Profilin Directly Promotes Microtubule Growth through Residues Mutated in Amyotrophic Lateral Sclerosis

Abstract

Profilin is an abundant actin monomer-binding protein with critical actin regulatory roles in vivo [1, 2]. However, profilin also influences microtubule dynamics in cells, which may be mediated in part through its interactions with formins that in turn bind microtubules [3, 4]. Specific residues on human profilin-1 (PFN1) are mutated in patients with amyotrophic lateral sclerosis (ALS) [5, 6]. However, the observation that some ALS-linked PFN1 mutants fail to alter cellular actin organization or dynamics [5-8] or in vitro actin-monomer affinity [9] has been perplexing, given that profilin is best understood as an actin regulator. Here, we investigated direct effects of profilin on microtubule dynamics and whether ALS-linked mutations in PFN1 disrupt such functions. We found that human, fly, and yeast profilin homologs all directly enhance microtubule growth rate by several-fold in vitro. Microtubule stimulatory effects were unaffected by mutations in the canonical actin- or poly-proline-binding sites of profilin. Instead, microtubule activities depended on specific surface residues on profilin mutated in ALS patients. Furthermore, microtubule effects were attenuated by increasing concentrations of actin monomers, suggesting competition between actin and microtubules for binding profilin. Consistent with these biochemical observations, a 2-fold increase in the expression level of wild-type PFN1, but not the ALS-linked PFN1 mutants, increased microtubule growth rates in cells. Together, these results demonstrate that profilin directly enhances the growth rate of microtubules. They further suggest that ALS-linked mutations in PFN1 may perturb cellular microtubule dynamics and/or the coordination between the actin and microtubule cytoskeletons, leading to motor neuron degeneration.

Keywords: ALS; actin; crosstalk; microtubule; neurodegeneration; profilin; structure; tubulin.

Figure 1. Profilin increases the elongation rate of microtubules in vitro

(A) Representative field of view from a control TIRF reaction containing biotinylated GMP-CPP MT seeds and 10 μM free tubulin (30% Alexa-647 labeled) (left), with kymograph (middle) and MT length traces (right) showing stochastic growth and shortening of MTs. (B) Representative fields of view for reactions as in (A) except in the presence of 5 μM wildtype PFN1. Length scale bars, 10 μm. Time scale bars, 300 s. (C) MT elongation rates from TIRF reactions as in (A) and (B), with and without 5 μM PFN1 (n = 73 MTs per condition). (D) Duration of growth phase for individual MT elongation events (n = 47–62 MTs per condition). (E) Maximum length to which MTs grew before undergoing catastrophe (n = 30 MTs per condition). (F) MT stability index: rescue/catastrophe frequency (measured from n = 50 MTs per condition). Error bars, SE. (G) MT elongation rates measured in TIRF assays as in (A) at variable concentrations of PFN1 (n = 69–78 MTs per condition). (H) Linear regression of MT elongation rates from TIRF assays as in (A) at variable concentrations of free tubulin, with and without 5 μM PFN1 (n = 30–73 MTs per condition). Shaded areas represent 95% confidence intervals and error bars indicate SE. (I) Total number of MTs present in reactions as in (H) (measured from n = 3 reactions per condition). (J) Inset, representative Coomassie-stained SDS-PAGE gel of supernatants and pellets from MT co-sedimentation assay containing 5 μM PFN1 and variable concentrations of MTs (0–50 μM), with corresponding binding curve. (K) Field of view from TIRF reactions containing biotinylated GMP-CPP MT seeds, 10 μM free tubulin (30% Alexa-647 labeled), and 5 μM PFN1. MTs were fixed with 0.1% glutaraldehyde and PFN1 was visualized by immunofluorescence. Magnified view of individual MTs decorated by PFN1 are shown below each field of view. Scale bars, 10 μm. (L) Quantification of PFN1 fluorescence along MTs by line scan analysis, from reactions as in (K) (n = 50 MTs per condition). (M) MT elongation rates from TIRF reactions performed as in (A) in the presence or absence of 5 μM human PFN1 (blue), S. cerevisiae Profilin (Pfy1) (pink), or D. melanogaster Profilin (Chickadee) (grey) (n = 57–72 MTs per condition). Error bars indicate 95% confidence intervals except where noted. Significant differences by one-way ANOVA with Bartlett’s correction for variance: ns, not significantly different from control; a, compared with control (P < 0.05); b, compared with PFN1 (P < 0.05). See also Figure S1 and Movie S1 and Movie S2.

Figure 2. Effects of ALS-associated Profilin mutants on microtubules in vitro

(A) Left: View of Profilin (blue) interacting with actin (grey) and with the poly-L-proline (PLP) region of VASP (yellow), modeled using PDB ID codes 2PAV [15] and 2BTF [10]. Middle: Magnified view showing Profilin surfaces contacting actin (cyan) and PLP (yellow), and highlighting the positions of key residues required for actin-binding (R88; green) and PLP-binding (Y6; orange). Right: Close up view of Profilin highlighting the positions of five ALS-linked residues: C71 (blue), mutations in which lead to protein aggregation; H120 (purple), which mediates actin-binding; and M114, E117, and G118, which mediate MT binding and regulation. (B) MT elongation rates from TIRF reactions containing biotinylated GMP-CPP MT seeds and 10 μM free tubulin (30% Alexa-647 labeled), in the presence and absence of 5 μM wildtype (blue), Y6D (chartreuse), or R88E (green) human PFN1 (n = 66–77 MTs per condition). (C) Representative MT length traces from TIRF reactions containing biotinylated GMP-CPP MT seeds and 10 μM free tubulin (30% Alexa-647 labeled) in the absence (control) or presence of 5 μM wildtype or ALS-mutant PFN1s. (D) MT elongation rates measured from TIRF reactions in the presence and absence of 5 μM wildtype or ALS-mutant PFN1s (n = 50–83 MTs per condition). (E) Time MTs spent growing before undergoing catastrophe (n = 50 MTs per condition). (F) MT stability index: number of rescue events/number of catastrophe events. Error bars, SE. (G) Binding of 5 μM wildtype or ALS-mutants of PFN1 to variable concentrations of MTs in co-sedimentation assays (n = 3 replicates per concentration). Error bars, SE. (H) Table of binding data from co-sedimentation analyses as in (G) comparing wildtype and ALS-mutant PFN1s. One-site binding parameters and fit-values for MT binding are derived from the curve fits in (G). All experiments were performed from at least three times. Error bars indicate 95% confidence intervals except where noted. Significant differences by one-way ANOVA with Bartlett’s correction for variance: ns, not significantly different from control; a, compared with control (P < 0.05); b, compared with PFN1 (P < 0.05). See also Figures S1 and S2 and Movie S1.

Figure 3. Profilin effects on microtubules are attenuated with the presence of actin monomers

(A) Representative views from TIRF assays containing biotinylated GMP-CPP MT seeds, 10 μM free tubulin (30% Alexa-647 labeled), 5 μM wildtype PFN1, and variable concentrations of monomeric actin bound to Latrunculin A (LatA). Scale bars, 10 μm. (B–H) MT elongation rates for TIRF conditions described in (A) in the presence of 5 μM wildtype PFN1 (B), Y6D PFN1(C), R88E PFN1 (D), M114T PFN1 (E), G118V PFN1 (F), E117G PFN1 (G), and H120E PFN1 (H) (n = 50 MTs per condition). The presence of LatA (without actin) did not alter MT dynamics (B-H). (I) Summary of dose-dependent loss of MT growth-enhancing effects for each Profilin mutant in the presence of actin monomers generated from data in (B–H). All experiments were performed at least three times. Error bars indicate 95% confidence intervals. Significant differences by one-way ANOVA with Bartlett’s correction for variance: ns, not significantly different from control; a, compared with control containing LatA (P < 0.05); b, compared with 5 μM wildtype PFN1 (P < 0.05). See also Figure S3.

Figure 4. Cellular effects of wildtype or ALS-relevant PFN1 mutations on MT dynamics

(A) Representative kymographs generated from live-cell tracking of EB1-GFP comets from control cells and cells expressing wildtype or mutant PFN1 incapable of binding actin monomers (R88E) (n = 50–193 cells per condition). Length scale bar, 5 μm. Time scale bar, 50 s. (B) Quantitative tracking analysis of EB1-GFP velocities from live-cell TIRF microscopy. Each data point graphed is the mean EB1-GFP velocity for a single cell, averaged from >3000 EB1-GFP comets tracked during a period of 10 min. (C) TIRF montage displaying the localization of actin (mChr-Utrophin) and MT plus-ends (EB1-GFP) in filopodia at the cell periphery. Scale bar, 10 μm. (D) Instances of MTs entering filopodia denoted by EB1-GFP (arrowheads) were counted over the lifetime of filopodia. EB1-GFP comets were present in individual filopodia were scored over the 10 min imaging period (n = 50 filopodia from 10 cells per condition). (E) Number of filopodia (n = 30–50 cells), and (F) filopodia lifetimes (n = 250 filopodia from 10 cells per condition). (G) Live-cell EB1-GFP velocity tracking analysis from control or siPFN1 cells. (H) Inset: Western blots of cell extracts were used to determine relative PFN1 and α-tubulin levels for ALS-associated PFN1 mutants and controls. Quantification of band intensities from Western blots PFN1 levels (grey) and for α-tubulin levels (teal) (n = 3 blots per treatment). Error bars represent SE. (I) Live-cell TIRF microscopy of EB1-GFP comets was used to track MT plus-ends from control or cells with enhanced levels of PFN1 or PFN1 mutations associated with ALS (n = 50–193 cells per condition). (G) Model for the distribution of Profilin to actin or MTs. MTs, actin, and actin nucleators (Formins and the Arp2/3 complex) compete for Profilin. Thus, Profilin is a molecular oscillator capable of regulating the dynamics of individual MTs, the availability of actin monomers to specific actin nucleation systems, and due to competition for Profilin between MTs and actin monomers, acts as a regulator of actin-MT crosstalk. All experiments shown were performed in at least three independent trials. Error bars indicate 95% confidence intervals. Significant differences by one-way ANOVA with Bartlett’s correction for variance: ns, not significantly different from control; a, compared with control (P < 0.05); b, compared with wildtype PFN1 (P < 0.05). See also Figure S4 and Movie S3.