Visualizing molecules of functional human profilin

Profilin-1 (PFN1) is a cytoskeletal protein that regulates the dynamics of actin and microtubule assembly. Thus, PFN1 is essential for the normal division, motility, and morphology of cells. Unfortunately, conventional fusion and direct labeling strategies compromise different facets of PFN1 function. As a consequence, the only methods used to determine known PFN1 functions have been indirect and often deduced in cell-free biochemical assays. We engineered and characterized two genetically encoded versions of tagged PFN1 that behave identical to each other and the tag-free protein. In biochemical assays purified proteins bind to phosphoinositide lipids, catalyze nucleotide exchange on actin monomers, stimulate formin-mediated actin filament assembly, and bound tubulin dimers (kD = 1.89 µM) to impact microtubule dynamics. In PFN1-deficient mammalian cells, Halo-PFN1 or mApple-PFN1 (mAp-PEN1) restored morphological and cytoskeletal functions. Titrations of self-labeling Halo-ligands were used to visualize molecules of PFN1. This approach combined with specific function-disrupting point-mutants (Y6D and R88E) revealed PFN1 bound to microtubules in live cells. Cells expressing the ALS-associated G118V disease variant did not associate with actin filaments or microtubules. Thus, these tagged PFN1s are reliable tools for studying the dynamic interactions of PFN1 with actin or microtubules in vitro as well as in important cell processes or disease-states.


Design of tagged (PFN1)
Critical aspects of PFN1 function occur through conserved binding sites for PIP lipids, actin monomers, PLP motifs or microtubules ( Figure 1A and B). To directly monitor PFN1 activities, we engineered two genetically encoded versions of the protein visible with either an mApple probe or self-labeling ligands that bind to Halo-tags. We chose the mApple probe over standard fluorescent proteins (e.g. GFP) because it is bright, relatively stable, and avoids overlap with the excitation and emission ranges of well-characterized actin-labels used in biochemical assays (e.g. Oregon-Green; OG, and pyrenelabeled actin; Shaner et al., 2008;Bindels et al., 2017). We fused each tag to the N-terminus of PFN1 flanked by a linking sequence, based on strategies used in the Michael Davidson Fluorescent Protein Collection (https://www.addgene.org/fluorescent-proteins/davidson/; Figure 1C), then expressed and purified recombinant versions of untagged PFN1 or mApple-PFN1 (mAp-PFN1) to compare the effects of the tag in several biochemistry assays. SDS-PAGE and gel filtration analyses revealed that the tag-free and tagged versions were of the expected size and highly pure ( Figure 1D and

PFN1 and mAp-PFN1 equally promote formin-based actin assembly
In addition to its role as a strong inhibitor of actin filament assembly (Figure 3), PFN1 can simultaneously bind actin monomers and proline-rich motifs (PLP) present in cytoskeletal regulatory proteins to stimulate actin filament assembly (Kovar et al., 2006;Paul and Pollard, 2008;Breitsprecher and Goode, 2013;Zweifel and Courtemanche, 2020). We used bulk pyrene fluorescence assays to assess whether mAp-PFN1 was capable of enhancing actin filament assembly through the PLP     Figures 3D and 4A). Actin filaments assembled with the constitutively active formin mDia1(FH1-C) and either protein resulted in a strong enhancement in actin polymerization ( Figure 4A). Using TIRF microscopy to directly monitor formin-based actin filament assembly we measured elevated levels of actin fluorescence from reactions that contained formin and either protein compared to controls ( Figure 4B-C and Figure 4-figure supplement 1A). TIRF reactions containing actin and mDia1(FH1-C) had an average of 659.3 ± 98.3 filaments, which was significantly higher (p=0.014, ANOVA) than control reactions lacking formin or either PFN1 ( Figure 4D). Consistent with previous reports that PFN1 also suppresses formin nucleation, reactions containing formin and either tag-free (42.7 ± 9.8 filaments) or mAp-PFN1 (48.3 ± 14.5 filaments) had statistically fewer filaments than control reactions containing formin and actin (p=0.0004, ANOVA) (Kovar et al., 2006;Breitsprecher et al., 2012;Henty-Ridilla et al., 2016).   To assess whether mAp-PFN1 was capable of stimulating formin-based actin assembly, we quantified actin filament elongation rates from TIRF movies ( Figure 4E and Figure 4-figure supplement 1B). We measured actin filaments elongating at two different speeds that correspond to the recorded rates of unassisted growth (10.1 ± 0.4 subunits s -1 µM -1 ) and mDia1(FH1-C)-assisted growth in the presence of PFN1 (PFN1: 42.7 ± 9.8 subunits s -1 µM -1 ; or mAp-PFN1: 48.3 ± 14.5 subunits s -1 µM -1 ) (Kovar et al., 2006;Henty-Ridilla et al., 2016). No substantial change to the number of mAp-PFN1 end-association events in the presence was observed with formin in two-color TIRF assays. These data demonstrate that PFN1 and mAp-PFN1 stimulate formin-based actin assembly to comparable levels.
The combined regulatory activities of PFN1 on microtubule polymerization may be explained by several mechanisms, including that it binds or stabilizes tubulin dimers at growing microtubule plusends. To test for a direct interaction between mAp-PFN1 and unlabeled tubulin dimers, we performed fluorescence polarization (k D = 1.89 ± 0.28 µM) ( Figure 5G). This affinity is 5.9-fold stronger than its affinity for polymerized microtubules (k D = ~10 µM) (Henty-Ridilla et al., 2017). We used TIRF microscopy to evaluate whether microtubule plus-ends were enriched with mAp-PFN1 ( Figure
To assess whether mAp-PFN1 also restored cell-based functions, we plated cells on micropatterns to standardize the shape and measured cell morphology ( Figure 6E and F and Figure 6figure supplement 1G), the fluorescence of actin filaments ( Figure 6G and H) or the fluorescence of microtubules ( Figure 6I and J) in endogenous, knockout, and PFN1 (-/-) cells transfected with tagfree or mAp-PFN1 plasmids. Notably, traditional assays for assessing the cytoskeleton in these cells are difficult to perform because N2a cells do not efficiently migrate or divide. PFN1 (-/-) knockout cell morphology significantly deviated from all cells expressing PFN1 (endogenous, tagged, or untagged; p=0.01; ANOVA; Figure 6E and F). In addition, knockout cells displayed strikingly aberrant actin filament (p=0.02; ANOVA; Figure 6G and H) and microtubule networks (p=0.002; ANOVA; Figure 6I and J). Compared to PFN1 (+/+) , the cell morphology, architecture of actin filaments, and shape of the    Figure 6E and J and Figure 6-figure supplement 1G). This demonstrates that mAp-PFN1 complements cytoskeleton-based activities in cells.

Live-cell visualization of Halo-PFN1
The discrete cellular localization of PFN1 has been difficult to determine because the endogenous protein is present at high cellular concentrations (121 µM; Figure 6D; Pollard et al., 2000;Funk et al., 2019;Skruber et al., 2020). Therefore, to further confirm the utility of the fluorescently-tagged protein, we performed live-cell imaging of PFN1 deficient N2a cells transfected with Halo-PFN1 and markers for actin or microtubules (Figure 8, Figure 8-figure supplements 1 and 2). We performed a titration of Janelia Fluor-646 (JF-646) ligand to limit the amount of cellular PFN1 signal and to test whether molecules of the protein could be visualized coincident with actin or microtubules ( Figure 8A). At high concentrations of JF-646, Halo-PFN1 was localized to the nucleus and throughout the cytoplasm ( Figure 8A). Single molecules of Halo-PFN1 were identified at 1-10 nM ligand, and localized to microtubules in ~60% of N2a cells ( Figure 8A-C).
To explore whether this localization was cell-type specific, we overexpressed Halo-PFN1 in additional mammalian cell types (Figure 8-figure supplement 1A). In NIH 3T3 cells, Halo-PFN1 localized to actin filaments in all cells assessed and only occurred with microtubules at sites where both polymers overlapped (Figure 8-figure supplement 1A, B). In contrast, Halo-PFN1 associated more frequently with microtubules than actin filaments for U2OS and N2a cells (Figure 8-figure supplement 1A-D). Notably, for N2a cells, the localization of Halo-PFN1 to actin filaments only occurred at sites where both polymers overlapped (Figure 8-figure supplement 1D). Expressing Halo-PFN1 over endogenous PFN1 did not stimulate further microtubule association (Figure 8-figure supplement 1E).
To shift cellular PFN1 pools from the endogenous localization, we transfected PFN1 (-/-) cells with Halo-PFN1 plasmids harboring point mutations that disrupt specific PFN1 functions (Ezezika et al., 2009;Rotty et al., 2015;Suarez et al., 2015;Henty-Ridilla et al., 2017;Liu et al., 2022). PFN1 colocalized with microtubules >50% of cells expressing the actin monomer binding deficient PFN1, Halo-PFN1(R88E) (n>50 cells from 5 separate experiments; Figure 8B-C). Line scans projected as kymographs further demonstrate that at least some protein is associated with microtubules ( Figure 8D and Figure 8-figure supplement 2). However, it is still difficult to resolve whether these molecules are transiently associated or move along the microtubule lattice Figure 8-video 1. Surprisingly, PFN1 colocalized with microtubules in >75% of cells expressing the formin/poly-L-proline binding deficient Y6D mutation ( Figure 8B and C; Ezezika et al., 2009;Henty-Ridilla et al., 2017;Liu et al., 2022). Microtubules in these cells appear less dynamic than wild-type or R88E-expressing cells, which may strengthen the hypothesis that PFN1-microtubule interactions stabilize microtubule dynamics  Figure 8D and Figure 8-video 1). Significantly less PFN1 was associated with microtubules in cells expressing the microtubule-binding deficient, and ALS-relevant, G118V mutation ( Figure 8B and C). These cells displayed PFN1 accumulations may be disease-relevant aggregates that are hypothesized to contribute to ALS onset (Wu et al., 2012;Figley et al., 2014). A line scan placed over a Halo-PFN1(G118V) expressing cell does not show much colocalization with microtubules ( Figure 8D, Figure 8-figure supplement 2, and Figure 8-video 1). Thus, the localization of molecules of PFN1 can be shifted from actin-or microtubule-binding activities with specific point mutations.
The online version of this article includes the following source data and figure supplement(s) for figure 6: Source data 1. Full blots and additional cell views associated with Figure 6.

Discussion
The lipid, actin, and microtubule regulating capabilities of profilin (PFN1) position it as a critical convergence point for major cell signaling pathways. Here we engineered and characterized genetically encoded tagged PFN1 proteins that are fully functional for interactions with important signaling lipids, binding and exchanging nucleotides on actin monomers, stimulating mDia1-based actin filament assembly, and binding to microtubules ( Figure 8E). In cells, Halo-PFN1 fully compensates for the loss of the endogenous protein and restores normal cell morphology and actin filament and microtubule array architectures. A titration of specific self-labeling JF-646 ligand allowed us to directly localize a subset of PFN1 molecules for the first time in living cells. Overall, these tools directly illuminate functions of PFN1 that could not be deduced previously through indirect mechanisms or observed due to high cellular concentrations.
Fluorescent PFN1 also offers many advantages for dissecting direct PFN1-binding relationships. First, mAp-PFN1 is well-suited for fluorescence polarization assays to determine the affinities of PFN1 for binding actin monomers or tubulin dimers. This assay may now be expanded to disease-relevant mutations in PFN1 or to determine the binding constants of canonical ligands (i.e. formin, VASP, PLP, and PIPs) . Second, PFN1 was tagged with two entirely different genetically encodable probes through the same flexible linker. This validates that the tagging strategy may be applied with new fluorophores of different wavelengths or sizes (Oliinyk et al., 2019). Furthermore, when paired with appropriate and well-placed fluorophores, functional fluorescently tagged PFN1 may now be combined with FRET. Such studies paired with elegantly engineered formins may elucidate more details of how PFN1 interacts with PLP-tracts of different lengths or compositions at nm resolution (Courtemanche and Pollard, 2013;Aydin et al., 2018;Zweifel and Courtemanche, 2020).
Perhaps the greatest benefit of these tools is the ability to decipher mechanisms of PFN1 function with actin or microtubules in important cell processes. Dye-ligand titrations to illuminate Halo-PFN1 effectively overcome many of the problems associated with visualizing highly concentrated fluorescent proteins in cells. While advantageous, it is worth reiterating that only a fraction of the total cellular PFN1 is visible with this approach. As a consequence, Halo-PFN1 is the most appropriate for deciphering high affinity cellular mechanisms that occur at high concentrations and discrete cellular locations. Specifically, Halo-PFN1 is suitable for detecting interactions with microtubules, vesicles, stress fibers, and centrosomes rather than with actin monomers in the cytoplasm. However, the utility of these probes in the cytoplasm or at the leading edge may be achieved with optimization, genetics, different cell types, or clever techniques that stimulate PFN1 (Lee et al., 2013;Skruber et al., 2020).
Finally, several disease-variants of PFN1 have been identified in cancer and neurodegenerative disorders (Michaelsen-Preusse et al., 2016;Pimm et al., 2020;Murk et al., 2021). Using a genetic approach, we observed less microtubule association in cells expressing the actin-and microtubulebinding deficient PFN1(G118V) ALS variant. This is the first examination of the effects of an ALSrelated PFN1 at endogenous levels. Future studies will detail how other functional or disease-related mutations influence the role of PFN1 in diverse signaling schemes, associated disorders, and different cell types.

Reagents
All materials were obtained from Fisher Scientific (Waltham, MA) unless otherwise noted.
The online version of this article includes the following video, source data, and figure supplement(s) for figure 8: Source data 1. Halo-profilin microtubule localization counts.   incubated for 30 min at room temperature, and pelleted at 400,000 × g. Collected supernatants and pellets were resuspended in equal volumes (100 µL) of buffer, precipitated in 10% (v/v) trichloroacetic acid, washed with cold 100% acetone, and dissolved in 50 µL of 50 mM Tris-HCl (pH 6.8), 8 M urea, 5% SDS, and 1 mM EDTA. Blots were probed as described in figure legends. Densitometry was performed in Fiji (Schindelin et al., 2012).

Fluorescence polarization binding assays
Reactions with actin (10 nM; unlabeled or OG-labeled) were performed in 1 × PBS (pH 8.0) and 150 mM NaCl and incubated at room temperature for 15 min before recording. Competitive experiments had 10 nM unlabeled actin, 10 nM GFP-Tβ4, and concentrations of either PFN1 protein. Direct tubulin-binding experiments were performed with 10 nM tubulin in 1 × BRB80 (80 mM PIPES, 1 mM MgCl 2 , and 1 mM EGTA; pH 6.8) with 150 mM NaCl, and concentrations of mAp-PFN1, at 4°C. No microtubules were found when the contents of the highest tubulin-containing reaction were spotted onto coverslips and screened for microtubules by epifluorescence microscopy (561 filter) at the end of the experiment. Time-lapse polarization was performed to determine the rate of nucleotide exchange on actin using 2 µM actin (unlabeled), 500 nM ATP-ATTO-488, and 1 µM PFN1 or mAp-PFN1 in NFG (5 mM Tris (pH 8), 0.2 mM CaCl 2 , 0.5 mM DTT) supplemented with MEI (1 mM MgCl 2 , 1 mM EGTA, and 10 mM imidazole-HCl (pH 7.0)), similar to Colombo et al., 2021. Polarization was recorded every 7.5-8 s for 30 min. Actin was dialyzed for 30 min in NFG and pre-cleared before use. Polarization of OG, GFP or ATTO-488 probes was determined at 440 nm/510 nm, whereas assays using the mAp-PFN1 were set to 568 nm/592 nm. Assays were performed in a plate reader (Tecan, Männedorf, Switzerland). All proteins and reagents were pre-cleared via centrifugation at 279,000×g. Non-linear curve fits were performed using data normalized so that the smallest mean in each data set was defined as zero. Data were fit to the following curve using least squares regression with no constraints: Y = Y 0 − B max *(X/(K D + X)).

Bulk actin assembly assays
Bulk assembly assays were performed by combining 2 μM Mg-ATP actin (5% pyrene labeled), proteins or control buffers, and initiation mix (2 mM MgCl 2 , 0.5 mM ATP, 50 mM KCl). Reactions for each replicate were initiated simultaneously by adding actin to reactions using a multichannel pipette. Total fluorescence was monitored at 365 nm/407 nm in a plate reader. Recorded values were averaged between three replicates.

Determining the concentration of PFN1 in N2a cells
The concentration of PFN1, mAp-PFN1, or Halo-PFN1 was determined using quantitative Western blots from 100 K confluent cells lysed in 200 µL 2 × Laemmli buffer. The total mass of PFN1 was determined from 50 µL of lysate and determined from PFN1 or mAp-PFN1 standard curves. The mean cell volume of a typical N2a cell was calculated as 196 µm 3 (1.96 × 10 -13 L) by taking the average XY area of ten well-spread N2a cells and then multiplying by the mean cell thickness in Z (~1 µm) from the same cells, similar to Christ et al., 2010;Cadart et al., 2017. Correction of lysates was not necessary as transfection efficiencies were 70-90% for these cells (Henty-Ridilla et al., 2016;Henty-Ridilla et al., 2017).
Cells were imaged by spinning disk confocal microscopy on an inverted Nikon Ti2-E microscope (SoRa; Nikon Instruments, Melville, NY) equipped with 488 nm, and 561 nm wavelength lasers, a CF160 Plan Apo 60 × 1.4 NA oil-immersion objective, a CSU-W1 imaging head (Yokogawa Instruments, Tokyo, Japan), a SoRa disk (Nikon Instruments, Melville, NY), and a Prime BSI sCMOS camera with a pixel size of 6.5 µm/pixel (Teledyne Photometrics, Tucson, AZ). Artificial intelligence denoise and 40 iterations of Richardson-Lucy deconvolution was applied to 7-10 µm Z-stacks acquired with identical laser exposures and power using Nikon Elements software. Images for cell morphology (saturated actin signal), actin fluorescence or microtubule fluorescence quantification were converted to 8-bit grayscale, binarized, and counted (RawIntDen) in Fiji software.

Data analyses and availability
GraphPad Prism 9 (GraphPad Software, San Diego, CA) was used for analyses and statistical tests. The design, sample size, and statistics used for each experiment are in each figure legend. All datasets passed tests for normality. Individual data points are shown in each figure shaded by replicate. Source datasets for all quantitative data are linked in the figure legends. The Henty-Ridilla Zenodo contains the original large image files and is available upon confirmation that users will follow CC-BY licensing guidelines for reuse at: http://doi.org/10.5281/zenodo.5329584. Scholar Award, Amyotrophic Lateral Sclerosis Association Starter Grant (20-IIP-506), and the National Institutes of Health, GM133485. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.