunit bonds (Nag et al., 2013).

G1 and G4 compete with actin monomers for binding, which induces severing, and also becomes stabilized by gelsolin type-1 Ca²⁺ binding sites formed between actin residue Glu 167 and gelsolin residues at the gelsolin/actin interfaces (Figure 3) (Nag et al., 2013; Burtnick et al., 2004). Glu167 has been shown to form ‘stiffness cation binding sites’ with adjacent monomers in SD 2 to regulate filament stiffness (Kang et al., 2012). Hence, gelsolin binding allows actin Glu167 to form a bond with residues at gelsolin Ca²⁺ binding sites. Potassium (K⁺⁾ and magnesium ions (Mg²⁺) get swapped with Ca²⁺ which binds between actin residue Glu167 and G1 residues (Asp109, Gly114, and Ala116) and adjacent actin residue Glu167՛ and G4 residues (Asp487, Gly492, and Pro494) (Figures 3A, B) (Nag et al., 2013; Burtnick et al., 2004; Choe et al., 2002; Barrie et al., 2024). This may modulate the bending and torsional rigidity of gelsolin-bound filaments. Furthermore, Fan et al. (2017) recently revealed that protonation of gelsolin residues (His29, Asp109, and His151) at acidic pH (<6.0) in Ca²⁺ free conditions induces local structural changes of G1+ and G1+/actin interfaces to enhance binding and pH-dependent severing. We can speculate that structural rearrangement on gelsolin/actin interfaces from Ca²⁺ and pH will also impact filament mechanics as it did for severing dynamics.

Although it is not well understood how the effects of gelsolin binding on single filament mechanics are linked to the mechanical properties of cells, we anticipate its role in regulating cell stress and elastic modulus by referring to other severing ABPs. Actin Depolymerization Factor (ADF) and cofilin causes fluidization of the actin cytoskeleton and cell softening by remodeling contracted, solid filaments into relaxed, fluid-like filaments (Vakhrusheva et al., 2022). Increased bending stiffness of single filaments by gelsolin may modulate the viscoelasticity of cytoskeletal networks, thereby regulating overall cell mechanics.

We investigated conformational changes of F-actin associated with gelsolin binding at varied pH using AFM imaging. Changes in filament height was evaluated from topography images (Figure 4A). Selected lateral line profiles along filaments were used to extract their height values (Figure 4B) (Nečas and Klapetek, 2012; Ikawa et al., 2007). In the absence of gelsolin, we observed average filament height values of 5 nm at pH 7.5 and pH 6.0 (Figure 4C). The addition of gelsolin at pH 7.5 resulted in a decrease of the average height value (4 nm), which was slightly lower than the control. In contrast, the average filament height at pH 6.0 increased to 6 nm. When comparing gelsolin-bound filaments at both pH conditions, we observed significantly higher filament heights in acidic pH. We also noted that our height values were lower for actin filaments than previous values reported in High Speed (HS) AFM studies (∼8–10 nm) (Matusovsky et al., 2019; Ngo et al., 2015). This may be attributed to their use of mica supported bilayers in solution.

To further understand the effects of pH on gelsolin-mediated conformational changes, we examined actin filament helical structure from topography images (Figure 5A). Longitudinal line profiles along filaments were utilized to extract half-pitch values (Figure 5B). We reported the half-pitch of actin filaments at neutral pH to be 39 ± 2 nm (Figure 5C), with gelsolin binding significantly lowering half-pitch to 32 ± 1 nm. Similarly, at pH 6.0 we also observed a significant decrease in actin half-pitch in the presence of gelsolin (39 ± 1 nm to 32 ± 1 nm). Control actin filaments exhibited a large variance and distribution in half-pitch values at both pH conditions. Our half-pitch values were higher than previously reported for actin (35–38 nm) and cofilin-bound filaments (26.9 nm) that underwent shortening and/or overtwisting (Matusovsky et al., 2019; Ngo et al., 2015; Shao et al., 2000; Jegou and Romet-Lemonne, 2020). Recent AFM studies observed a wide distribution for half-pitch and height values, indicative of structural polymorphism (Matusovsky et al., 2019; Ngo et al., 2015). This can be supported by a Molecular Dynamics (MD) simulation study demonstrating how gelsolin segement-1 modulated the orientation, and filament twist of different actin variants (Lee and Kang, 2020). Our data suggest that overtwisting of filaments may occur in the presence of gelsolin at pH 7.5 and 6.0.

The helical structures of actin filaments have implications in mechanical properties including filament torsional and bending rigidity (Jegou and Romet-Lemonne, 2020). Applied mechanical torque or twisting by ABPs modifies helical pitch uniformly or non-uniformly along actin filaments. While the previous study by Prochniewicz et al. (Prochniewicz et al., 1996) demonstrates gelsolin’s effects on filament torsional rigidity, further mechanical experiments utilizing magnetic tweezers or optical tweezers can determine the point where torsional rigidity and helical pitch are no longer uniform. Moreover, we speculate that gelsolin causes overtwisting of filament half-pitch as this is the most probable consequence of disrupting longitudinal contacts postulated by models (Schramm et al., 2017).