Purification of actin monomers (G-actin) from rabbit skeletal muscle acetone powder (PelFreeze Biologicals Inc., Rogers, AR, USA) was performed through gel filtering G-actin over Sephacryl S300 size exclusion column equilibrated in buffer A (0.2 mM CaCl₂, 1 mM NaN₃, 2 mM Tris-HCl pH 8.0, 0.2 mM ATP, and 0.5 mM DTT) as previously described (Kang et al., 2012; Castaneda et al., 2018; Castaneda et al., 2019; Heidings et al., 2020). G-actin bound with Ca²⁺ was subjected to cation exchange by ethylene glycol-bis(β-aminoethyl ether)-N,N,N,N-tetraacetic acid (EGTA) to Mg²⁺ with the addition of 0.2 mM EGTA and MgCl₂ concentration equal to the initial G-actin concentration plus 10 μM. Following the cation exchange, polymerization of G-to actin filaments (F-actin) was performed (Kang et al., 2012; Castaneda et al., 2018; Castaneda et al., 2019; Heidings et al., 2020).

To form actin bundles, unlabeled F-actin was allowed to polymerize for 1 to 2 h and subjected to high concentrations of either cations or crowding agents. Cation-induced bundles were formed by the use of 0.1 volume 10X MI buffer (300 mM Mg²⁺, 100 mM imidazole, pH 7.0, 10 mM ATP, and 10 mM DTT). Formation of depletion-induced bundles was achieved by the addition of crowding buffer [Ficoll 70 20% w/w or polyethylene glycol (PEG) 5% w/w] in 1X KMI (50 mM KCl, 2 mM Mg²⁺, 10 mM imidazole pH 7.0, 1 mM ATP, and 1 mM DTT). The crowding agent concentrations for the experiments were Ficoll 70 (Thermo Fisher Scientific Inc., Waltham, MA, USA), 1% w/w (∼0.16 mM), 5% w/w (∼0.79 mM), and 10% w/w (∼1.56 mM), or PEG, 8 kDa (Thermo Fisher Scientific Inc., Waltham, MA, USA), at 1% w/w (∼1.38 mM), 5% w/w (∼6.94 mM), and 10% w/w (∼13.88 mM). These specific crowding conditions were chosen based on the intracellular volume occupancy of ∼5–40% (Ellis and Minton, 2003; Kuznetsova et al., 2014), while (Mg²⁺) (10–50 mM) was selected based on previous work shown in Castaneda et al. (2018) and intracellular (Mg²⁺) (Romani, 2011).

Mica substrates were freshly cleaved, and the addition of 50 μl of positively charged binding agent (3-aminopropyl)triethoxysilane (APTES) (0.1% v/v) was done by pipetting onto the mica surface and allowing to bind for 10 min (Liu et al., 2005; Lyubchenko, 2011) to favor actin bundle adhesion on the substrate. Prior to the addition of bundle samples, the APTES-coated mica was rinsed with a gentle stream of ddH₂O and dried with compressed air (Liu et al., 2005; Lyubchenko, 2011). AFM experiments were performed on the coated substrates to check their topographical and mechanical homogeneities and their low rugosity as shown in Supplementary Figure S1. Following the drying of the mica surface, 5 μl of the bundle sample in the respective crowding or cation environment, at a concentration of ∼10 μM, was placed on top the APTES-coated mica surface and allowed to bind for ∼5 min (Liu et al., 2005). Then, a corresponding sample buffer was added onto the bundle-bound mica surface at a volume of ∼50 μl.

The nanomechanical and height changes of the bundles in the liquid environment were determined using a Dimension FastScan AFM (Bruker, Santa Barbara, CA, USA) at a constant temperature of 20°C. Imaging was first conducted in Peak Force Quantitative Nanomechanical Mapping (PFQNM) tapping mode (256 pixels × 256 pixels) at a scan size ranging from 1 to 5 μm. The AFM cantilever tips used for the experiments were gold-coated FASTSCAN-C tips (Bruker, Santa Barbara, CA, USA) with a triangular tip shape, calibrated tip radius of ∼5 nm, nominal spring constant of 0.8 N/m, and nominal resonant frequency of ∼300 kHz. The nanomechanical experiments were performed by force curve measurements after proper calibration of the setup. Cantilever tips were calibrated as previously described in Heu et al. (2012). Briefly, prior to each measurement, the deflection sensitivity (by capturing three force curves on a non-compliant part of the sample and averaging the three corresponding slopes on the linear portions of the curves) and the spring constant (by tuning the cantilever at least 10 µm away from the surface and spotting the resonance peak) were calculated for each probe in dilute, crowded, or cation buffer conditions. The tip radius for each probe was determined before and after experiments using a reference titanium roughness sample (TipCheck Sample, Bruker, USA). A PeakForce frequency of 0.25 kHz was used in order to maximize the contact time between the tip and the sample, with a PeakForce amplitude of ∼1 µm. The loading force was adjusted to 800 pN. A small approach velocity of 6 μm/s was used to minimize the contribution of viscosity to the mechanical response. Thus, the hydrodynamic damping hysteresis was almost suppressed and had no impact on the relative Young’s modulus measurements. For the determination of the relative Young’s modulus, the retraction curves were used in response to the retraction and the approach curves showing a similar negligible contribution of viscosity for cation- and depletion-induced bundles in crowding and cation environments. Regarding the geometry of the tip and the negligible adhesion in the force-curve, the relative Young’s modulus was determined using a classic Hertz model to fit the force curves: F =   4 3   E ( 1 − v 2 ) R δ 3 / 2 (1) where E is the relative Young’s modulus of the actin bundle, v is the estimated Poisson ratio of the bundle, R is the nominal radius of the tip, and δ is the indentation depth. The Poisson ratio of actin has been previously investigated to be ∼0.3 for actin filaments and crosslinked bundles (Tseng et al., 2002; Lin et al., 2017; Wang et al., 2020). We estimated our E values with an assumed Poisson value of 0.3. For each condition, at least 150 curves were analyzed. Analysis of the collected force curve bundle measurements was achieved by the use of NanoScope Analysis v. 2.0 software (Bruker). Alterations to the height of individual bundles in crowded or cation environments were analyzed by the profile extracting tool in Gwyddion software (Nečas and Klapetek, 2012).

The statistical significance for each of the actin bundle E and the height measurements were determined using OriginLab v.8.5 software. Multiple analysis of variance and post-hoc Tukey test determined the probability (p-value) showing the significant modulations between samples (notations for p-values: n.s., not significant, *p < 0.05, **p < 0.01, ***p < 0.001).

We used AFM to determine the height and nanomechanical properties of cation-induced actin bundles in the solution of macromolecular crowding agents. Cation-induced bundles were formed by the addition of physiologically relevant divalent cation (10–50 mM Mg²⁺), and then bundles were placed in buffer solutions with varying concentrations of Ficoll or PEG (1–10% w/w). Dilute buffer conditions displayed a mixture of both cation-induced actin bundles and actin filaments (Figure 1A). Furthermore, cation-induced bundle control exhibited short periodic striations along the bundle surface not visualized with the addition of Ficoll or PEG (Figure 1A). Upon addition of Ficoll and PEG (1–10% w/w), the bundles exhibited an increase in height (Figure 1A). Bundle control displayed a height distribution centered at ∼8 nm; however, bundle height was shown to increase with a shift in the distribution center to ∼30 nm in Ficoll (10% w/w) (Supplementary Figure S2A). In addition, cation-induced bundles in varying PEG environments were similarly shown to have a distribution centered at ∼8 nm (Supplementary Figure S2B). When increasing the concentration of PEG to 10% w/w, the bundle height distribution centered at ∼20 nm; however, bundle heights of ∼45 nm can occasionally be observed (Supplementary Figure S2B).

The nanomechanical properties of bundles, such as relative Young’s modulus ( E ), can be obtained from the force curve measurements provided by PFQNM mode for each pixel of the obtained image. Force curve measurements were taken along the center of cation-induced bundles and analyzed for changes to E (Figure 1B). Interestingly, histograms of cation-induced bundle E in crowding revealed that increasing Ficoll and PEG concentrations narrow and shift the distribution to reduced E values (Supplementary Figure S3). The cation-induced bundle control showed an averaged value of E ∼60 ± 9.9 MPa (Figure 2A). The addition of the lowest Ficoll condition (1% w/w) leads to ∼8% reduction in bundle E (∼49 ± 5.8 MPa) (Figure 2A). However, increasing Ficoll up to 10% w/w, the bundles exhibit a significant decrease in E by approximately tenfold (∼5.0 ± 3.7 MPa) (Figure 2A). For cation-induced bundles in PEG, the results showed a more drastic reduction in bundle E with the initial presence of PEG (1% w/w), reducing bundle E by ∼45% (∼32 ± 17 MPa) (Figure 2B). In addition, as the PEG concentration was increased to 10% w/w, a reduction of approximately fourfold in bundle E , with a value of ∼15 ± 9.2 MPa, was determined, as compared to bundle control (∼60 MPa) (Figure 2B).

We set out to determine the height and nanomechanical changes to depletion-induced bundles in solutions with varying (Mg²⁺) using AFM. We formed bundles by the addition of crowding agents Ficoll (20% w/w) or PEG (5% w/w) and subjected the bundles to divalent cation (Mg²⁺) (10–50 mM). Both Ficoll- and PEG-induced actin bundle controls did not visibly display short periodic striations as previously shown in cation-induced bundles, possibly due to the initial presence of crowding agents (Figures 3A,B). The Ficoll-induced actin bundle height was shown to increase at 50 mM Mg²⁺, while the PEG-induced bundle height increased at both 30 and 50 mM Mg²⁺ (Figures 3A,B). The Ficoll-induced bundle control height was shown to center at ∼7 nm and showed a broadening of height distribution at the lowest 10 mM Mg²⁺ (Supplementary Figure S4A). At 30 mM Mg²⁺, bundle height distribution was shown to be mainly centered at ∼8 nm, while 50 mM Mg²⁺ bundle height distribution is shown to center at ∼14 nm (Supplementary Figure S4A). The PEG-induced bundle control height was measured to be distributed at ∼10 nm (Supplementary Figure S4B). However, increasing (Mg²⁺) led to significant alterations in PEG-induced bundle height observed at 30 mM (Mg²⁺), with a distribution centered at ∼30 nm and with occasional bundles observed at ∼60 nm (Supplementary Figure S4B). Upon increasing the (Mg²⁺) to 50 mM, the bundle height shifts in distribution to ∼14 nm (Supplementary Figure S4B).

Next, we analyzed the relative Young’s modulus ( E ) of depletion-induced actin bundles in varying (Mg²⁺) (10–50 mM) (Figure 3C). We collected and analyzed force curve measurements to determine the variations of depletion-induced bundles E as previously performed with cation-induced bundles. The histograms of Ficoll-induced actin bundles demonstrated that the bundles experience a narrowing in overall distribution and a shift to lower values of E as the (Mg²⁺) increases (Supplementary Figure S5A). In addition, PEG-induced bundles showed a similar behavior of narrowing and reduced E values with rising (Mg²⁺) (Supplementary Figure S5B). The Ficoll-induced bundle control showed an average of E ∼18 ± 6.6 MPa (Figure 4A). The addition of initial (Mg²⁺) (10 mM) demonstrated a slight reduction in Ficoll-induced bundle E , with a reduction of ∼10% (∼16 ± 2.2 MPa) (Figure 4A). However, increasing the (Mg²⁺) up to 30 mM showed the greatest change with bundle E ∼7 ± 4 MPa, an approximately twofold reduction (Figure 4A). On the other hand, the PEG-induced bundle control E showed an average of ∼35 ± 15 MPa (Figure 4B). At the lowest (Mg²⁺) (10 mM), bundle E was measured to be ∼20 ± 9.7 MPa, while increasing (Mg²⁺) to 30 and 50 mM displayed a significant reduction in E of bundles, with the lowest E ∼4 ± 1 MPa for 30 mM Mg²⁺ (Figure 4B).