Article

Two Deafness-Causing Actin Mutations (DFNA20/ 26) Have Allosteric Effects on the Actin Structure Lauren Jepsen,1,2,3 Karina A. Kruth,4 Peter A. Rubenstein,4 and David Sept2,3,* 1 Bioinformatics Graduate Program, 2Department of Biomedical Engineering, and 3Center for Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, Michigan; and 4Department of Biochemistry, University of Iowa Carver College of Medicine, Iowa City, Iowa

ABSTRACT Point mutations in g-cytoplasmic actin have been shown to result in autosomal-dominant, nonsyndromic, earlyonset deafness. Two mutations at the same site, K118M and K118N, provide a unique opportunity to compare the effects of two dissimilar amino acid substitutions that produce a similar phenotype in humans. K118 resides in a helix that runs from K113 to T126, and mutations that alter the position, dynamics, and/or biochemistry of this helix can result in a wide range of pathologies. Using a combination of computational and experimental studies, both employing yeast actin, we find that these mutations at K118 result in changes in the structure and dynamics of the DNase-I loop, alterations in the structure of the H73 loop as well as the side-chain orientations of W79 and W86, changes in nucleotide exchange rates, and significant shifts in the twist of the actin monomer. Interestingly, in the case of K118N, the twist of the monomer is nearly identical to that of the F-actin protomer, and in vitro polymerization assays show that this mutation results in faster polymerization. Taken together, these results indicate that mutations at this site give rise to a series of small changes that can be tolerated in vivo but result in misregulation of actin assembly and dynamics.

INTRODUCTION Point mutations in g-cytoplasmic actin (DFNA20/26) have been shown to cause autosomal-dominant, nonsyndromic, early-onset deafness (1–3). Of the 11 known DFNA20/26 actin mutations, two are unique in that they occur in the same residue. Interestingly, two families show mutations (K118M in one family and K118N in the other) at the same amino acid, which provides an interesting opportunity to compare the effects of two dissimilar amino acid substitutions that produce a similar phenotype in humans. In the first case, a charged but amphiphilic residue is replaced by a more hydrophobic species, whereas in the second case the mutation replaces the original residue with a smaller, neutral, but still hydrophilic residue. The similar phenotype produced by such different amino acid substitutions raises a particularly interesting question regarding the role of K118 in actin function. As outlined in our earlier work (4), K118 may play a previously unsuspected role in actin filament stability. K118 lies within a helix spanning from K113 to T126, extending from the inside to the outside of the actin filament, respec-

Submitted April 18, 2016, and accepted for publication June 9, 2016. *Correspondence: [email protected] Lauren Jepsen and Karina A. Kruth contributed equally to this work. Editor: Enrique De La Cruz. http://dx.doi.org/10.1016/j.bpj.2016.06.012

tively (Fig. 1). The residues of this helix are populated by mutations that lead to autosomal-dominant actinopathies, and our recent work demonstrated that the most interior residue of the helix, K113, is part of an interstrand ionic bond involving E195 of an actin protomer in the opposing strand of the two-stranded actin helix (5). These results suggest that mutations within this helix, such as those at K118, may subsequently alter protomer-protomer interactions at this interface. Furthermore, K118 is located at or near the binding site for the Arp2/3 complex (6,7) and formins (8,9), two important regulators of actin filament dynamics (the first mediates filament branching from a mother filament, and the second nucleates de novo filament formation). The colocalization of so many pathogenic mutations in one helix, coupled with the location of K118 at or near the binding sites of two important filament regulatory proteins, suggests that these mutations may be part of an allosteric regulatory system that controls actin filament stability. Using yeast actins, each carrying one of these two mutations (K118M and K118N), we previously demonstrated that the M and N mutations resulted in very different effects on actin behavior (4). By itself, K118M actin polymerized with the same kinetics as wild-type (WT) actin, whereas K118N polymerized twice as fast. This result was unexpected because the side chains for the amino acids at this site lie on the surface of the protein, away from the regions of globular actin

Ó 2016 Biophysical Society.

Biophysical Journal 111, 323–332, July 26, 2016 323

Jepsen et al.

FIGURE 1 K118 helix in F- and G-actin. (A and B) The helix containing K118 (red) is shown in the context of (A) the F-actin filament and (B) monomeric G-actin. K118 is positioned away from the actin-actin interface in the filament, whereas K113 extends toward this interface, where it makes interstrand ionic interactions. Two tryptophans, W79 and W86, lie within the vicinity of K118 on a neighboring helix. To see this figure in color, go online.

(G-actin) that form protomer-protomer interfaces within the actin filament. Additional experiments in which the K118M/ N mutant actins were polymerized in the presence of the Arp2/3 complex revealed that the mutations resulted in differential effects on Arp2/3-mediated branch formation. As measured by light scattering, polymerization of K118N actin showed almost no stimulation by addition of the Arp2/3 complex. The K118M actin, in contrast, showed some weak Arp2/ 3-dependent stimulation of polymerization, but it was dramatically reduced compared with the WT. We then examined Arp2/3-mediated branching behavior in K118M actin using total internal reflection fluorescence microscopy, which revealed that the branching frequency was drastically reduced. Furthermore, when branching did occur, the vast majority of branches formed near the pointed end of the filament instead of near the barbed end of the filament, where branching normally occurs. Because of the potential importance of the K118 residue for actin allosteric regulation and its association with human disease, we wished to gain further insight into the structural consequences of the mutations that resulted in this altered polymerization behavior and its regulation by the Arp2/3 complex. Using the mutant yeast actins we previously generated, we employed a combination of spectroscopy methods, protease digestion, and molecular modeling to elucidate a structural basis for the kinetics results we previously obtained. In the process, we also gained further insight into the dynamics and behavior of the pathogenic regulatory helix in which K118 lies, and how these mutations influence the overall actin structure. MATERIALS AND METHODS Yeast growth Growth curves for both WT and mutant yeast cells were obtained by growing yeast in liquid YPD medium at 30
C with shaking. Cultures were inoculated to a cell density of OD600 ¼ 0.1 from liquid overnight cul-

324 Biophysical Journal 111, 323–332, July 26, 2016

tures. The OD600 was measured every hour for the first 10 h and then periodically until a plateau was reached. The doubling times were determined from a semi-log plot of the first 10 points, which was then subjected to regression analysis to generate the best slope.

Actin purification Yeast cakes for WT actin preparations were purchased from a local bakery. Mutant yeast strains were created as previously described (1,10). WT and mutant actins were purified as previously described (4). Briefly, actin was purified from clarified cell lysate via DNase I affinity chromatography (with DNase I (grade D; Worthington, Lakewood, NJ) conjugated to Affi-Gel 10 (Bio-Rad)), DE52 anion exchange chromatography (Whatman, Pittsburgh, PA), and polymerization/depolymerization cycling as previously described (11). The concentration of the G-actin was determined by measuring A290 nm using ε290 nm ¼ 25,600 M1 cm1, and the actin was stored in G buffer (10 mM Tris-HCl, pH 7.5, 0.2 mM CaCl2, 0.1 mM ATP, and 0.2 mM dithiothreitol (DTT)) at 4
C for no more than 4 days.

Protease digestion A 0.4 mg/mL sample of actin was digested at room temperature using 0.05 mg/mL trypsin (containing the chymotrypsin inhibitor tosyl phenylalanyl chloromethyl ketone), 0.005 mg/mL subtilisin, or 0.25 mg/mL chymotrypsin (containing the trypsin inhibitor tosyllysine chloromethylketone) for 20, 15, or 10 min, respectively. Reactions were quenched with the addition of 4 SDS buffer, followed by immediate boiling of the sample for 2 min at 100
C. The samples were then subjected to SDS polyacrylamide gel electrophoresis on 12% gels and the proteins were visualized by Coomassie Blue staining to assess potential differences in the protease cleavage patterns.

Circular dichroism Circular dichroism (CD) spectra were captured on a Jasco J-815 CD spectrometer using 5 mM samples of G-actin. Spectra were acquired at 25
C from 190 to 250 nm at a speed of 20 nm/min. At least three spectra were obtained for each sample for at least three different actin preparations. The spectra were baseline-corrected by subtracting the spectrum for the G-buffer blank (10 mM Tris-HCl, pH 7.5, 0.2 mM CaCl2, 0.1 mM ATP, and 0.2 mM DTT), and averaged. Normalization of the K118N spectra to WT spectra was achieved by determining the percentage difference in

Effects of Actin Deafness Mutations millidegrees at 210 and 222 nm, and subtracting that percentage from each data point in the spectrum. The apparent melting temperature (T1/2) of WT, K118M, and K118N actins was determined by measuring the change in ellipticity of the respective G-actin sample in a temperature range of 25–90
C. Ellipticity was measured at 222 nm and temperature was increased at a rate of 2
C per minute. To approximate the temperature at which 50% of the actin was denatured, the data were fit to a two-state model, and the apparent T1/2 value was determined by fitting the data to the Gibbs-Helmholtz equation:


DGi ¼ DHi

     T T 1 þ DCpi T  Tmi  T ln ; Tmi Tmi




where DHi is the van’t Hoff enthalpy and DCpi is the heat capacity for transition (12). However, because actin cannot refold upon cooling, it is not possible to obtain a reversible melting temperature for G-actin, and an apparent melting temperature, T1/2, was calculated as opposed to a true melting temperature, Tm. Therefore, the above equation was used solely for the purpose of approximating relative protein stability and not for thermodynamic analysis.

Nucleotide exchange Unbound ATP was removed from a 25 mM actin solution using a Micro BioSpin P-30 column (Bio-Rad, Hercules, CA), and the actin was incubated with 0.3 mM 1,N6-ethenoadenosine 5-triphosphate at 4
C overnight. Excess nucleotide was removed using a Micro Bio-Spin P-30 apparatus the next day. The subsequent actin solution was diluted with ATP-free G-buffer to a concentration of 1 mM. The exchange was triggered by the addition of ATP to a concentration of 33 mM to the labeled actin. The decay in fluorescence that accompanied release of the etheno-nucleotide from the actin was monitored as a function of time with excitation and emission wavelengths at 340 and 410 nm, respectively. Nucleotide exchange rates were derived by fitting the data to a single exponential expression, A(t) ¼ Aoekt, where Ao represents the initial fluorescence, k represents the decay constant, and t represents time.

Tryptophan fluorescence The intrinsic tryptophan fluorescence was measured using 5 mM WT, K118M, or K118N actins. Assays were performed at 25
C in a final volume of 120 mL in a microcuvette housed in a thermostat-controlled sample compartment of a FluoroMax-3 spectrometer (Jobin Yvon-Spex, Kyoto, Japan). Intrinsic tryptophan fluorescence was measured using an excitation wavelength of 295 nm, and emission was measured in 1 nm intervals from 305 to 400 nm. G-actin measurements were obtained by scanning G-actin samples diluted in G buffer (10 mM Tris-HCl, pH 7.5, 0.2 mM CaCl2, 0.1 mM ATP, and 0.2 mM DTT). We then added 6 mL of 20 F-salts to the G-actin sample to induce polymerization, and after ~20 min, when polymerization had reached the steady-state phase, we rescanned samples to obtain the F-actin measurements. We repeated all scans three times and smoothed them using the Savitzky-Golay method (13), fitting five points in either direction to a trinomial function, and measuring at least three separate samples for each actin. The lmax was then determined for each actin as the wavelength that produced the greatest fluorescence intensity. This experiment was repeated for three different actin preparations, and similar results were achieved each time.

Acrylamide quenching We prepared 5 mM samples of G- or F-actin with acrylamide quencher by first diluting G-actin in either G-buffer (10 mM Tris-HCl, pH 7.5, 0.2 mM

CaCl2, 0.1 mM ATP, and 0.2 mM DTT) or F-buffer (10 mM Tris-HCl, pH 7.5, 0.2 mM CaCl2, 0.1 mM ATP, 2 mM MgCl2, 50 mM KCl, and 0.2 mM DTT), respectively. We then added 3 M acrylamide, diluted in either G- or F-buffer, to bring the final volume of the actin sample to 120 mL and to reach a final actin concentration of 5 mM. The acrylamide was used at concentrations of 0, 25, 50, 75, 100, 150, 200, 250, 300, and 400 mM. To minimize the effect of acrylamide on polymerization of the actin, quencher was only added to F-actin samples ~1 h after polymerization was induced. Intrinsic tryptophan fluorescence was then measured using an excitation wavelength of 295 nm, and emission was measured at 1 nm intervals from 300 to 400 nm. Each sample was scanned three times and the results were smoothed using the Savitzky-Golay method (13) as above. The results were then plotted using the Stern-Volmer equation:

F0 ¼ 1 þ KSV ½Q; F where F0 is the maximum intensity of fluorescence without acrylamide and F represents the maximum intensity of fluorescence in the presence of a given concentration of quencher, [Q] (in this case acrylamide) (14). The ratio of F0/F was plotted against the concentration of acrylamide and the points were fit to a line with the y intercept forced to equal one. The resulting slope was determined to be the quenching constant, KSV. This experiment was repeated using three separate actin preparations, with similar results.

Molecular simulations The Protein Data Bank structure for yeast G-actin was taken from PDB: 1YAG (15). Missing atoms and/or residues were completed using VMD (16) and the nucleotide was set to be ADP with Mg2þ. Models for the K118M/N mutants were created from this structure, again using VMD. To prepare these structures for molecular-dynamics (MD) simulations, we first solvated the systems with TIP3P water and added Naþ and Cl to both neutralize the system charge and set the ionic strength to 50 mM. MD simulations were carried out using NAMD (17). After heating and equilibration steps, we performed simulations using the AMBER99SB force field (18) at 300 K in an NpT ensemble with 1 atm pressure. Bonded hydrogens were fixed so that we could use 2 fs time steps. We employed ˚ cutparticle mesh Ewald (19) for long-range electrostatics, and used a 10 A ˚ switch distance for van der Waals interactions. off and 8.5 A The MD simulations were carried out in two stages. First, to enhance the sampling of conformational space and remove potential crystal artifacts, we used accelerated MD (aMD) (20). Since the WT structure will serve as our reference point, we performed 1 ms of aMD simulation and 300 ns of conventional MD (cMD) for the WT, and 100 ns of aMD for the K118M/N proteins. We performed a principal component analysis (PCA) on the combined WT simulations to assess the large-scale dynamics of the protein and gain some sense of the extent of the conformational space. This analysis showed the presence of an equilibrium conformation of WT G-actin with a single equilibrium structure. The K118M/N mutants were projected onto this same conformational space, and the results indicated that their equilibrium structures were shifted slightly. Starting from these equilibrium structures, we subsequently performed 50 ns of cMD simulations for K118M and K118N to properly sample in the vicinity of the equilibrium conformations. All analysis was done using R (http://www.r-project.org) and Bio3D (21,22).

Phosphate release The production of free phosphate was measured using the EnzChek Phosphate Assay Kit (Molecular Probes). Briefly, the enzyme purine nucleoside phosphorylase was used to convert the substrate, 2-amino-6-mercapto-7methyl-purine riboside (MESG), to ribose 1-phosphate and 2-amino-6-mercapto-7-methylpurine by catalyzing the phosphorolytic cleavage of MESG.

Biophysical Journal 111, 323–332, July 26, 2016 325

Jepsen et al. The resulting product shifts the spectrophotometric peak absorbance from 330 nm for the substrate to 360 nm for the product. A UV-Vis spectrometer can then be used to measure the rate of formation of this compound, which is directly proportional to the amount of free phosphate in solution. To perform the phosphate-release assays, a fresh set of standards of known phosphate concentration (0, 2, 5, 8, 10, 15, and 20 mM) were created before every assay and used within 1 h of preparation. The A360 of each standard was measured and graphed against the known phosphate concentration to determine the correlation between the absorbance signal and the phosphate concentration. A linear fit of the data points was then performed, and the resulting slope of the line was used to convert A360 to mM Pi. The raw absorbance was then converted to mM Pi and graphed versus time. These experiments were performed at 25
C using a Hitachi U-2900 spectrophotometer. An initial baseline absorbance reading was acquired with a 5 mM G-actin sample, after which polymerization was induced by the addition of F-salts to a final concentration of 2 mM MgCl2 and 50 mM KCl. Absorbance was measured over ~25 min, with measurements taken every 5 s, until the rate at which the absorbance increased was linear and the actin had reached the treadmilling phase. The rates of phosphate release for elongation and steady-state phases were determined by fitting the data points from ~50–100 s and 800–1000 s, respectively, to a line. The slope of the resulting line was then determined to be the rate of phosphate production.

RESULTS Our previous observation that K118N actin polymerized approximately twice as fast as WT or K118M actin was puzzling because residue 118 lies on the surface of the actin monomer in subdomain 1, extending out into the surround-

ing solvent and away from the nearest actin-actin interface in the filament. This result, coupled with our observation that the K118M mutation caused severe growth defects, suggested that the two mutant residues might generate allosteric effects and propagate protein conformational changes, altering both the structure of the monomer and the protomer contacts in F-actin. Changes in the W79 and K118 helices In the actin monomer, the helix containing K118 is in close proximity to the helix containing tryptophans at positions 79 and 86. We find that whereas the local environments around W79 and W86 are very similar in the WT and K118M structures, the side chain of W86 is flipped in the K118N structure (Fig. 2). The flipped W86 side chain in K118N is a result of localized conformational changes due to the loss of interactions between K118 and W79. The aliphatic nature of the lysine results in the K118 side chain being packed against the ring of W79, with the terminal amino group making contact with the surrounding solvent. This interaction is lost upon mutation to asparagine, since it is both smaller and has less nonpolar surface area. Additionally, asparagine has a reduced helical propensity, weakening the K118 helix and allowing N115 to make new significant

FIGURE 2 Localized structural changes around the K118 and W79 helices. For the WT, K118M, and K118N structures, we show the average conformation of the H73 loop (orange), orientation of the W86 indole ring (yellow arrow), bends of the W79 and K118 helices, and hydrogen-bonding interactions. Helix bend angles are illustrated using the Bendix plugin in VMD (39). Amino acid distances are listed in Table S1. To see this figure in color, go online.

326 Biophysical Journal 111, 323–332, July 26, 2016

Effects of Actin Deafness Mutations

hydrogen bonds with W79, N111, and the backbone of V76 (Fig. 2; Table S1 in the Supporting Material). These new interactions shift the K118 helix up toward the W79 helix and push W79 back into the protein, crowding W86 and causing its side chain to flip. The shift of this helix also slightly shifts the bend of both the K118 and W79/W86 helices, becoming slightly straighter about K118 but more bent in the vicinity of W86. Since changes in the orientation and environment of tryptophan can affect its intrinsic fluorescence, we used fluorescence spectroscopy to measure the tryptophan fluorescence of the three different actins from 305 to 400 nm using an excitation wavelength of 300 nm. Typically, the lmax for tryptophan fluorescence occurs in WT actin at 328 nm, but we found that both mutations affected the tryptophan fluorescence spectrum (Fig. 3). The K118M mutation resulted in a quenching of the fluorescence intensity, but lmax remained unchanged (328 nm). This result is probably not due to an induced protein conformation change, but rather to the properties of the methionine itself, since the sulfur atom is known to cause quenching (23). In contrast, K118N actin had a higher maximum fluorescence intensity than the WT, and there was a distinct and repeatable red shift in lmax from 328 nm to 334 nm. Doyle et al. (24) and Turoverov and Kuznetsova (25) previously showed that the fluorescence of W79 is quenched by a nonradiative energy transfer to W86 in WT G-actin. Even though W79 has much greater solvent accessibility than any other tryptophan in WT and K118M G-actin structures, it does not contribute to overall fluorescence due to this quenching. Since the indole ring of W86 is flipped in the K118N structure (Fig. 2), W79 can now contribute more to the overall fluorescence spectrum, consistent with the red-shifted spectrum we observe.

Fluorescence (a.u.)

5e+07 4e+07

Structure WT K118M K118N

3e+07

Actin G F

2e+07 1e+07 0e+00 300

325

350

375

Wavelength (nm)

400

FIGURE 3 Intrinsic tryptophan fluorescence of G- and F-actin. Intrinsic tryptophan fluorescence was measured for both G- and F-actins to assess potential conformational changes in the protein. The above scans are representative of scans that were acquired in triplicate from three separate actin preparations. To see this figure in color, go online.

Red shifts can also indicate an increase in the solvent accessibility of one or more tryptophan residues (26). Indeed, if we look at the total combined solvent-accessible surface area of the tryptophans in our simulations, we find that the distributions are not normal (Gaussian), and we see median ˚ 2 for the WT and K118M, respecareas of 12.8 and 14.6 A tively (this does not include W79 due to quenching). However, in the case of K118N and the flipped W86 side chain, the total tryptophan area should now include W79, which in˚ 2 (Fig. S1). Consistent with creases the median area to 26.0 A this finding, an acrylamide quenching analysis based on the Stern-Volmer equation (see Materials and Methods) shows that the KSV values for the WT and K118M G-actins are almost identical, at 0.00227 and 0.00240, respectively. However, for K118N G-actin, the KSV is 0.00359, which is nearly twice as large, indicating a much greater accessibility to the acrylamide quencher (Fig. S2). Changes in the actin monomer The localized changes we see in the W79 and K118 helices lead to larger-scale changes in the protein. Using the results of our MD simulations, we examined the potential effects of the K118M/N mutations on the overall structure of G-actin. The two major conformational changes of the actin monomer (a propeller twist and a scissor-like motion of the nucleotide-binding cleft) correspond to relative changes in the four subdomains (Fig. 4). The propeller twist is characterized by the dihedral angle between the cores of subdomains 2-1-3-4. After an extensive simulation, we found that our WT yeast structure settled on an average twist of 12.9
(standard deviation (SD) of 2.8
). This conformation is slightly flatter than what is found in crystal structures of actin, which show a range of twists from 16
to 25
(27). The K118M mutant adopts a more twisted structure, 28.2
5 2.6
(SD), whereas the K118N conformation is significantly flatter, with a twist of 8.1
5 1.8
(SD). This twist angle of K118N is consistent with what is found in the F-actin model structures (28,29), and this likely contributes to the observation that K118N polymerizes much faster than either WT or K118M G-actin. Next, we sought to ascertain whether the effects of these two mutations might be due to a propagated conformational change in the actin. As a first step, we assessed the susceptibility of the mutant actins to different proteases whose cleavage sites were distant from the site of the mutations. Actin contains cleavage sites for trypsin, subtilisin, and chymotrypsin near the DNase I loop (D-loop) in actin subdomain 2, and some previously studied mutations alter the conformation of these sites to the point that susceptibility to cleavage by these proteases is affected (30,31). However, we observed no changes compared with WT actin when we subjected K118M and K118N actins to these proteases, suggesting that the mutations caused no significant conformational changes in this part of the protein (Fig. S3). This

Biophysical Journal 111, 323–332, July 26, 2016 327

Jepsen et al.

FIGURE 4 Structural changes in actin. Two metrics of actin subdomain arrangement are highlighted: the propeller twist (yellow), measured as the dihedral angle between the cores of subdomains 2 (red), 1 (orange), 3 (green), and 4 (blue), and the phosphate-clamp distance (black), measured as the distance between the Ca atoms of D157 and G15. The values listed in the table reflect the mean 5 SD as calculated from the simulations. To see this figure in color, go online.

observation is also consistent with our MD simulations. Although we observed changes in the arrangement of the four subdomains, we did not see any major structural changes that would likely affect proteolysis. We next attempted to detect structural changes via alterations in the CD spectra of the mutant proteins that, if they occurred, would indicate a mutation-based alteration in secondary structure. Using seven replicates over three different protein preparations, we found no significant effect of either mutation on the G-actin CD spectra between 190 and 240 nm (Fig. S4 A), although K118N actin did appear to have a slightly stronger signal at ~210 nm and 222 nm. To distinguish whether the difference was due to a concentration effect or the protein structure per se, we normalized the intensity of the K118N curve to that of WT actin and overlaid the two curves to observe potential differences in the spectrum shape. The spectra were virtually identical (Fig. S4 B), which reinforces the hypothesis that the protein secondary structures are not different. TABLE 1

Changes in F-actin We next wanted to determine how polymerization affected tryptophan fluorescence with the WT and K118M/N actins,

Experimentally Measured Parameters for Expressed Actins

lmax (nm) (n ¼ 6) WT K118M K118N

As an additional assessment of protein structure, we also used CD as a function of temperature to test for changes in protein stability brought about by the mutations. Although data from such a test were previously published (1,10), the discovery of cofilin in the initial K118N actin preparations (since removed) required us to repeat any assays that were performed with the contaminated samples. G-actin was heated from 25
C to 90
C and the resulting change in ellipticity was measured and fit to a two-state model (Fig. S5). T1/2 was determined as the temperature at which half the protein was folded, which for the WT, K118M, and K118N actins was ~64
C, 62
C, and 59
C, respectively (Table 1). Both the K118M and K118N mutations slightly but reproducibly reduced the protein stability, indicating that the mutations may cause localized structural effects.

327.8 5 0.12 327.5 5 0.08 334.0 5 0.16

G-actin KSV

F-actin KSV

0.00227 5 0.00001 0.00219 5 0.00003 0.00240 5 0.00002 0.00228 5 0.00001 0.00359 5 0.00002 0.00282 5 0.00003

t1/2 of Nucleotide Exchange (s) (n ¼ 6)

T1/2 (
C) (n ¼ 6)

37.1 5 0.41 41.8 5 0.86 39.4 5 1.1

63.7 5 0.08 61.2 5 0.33 59.1 5 0.20

Elongation Steady-State Pi Release Pi Release Rate (mM/s) (n ¼ 3) Rate (mM/s) (n ¼ 3) 0.0180 5 0.0006 0.0203 5 0.0014 0.0335 5 0.0018

0.0024 5 0.0001 0.0022 5 0.0001 0.0031 5 0.0003

From left to right: peak wavelength for tryptophan fluorescence, quenching constants for G- and F-actin, halftime for nucleotide exchange, and rate constants for Pi release for both elongation and steady-state growth. All errors listed are standard errors of the mean. For the KSV values, these are the standard errors resulting from linear regression.

328 Biophysical Journal 111, 323–332, July 26, 2016

Effects of Actin Deafness Mutations

as polymerization normally results in an ~15% decrease in intrinsic tryptophan fluorescence (32). Our previous results showed that the K118N mutation accelerated actin polymerization, but provided no substantial information about the nature of the final filament. We therefore asked whether this polymerization-dependent fluorescence change also occurred in the mutant actins. We found that the spectra of the WT and K118M actins were almost identical (Fig. 3), indicating that the chemical environments of the tryptophans after the conformational change from G- to F-actin were very similar for these two actins. In contrast, for K118N, polymerization caused a blue shift of the otherwise red-shifted G-actin fluorescence peak, returning the lmax to near that of the WT and K118M proteins. The polymerization-dependent decrease in fluorescence was almost to the same level as that seen with the other two actins. The tryptophan environment in K118N F-actin is thus very similar to that of the WT and K118M F-actins, suggesting that whatever conformational or environmental difference the asparagine mutation causes in G-actin, it reverts to almost normal in F-actin due to the constraints of the filament structure (Fig. S6). From these data, it appears that the K118N mutation mainly affects the G-actin conformation and the actin’s ability to polymerize, but the final filament structure of K118N F-actin is likely similar to that of the WT and K118M actins. A summary of the wavelength scan results is presented in Table 1. Acrylamide quenching in F-actin We repeated the fluorescence quenching assays with F-actin samples of the three actins (Fig. S7). Just as with G-actin, the KSV values for the WT and K118M actins were nearly identical, at 0.00220 and 0.00228, respectively. The KSV for K118N F-actin, however, was 0.00282, which is a significant (21%) decrease in quencher accessibility compared with K118N G-actin. These results are in agreement with our previous tryptophan fluorescence data, which indicated that the K118N tryptophans were in an environment more similar to the WT upon polymerization. However, the KSV was still 29% higher for K118N F-actin than for WT F-actin, and thus the environment of the tryptophans had not completely resumed a WT-like state. A summary of the results from the quenching assay is shown in Table 1. Nucleotide exchange Because structural effects in actin are often propagated from the outside of the protein to the nucleotide-binding cleft, and because the nucleotide dynamics of actin can influence polymerization, we measured the first-order rate of fluorescent etheno-ATP exchange from WT, K118M, and K118N G-actin in the presence of a large excess of ATP. The t1/2 values for WT and K118N were not statistically significant (~37.1 and 39.4 s, respectively; Table 1). The t1/2 for K118M

was slower (41.8 s), and this was statistically significant from the WT (p ¼ 0.002) and seemed to indicate slight effects of the mutations on the dynamics of the ATP-binding pocket. Although these shifts in t1/2 are relatively small, the trend is consistent with our simulation results. The phosphate clamp is a structural measure of how tightly the nucleotide is coordinated in the binding site, and is characterized by the distance between the Ca atoms of D157 and G15 (Fig. 4). We find phosphate clamp distances (5 SD) of ˚ and 5.8 5 0.4 A ˚ in WT and K118N G-actin, 6.4 5 0.5 A respectively, but the coordination is tighter in K118M ˚ ). If we calculate the Pearson correlation of (5.2 5 0.3 A the t1/2 values and phosphate-clamp distances, we get a significant negative correlation (0.99992, n ¼ 3, p ¼ 0.0078), suggesting that both mutants coordinate the nucleotide more tightly, resulting in longer exchange times. Phosphate release It is possible that the mutations might cause alterations in the dynamics of the filament that could be ascertained by measuring the release of Pi from the nucleotide-binding cleft. Therefore, we next measured the rate of free phosphate release during actin polymerization using a coupled enzymatic reaction method. Because there is a delay between ATP hydrolysis and release of the free phosphate in some actins, this assay does not provide results on the rate of ATP hydrolysis itself. Using cofilin-free G-actin, we reassessed the phosphate release rates (Fig. S8). We found that for all three actins, the rate of phosphate release closely follows the polymerization curve during the initial elongation phase of polymerization, suggesting that the rate of phosphate release is proportionate to the polymerization rate. However, as polymerization levels off to the treadmilling phase where there is no net polymerization, the rate of phosphate release becomes linear, according to the rate at which monomers cycle through the filament. Both WT and K118M show identical profiles, suggesting that the rate of phosphate release is unaffected by the K118M mutation. Although K118N actin shows a proportionate increase in the phosphate release rate during elongation, it shows a small but reproducible increase in rate during treadmilling, suggesting that the K118N filament is more dynamic than its WT counterpart. Our simulations did not probe phosphate release directly, but the looser phosphate clamp in K118N would also be consistent with faster Pi release. These results are also summarized in Table 1. DISCUSSION The fact that two very different mutations at residue 118 in actin lead to hereditary deafness provides a unique opportunity to gain insight into how this part of the actin structure relates to overall actin function. In the following discussion, we focus first on local and allosteric effects on the actin

Biophysical Journal 111, 323–332, July 26, 2016 329

Jepsen et al.

monomer, and then on the longer-range changes that occur in the actin filament. Finally, we discuss how these mutations may ultimately result in disease.

tide exchange rates, or susceptibility to limited protease digestion. Effects in F-actin

Effects on G-actin The altered intrinsic tryptophan fluorescence spectra and decreased thermal stability suggest that both mutants cause localized changes in the actin structure. Our results show that the mutation to asparagine causes changes within the K118 helix, allowing N115 to form new hydrogen bonds with W79, N111, and the backbone of V76, resulting in the helix being shifted up toward W79. The movement of the K118 helix forces W79 into the space previously occupied by W86, causing the W86 side chain to flip. These changes are consistent with the red-shifted tryptophan peak fluorescence observed for K118N actin and the increased susceptibility of K118N actin to acrylamide quenching. Since W86 has previously been shown to quench W79 by nonradiative energy transfer, the flipping of the W86 indole ring would increase the contribution of W79 to the observed fluorescence. Additionally, the polar group of the asparagine side chain would be closer to the tryptophan than is the lysine amino group, creating a more hydrophilic environment. In the case of K118M, there is evidence that methionine can interact with tryptophans to contribute to protein stability (33), and thus an interaction between M118 and W79 could compensate for the loss of an interaction between W79 and the methylene carbons of K118. Such an interaction would also be consistent with an altered tryptophan electronic environment that would explain the different fluorescence spectra of K118M. The fluorescence spectra of K118M showed no shift in peak fluorescence, but it did show a distinct quenching of this peak. This effect is likely due to the presence of the M118 sulfur, which has been shown to have a quenching effect on tryptophan fluorescence (23). The changes that we see within the K118 helix ultimately lead to changes in subdomain arrangement. The hydrogen bond between N115 and V76 in the K118N mutation results in a loss of secondary structure in the H73 loop (residues 70–78). As the H73 loop is part of a network of structural loops within actin that give structure to the nucleotide-binding cleft (34), the loss of secondary structure would alter the nucleotide site and should affect both the monomer twist and the D-loop structure. Indeed, we observed a significant flattening of the monomer in the case of K118N and further twisting of the K118M structure, although the molecular basis for the changes in K118M is not as obvious. Interestingly, even with substantial subdomain rearrangements and the changes in the H73 loop, we observed no other significant change in either mutant with regard to folding or overall secondary structure based on CD measurements, MD simulations, nucleo-

330 Biophysical Journal 111, 323–332, July 26, 2016

Although we see differences in monomer structure between the WT and the K118M/N mutants, it appears that these differences are lessened once actin is incorporated into the filament. Our fluorescence results show that putting the mutant monomer into the constraints of the filament structure significantly reverted the K118N protein to a more normal structure. Similarly, our measurement of phosphate release rates during polymerization showed that there does not appear to be a significant effect on nucleotide-binding dynamics as a result of the K118M/N mutations, since phosphate release closely reflected the overall rate of polymerization in all cases. The rate of phosphate release during the steady-state treadmilling phase, where monomer addition at one end of the filament is matched by monomer release at the other, was slightly faster for K118N than for WT actin, a feature that may result from the looser nucleotide coordination we observed in the K118N simulations. For yeast actin, there is no lag in Pi release subsequent to ATP hydrolysis (35). This finding, coupled with the fact that for all actins studied here, Pi release and polymerization occurred at basically the same rate, implies that a slightly increased rate of Pi release during treadmilling would give a faster rate of monomer addition and Pi release. Such a result would be consistent with our hypothesis that the N mutation causes an increase in conformational flexibility that allows the protein to more easily achieve a polymerization-competent state. Our polymerization studies showed that although the N and M mutations did not significantly alter the critical concentration for polymerization, the K118N mutation caused the actin to polymerize approximately two times faster than either the WT or K118M species. The maintenance of the critical concentration suggests that the increased polymerization rate of the N mutant results from an increased ease of spontaneous filament nucleation, and this is consistent with our simulation results showing that the monomer structure of K118N is very close to that of the F-actin protomer. The cytoskeletal dysfunction associated with both mutations, coupled with the polymerization data, suggests that the mutations caused improper regulation of filament formation and turnover, and that significant inherent polymerization defects were not the cause of the resulting altered cytoskeletal behavior. Our previous work indicated that the K118 helix is at the strand-strand interface of the filament and connects the binding sites for the Arp2/3 complex and formin with an interstrand ionic interaction that plays a role in filament stability and polymerization (5). Our current work, especially with the K118N mutant, provides key insight into how altering a residue at a regulatory protein-binding site can propagate a series of conformational changes to the actin

Effects of Actin Deafness Mutations

monomer in terms of propeller twist, H73 loop, and D-loop alteration. All of these factors are important determinants of polymerization, and our results show that the mutation shifts the actin monomer into a conformation that polymerizes faster and is closer to the F-actin protomer structure. In essence, the mutation could function to deregulate polymerization by converting the protein to a molecule that is more likely to undergo spontaneous polymerization.

Potential connection with disease A hallmark of proper cytoskeletal function is very tight regulation of actin polymerization and depolymerization. Patients with DFNA20/26 initially form a normally functioning auditory apparatus and gradually develop deafness as they age. The most widely accepted hypothesis regarding this phenotype is that the repair processes that maintain the structure over time, against various physical insults and normal protein turnover, fail with the mutant actins. This type of repair would represent a fine-tuning process, and if there is limited access to G-actin and the regulatory proteins that control it after the structure has been built, decreased regulation due to protein mutation could lead to the timedependent disappearance of hearing that is described. Our results indicate that the effects K118 mutations have on yeast actin may be even more pronounced in the context of the auditory hair cell. The vast majority of the DNFA20/ 26 mutations are in g-cytoplasmic actin. Our previous work (36) demonstrated that this actin isoform polymerizes more slowly than either yeast actin or b-cytoplasmic actin, the other actin isoform in the hair cell. Such an inherent bias would make spontaneous actin polymerization even less likely, underscoring the importance of a strict regulation of filament nucleation. Apart from regulation of the filament, higher-order regulation of filament bundles could also be affected. The bundling protein espin is an attractive candidate in this regard because it is the major bundling protein within stereocilia, and mutations in espin are also linked to autosomal-recessive hearing loss (37). Previous work showed how mutations that weaken the cross-linking activity of espin have significant effects on the viscoelastic properties of actin bundles (38), and there is a possibility that K118 mutations could result in similar defects. It is hoped that these issues will be addressed in future investigations.

SUPPORTING MATERIAL

formed and analyzed the experimental work. All authors reviewed the collective results and approved the final version of the manuscript.

ACKNOWLEDGMENTS We thank Karthikeyan Diraviyam for assistance with some of the molecular simulations. This work was supported by a grant from the National Institute on Deafness and Other Communication Disorders, National Institutes of Health (DC008803 to P.A.R.). L.J. received support from the Training Program in Bioinformatics through the National Institutes of Health (T32 GM070499). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

REFERENCES 1. Morı´n, M., K. E. Bryan, ., M. A. Moreno-Pelayo. 2009. In vivo and in vitro effects of two novel gamma-actin (ACTG1) mutations that cause DFNA20/26 hearing impairment. Hum. Mol. Genet. 18:3075– 3089. 2. Rendtorff, N. D., M. Zhu, ., L. Tranebjaerg. 2006. A novel missense mutation in ACTG1 causes dominant deafness in a Norwegian DFNA20/26 family, but ACTG1 mutations are not frequent among families with hereditary hearing impairment. Eur. J. Hum. Genet. 14:1097–1105. 3. Zhu, M., T. Yang, ., K. H. Friderici. 2003. Mutations in the gammaactin gene (ACTG1) are associated with dominant progressive deafness (DFNA20/26). Am. J. Hum. Genet. 73:1082–1091. 4. Kruth, K. A., and P. A. Rubenstein. 2012. Two deafness-causing (DFNA20/26) actin mutations affect Arp2/3-dependent actin regulation. J. Biol. Chem. 287:27217–27226. 5. Lee, C. Y., J. Lou, ., L. V. McIntire. 2013. Actin depolymerization under force is governed by lysine 113:glutamic acid 195-mediated catchslip bonds. Proc. Natl. Acad. Sci. USA. 110:5022–5027. 6. Goley, E. D., A. Rammohan, ., M. D. Welch. 2010. An actin-filamentbinding interface on the Arp2/3 complex is critical for nucleation and branch stability. Proc. Natl. Acad. Sci. USA. 107:8159–8164. 7. Pfaendtner, J., N. Volkmann, ., G. A. Voth. 2012. Key structural features of the actin filament Arp2/3 complex branch junction revealed by molecular simulation. J. Mol. Biol. 416:148–161. 8. Otomo, T., D. R. Tomchick, ., M. K. Rosen. 2005. Structural basis of actin filament nucleation and processive capping by a formin homology 2 domain. Nature. 433:488–494. 9. Thompson, M. E., E. G. Heimsath, ., F. J. Kull. 2013. FMNL3 FH2actin structure gives insight into formin-mediated actin nucleation and elongation. Nat. Struct. Mol. Biol. 20:111–118. 10. Bryan, K. E., K. K. Wen, ., P. A. Rubenstein. 2006. Effects of human deafness gamma-actin mutations (DFNA20/26) on actin function. J. Biol. Chem. 281:20129–20139. 11. Cook, R. K., W. T. Blake, and P. A. Rubenstein. 1992. Removal of the amino-terminal acidic residues of yeast actin. Studies in vitro and in vivo. J. Biol. Chem. 267:9430–9436.

Eight figures and one table are available at http://www.biophysj.org/ biophysj/supplemental/S0006-3495(16)30457-X.

12. Sorensen, B. R., L. A. Faga, ., M. A. Shea. 2002. An interdomain linker increases the thermostability and decreases the calcium affinity of the calmodulin N-domain. Biochemistry. 41:15–20.

AUTHOR CONTRIBUTIONS

13. Savitzky, A., and M. J. E. Golay. 1964. Smoothing and differentiation of data by simplified least squares procedures. Anal. Chem. 36:1627– 1639.

All authors were involved in the design of the study. L.J. and D.S. implemented and analyzed the simulation work, and K.A.K. and P.A.R. per-

14. Eftink, M. R., and C. A. Ghiron. 1981. Fluorescence quenching studies with proteins. Anal. Biochem. 114:199–227.

Biophysical Journal 111, 323–332, July 26, 2016 331

Jepsen et al. 15. Vorobiev, S., B. Strokopytov, ., S. C. Almo. 2003. The structure of nonvertebrate actin: implications for the ATP hydrolytic mechanism. Proc. Natl. Acad. Sci. USA. 100:5760–5765.

28. Fujii, T., A. H. Iwane, ., K. Namba. 2010. Direct visualization of secondary structures of F-actin by electron cryomicroscopy. Nature. 467:724–728.

16. Humphrey, W., A. Dalke, and K. Schulten. 1996. VMD: visual molecular dynamics. J. Mol. Graph. 14:33–38, 27–28.

29. Oda, T., M. Iwasa, ., A. Narita. 2009. The nature of the globular- to fibrous-actin transition. Nature. 457:441–445.

17. Phillips, J. C., R. Braun, ., K. Schulten. 2005. Scalable molecular dynamics with NAMD. J. Comput. Chem. 26:1781–1802.

30. Strzelecka-Go1aszewska, H., J. Moraczewska, ., M. Mossakowska. 1993. Localization of the tightly bound divalent-cation-dependent and nucleotide-dependent conformation changes in G-actin using limited proteolytic digestion. Eur. J. Biochem. 211:731–742.

18. Hornak, V., R. Abel, ., C. Simmerling. 2006. Comparison of multiple Amber force fields and development of improved protein backbone parameters. Proteins. 65:712–725. 19. Essmann, U., L. Perera, ., L. G. Pedersen. 1995. A smooth particle mesh Ewald method. J. Chem. Phys. 103:8577–8593. 20. Wang, Y., C. B. Harrison, ., J. A. McCammon. 2011. Implementation of accelerated molecular dynamics in NAMD. Comput. Sci. Discov. 4:4. 21. Grant, B. J., A. P. Rodrigues, ., L. S. Caves. 2006. Bio3d: an R package for the comparative analysis of protein structures. Bioinformatics. 22:2695–2696. 22. Skjærven, L., X. Q. Yao, ., B. J. Grant. 2014. Integrating protein structural dynamics and evolutionary analysis with Bio3D. BMC Bioinformatics. 15:399. 23. Yuan, T., A. M. Weljie, and H. J. Vogel. 1998. Tryptophan fluorescence quenching by methionine and selenomethionine residues of calmodulin: orientation of peptide and protein binding. Biochemistry. 37:3187–3195. 24. Doyle, T. C., J. E. Hansen, and E. Reisler. 2001. Tryptophan fluorescence of yeast actin resolved via conserved mutations. Biophys. J. 80:427–434. 25. Turoverov, K. K., and I. M. Kuznetsova. 2003. Intrinsic fluorescence of actin. J. Fluoresc. 13:41–57. 26. Vivian, J. T., and P. R. Callis. 2001. Mechanisms of tryptophan fluorescence shifts in proteins. Biophys. J. 80:2093–2109. 27. Oda, T., and Y. Mae´da. 2010. Multiple conformations of F-actin. Structure. 18:761–767.

332 Biophysical Journal 111, 323–332, July 26, 2016

31. Yao, X., S. Grade, ., P. A. Rubenstein. 1999. His(73), often methylated, is an important structural determinant for actin. A mutagenic analysis of HIS(73) of yeast actin. J. Biol. Chem. 274:37443–37449. 32. Bryan, K. E., and P. A. Rubenstein. 2005. An intermediate form of ADP-F-actin. J. Biol. Chem. 280:1696–1703. 33. Valley, C. C., A. Cembran, ., J. N. Sachs. 2012. The methionine-aromatic motif plays a unique role in stabilizing protein structure. J. Biol. Chem. 287:34979–34991. 34. Zheng, X., K. Diraviyam, and D. Sept. 2007. Nucleotide effects on the structure and dynamics of actin. Biophys. J. 93:1277–1283. 35. Yao, X., and P. A. Rubenstein. 2001. F-actin-like ATPase activity in a polymerization-defective mutant yeast actin (V266G/L267G). J. Biol. Chem. 276:25598–25604. 36. Bergeron, S. E., M. Zhu, ., P. A. Rubenstein. 2010. Ion-dependent polymerization differences between mammalian beta- and gammanonmuscle actin isoforms. J. Biol. Chem. 285:16087–16095. 37. Donaudy, F., L. Zheng, ., P. Gasparini. 2006. Espin gene (ESPN) mutations associated with autosomal dominant hearing loss cause defects in microvillar elongation or organisation. J. Med. Genet. 43:157–161. 38. Lieleg, O., K. M. Schmoller, ., A. R. Bausch. 2009. Structural and viscoelastic properties of actin networks formed by espin or pathologically relevant espin mutants. ChemPhysChem. 10:2813–2817. 39. Dahl, A. C., M. Chavent, and M. S. Sansom. 2012. Bendix: intuitive helix geometry analysis and abstraction. Bioinformatics. 28:2193– 2194.