Plant Villin Headpiece Domain Demonstrates a Novel Surface Charge

Feb 14, 2018 - Plants utilize multiple isoforms of villin, an F-actin regulating protein with an N-terminal gelsolin-like core and a distinct C-termin...
1 downloads 11 Views 3MB Size
Download PDF
Article Cite This: Biochemistry XXXX, XXX, XXX−XXX

pubs.acs.org/biochemistry

Plant Villin Headpiece Domain Demonstrates a Novel Surface Charge Pattern and High Affinity for F‑Actin Heather L. Miears,† David R. Gruber,† Nicholas M. Horvath,† John M. Antos,† Jeff Young,‡ Johann P. Sigurjonsson,† Maya L. Klem,† Erin A. Rosenkranz,† Mark Okon,§ C. James McKnight,∥ Liliya Vugmeyster,⊥ and Serge L. Smirnov*,† †

Department of Chemistry, Western Washington University, 516 High Street, Bellingham, Washington 98225-9150, United States Department of Biology, Western Washington University, 516 High Street, Bellingham, Washington 98225-9160, United States § Department of Biochemistry and Molecular Biology, Department of Chemistry, and Michael Smith Laboratories, University of British Columbia, Vancouver, British Columbia V6T 1Z3, Canada ∥ Department of Physiology and Biophysics, Boston University School of Medicine, 700 Albany Street, Boston, Massachusetts 02118-2526, United States ⊥ Department of Chemistry, University of Colorado at Denver, Denver, Colorado 80204, United States ‡

S Supporting Information *

ABSTRACT: Plants utilize multiple isoforms of villin, an F-actin regulating protein with an N-terminal gelsolin-like core and a distinct Cterminal headpiece domain. Unlike their vertebrate homologues, plant villins have a much longer linker polypeptide connecting the core and headpiece. Moreover, the linker−headpiece connection region in plant villins lacks sequence homology to the vertebrate villin sequences. It is unknown to what extent the plant villin headpiece structure and function resemble those of the well-studied vertebrate counterparts. Here we present the first solution NMR structure and backbone dynamics characterization of a headpiece from plants, villin isoform 4 from Arabidopsis thaliana. The villin 4 headpiece is a 63-residue domain (V4HP63) that adopts a typical headpiece fold with an aromatics core and a tryptophan-centered hydrophobic cap within its C-terminal subdomain. However, V4HP63 has a distinct N-terminal subdomain fold as well as a novel, high mobility loop due to the insertion of serine residue in the canonical sequence that follows the variable length loop in headpiece sequences. The domain binds actin filaments with micromolar affinity, like the vertebrate analogues. However, the V4HP63 surface charge pattern is novel and lacks certain features previously thought necessary for highaffinity F-actin binding. Utilizing the updated criteria for strong F-actin binding, we predict that the headpiece domains of all other villin isoforms in A. thaliana have high affinity for F-actin.

V

discovered in gelsolin and present in severin, supervillin, protovillin, and other proteins. In addition to the core, villin and many homologues (e.g., protovillin, supervillin, archvillin) contain a distinct C-terminal headpiece domain. The headpiece in turn consists of two distinct subdomains: a more thermodynamically stable C-terminal subdomain and less stable N-terminal one.6 The homologous headpiece is also present at the C-termini of certain proteins that lack the gelsolin core, as exemplified by dematin found in red blood cells of vertebrates.7 In vertebrate villins, the C-terminal headpiece domain is linked to the N-terminal gelsolin core through an unstructured, nonconserved 40-residue polypeptide (the “linker”), the longest interdomain linker in the protein.8 Analogous core-toheadpiece linkers of highly varying length, amino-acid

illins play key roles in cytoskeleton regulation in plants and vertebrates. Vertebrate villin is found in microvilli, which are a major component of brush border epithelial cells in the small intestine and kidney.1 Microvilli, cellular extensions of 1−5 μm scale of length, serve to enlarge the epithelial surface area. Structurally, microvilli are supported by dynamic actin filaments (F-actin) bundled together by villin and another protein fimbrin.2,3 Vertebrate villins and homologues have been extensively studied biochemically. On the other hand, understanding of the properties of principal domains and fragments within the more recently discovered plant villins is lacking.4 This paper presents the first structure and function study of a plant villin headpiece, a key domain within villin polypeptides (Figure 1). Villin belongs to the gelsolin and headpiece-containing families of actin regulating proteins2,5 (Supplementary Figure 1). A majority of villin homologues share the gelsolin-like “core”: a conserved multicopy repeat sequence initially © XXXX American Chemical Society

Received: August 31, 2017 Revised: February 7, 2018 Published: February 14, 2018 A

DOI: 10.1021/acs.biochem.7b00856 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

homology were identified: those from chicken villin, dematin, and supervillin (Supplementary Table 1). We also added to this comparison set protovillin18 headpiece the only nonvertebrate homologue with known F-actin binding properties.19 This collection of headpiece homologues represents domains showing high-affinity specific binding of F-actin (vertebrate villin and dematin), low affinity specific binding (protovillin), and no specific binding (supervillin). The key residues (basic, acidic, and aromatic) previously proposed as necessary and sufficient for F-actin binding6 are only partially present in the headpiece domain of villin 4 and other plant villin isoforms. Therefore, it is unclear from the sequence analysis whether villin 4 headpiece and other plant villin headpieces are expected to bind F-actin specifically and with significant affinity. In fact, the N-terminal boundary of villin 4 headpiece proved difficult to predict due to an insufficient level of local sequence identity. Therefore, for our investigation we chose to express and study a 76-residue C-terminal construct (atVHP76), 10− 15 residues longer than any previously characterized folded headpiece domain. In addition, a shorter, 60-residue, Cterminal construct (atVHP60) was expressed and tested with the intent to isolate completely the potential headpiece domain from the linker residues. Our solution NMR data for atVHP76 and atVHP60 showed that villin 4 contains a folded, 63-residue, C-terminal headpiece domain, which we refer to as V4HP63. This paper presents our characterization of the villin 4 headpiece for specific interaction with F-actin, its solution NMR structure, backbone dynamics, and molecular dynamics simulations of overall motions. Our structure/dynamics and function data for the plant villin C-terminus allows deeper evolutionary and functional understanding of the headpiece domains in general. To the best of our knowledge, this is the first investigation of an in vitro F-actin binding by any plant villin headpiece domain and first tertiary structure and dynamics of a villin headpiece from a nonvertebrate species.

Figure 1. C-Terminus of villin 4, the atVHP76 and atVHP60 fragment locations. (A) Numeration of residues as in REFSEQ accession NM_119162.5. The lower position enumeration in italics is as in atVHP76 (first six positions are occupied by the 6xHis tag). The coreto-headpiece linker consists of a larger, N-terminal basic (blue) and smaller, C-terminal acidic (red) regions. (B) The full amino acid sequence of atVHP76 is shown at the bottom. Numbers 1 and 82 indicate the N- and C-terminal residues respectively (H1 and F82) within the fragment. The residues comprising V4HP63 are underlined (positions 912−974 in villin 4). Highlighted in bold are (atVHP76 enumeration): W70, the hydrophobic cap residue; E45 (red) and K76 (blue), the salt bridge residues.

composition, and degree of sequence conservation are also present in most villin homologues (Supplementary Figure 1). Vertebrate villin nucleates and bundles actin filaments to support the microvilli under normal (low-calcium) conditions as well as severs the filaments to destabilize the microvilli under stress (signaled by high calcium levels).1,9,10 Vertebrate villins employ at least two actin binding sites: one site is localized on the headpiece, and the other actin site is associated with the core.6,11 An alternative model states that two villin molecules can dimerize exposing their headpiece domains for F-actin bundling.9 Possible other actin binding sites were proposed as well, e.g., a cryptic site on the linker.8 Uniquely, supervillin does not utilize its C-terminal headpiece or gelsolin core for interactions with F-actin but rather its large disordered Nterminus12,13 (Supplementary Figure 1). In plants, villins are distributed throughout diverse tissue types and organs and play more general and potentially diverse roles than they do in vertebrates. In general, plant villins operate under a broader range of ionic concentrations (e.g., Na+, K+)14 and assist in the upkeep of dramatically larger cellular structures (e.g., mm-long root hairs) than vertebrate villins do.4,15 Comparison of the plant and vertebrate villins yield unevenly distributed sequence homology within the core and headpiece domains with the linkers being largely unrelated (Supplementary Figure 1). The amino acid sequences of plant villins are predicted to include an unusually large, novel linker sequence of 100−200 residues (isoform-specific) (Figure 1).4 This paper targets villin 4, one of the five isoforms of the model plant Arabidopsis thaliana. Villin 4 is expressed in diverse tissues but has been studied most in root hairs. Plants lacking villin 4 function demonstrate disintegrated F-actin bundles and compromised root hairs.16,17 To understand the mechanism of action of villin 4, we set out to investigate its headpiece domain’s structure and function. Initially, we attempted predicting the villin 4 headpiece structure and F-actin binding capacity based on the properties of the previously characterized homologues. Three candidate vertebrate villin headpiece domains with known tertiary structure, F-actin binding properties, and highest sequence

■

MATERIALS AND METHODS Construction of the atVHP76 and atVHP60 Expression Vectors. The atVHP76 fragment was designed to include the 76 C-terminal residues of Arabidopsis thaliana villin-4 (REFSEQ accession NM_119162.5) fused with an N-terminal 6xHis tag. The atVHP60 fragment has a similar design but contains only the 60 residues on the villin-4 C-terminus. Both constructs were cloned into the pET-24a vector (Novagen) using commercial services (Genscript) with primers containing an NdeI site at the 5′ end and a HindIII site at the 3′ end of the TAG stop codon. Protein Expression for atVHP76 and atVHP60. The standard expression procedure was applied with the BL21(DE3) bacterial expression system transformed with the pET24a vectors and induced with IPTG for protein expression. Other details of the procedure are identical to those described previously.20 For expression of the isotopically labeled fragments, the cells were transferred to the minimal M9 media, equilibrated, and induced with IPTG as described before.21 Protein Purification for atVHP76 and atVHP60. After 4 h (in both rich media and minimal media), cells were harvested by low speed centrifugation (5000g for 30 min). Cell pellets were resuspended in lysis buffer (50 mM sodium phosphate, 300 mM NaCl, 10 mM Imidazole, pH 8.0) along with addition of lysozyme (50 μg/mL) and were left to shake at 4 °C for 30 min. Cells were then lysed via sonication in three 30-s intervals B

DOI: 10.1021/acs.biochem.7b00856 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

150000g for 1.5 h at 24 °C. Supernatant was removed and pellets were gently washed with F-actin buffer (79 mM TrisHCl, 50 mM KCl, 0.2 mM CaCl2, 2 mM MgCl2, 1 mM ATP, pH 8.0). Supernatants were analyzed for any significant presence of nonpolymerized actin with and without the headpiece constructs (none found). The pellets were always transparent and color-free, consistent with the absence of denatured protein. Pellets were resuspended with 7% acetic acid solution, and an aliquot of all samples, including controls, was removed for protein analysis by HPLC. A Luna 5 μ C18(2) 150 × 4.6 mm (Phenomenex) column connected to a Ultimate 3000 HPLC (Dionex) was pre-equilibrated with 10% solution of buffer A (0.1% TFA, ddH20), and samples were run from a 10−90% gradient of buffer A to B (0.1% TFA, 90% Acetonitrile). The resulting chromatograms were both visualized and analyzed with Chromeleon software. The Factin binding activity was also tested qualitatively through 15% SDS-PAGE analysis of the pull-down mixtures (not shown). We also produced a tagless version of atVHP76 to check for the formally possible effects of the 6xHis tag on F-actin binding. The tagless atVHP76 vector had a TEV protease cleavage site (ENLYFQS) cloned in between the 6xHis tag and the 76residue headpieace. The construct was expressed and purified by exactly the same steps as the two 6xHix-tagged fragments. Following that, standard TEV protease cleavage and purification with Ni-NTA resin resulted in separation of the cleaved tag and TEV protease (also 6xHis-tagged) from the tagless atVHP76. Prepared in this fashion, the tagless atVHP76 polypeptide has one non-native residue, serine, at its Nterminus. The identity of tagless atVHP76 fragment was confirmed with mass spectroscopy as described above for the tagged analogs. F-actin binding by tagless atVHP76 construct was tested exactly as for the tagged counterparts (described above in this section). NMR Data Collection and Processing. NMR samples contained 0.1−1.0 mM of atVHP76 and atVHP60 (unlabeled as well as labeled with 15N and 13C) and were suspended in NMR buffer (10% 2H2O, 20 mM PIPES, pH 6.8, 50 mM NaCl, 0.02% NaN3). No correction for the effect of 2H2O was made during the pH adjustment. 2D and 3D NMR spectra for atVHP76 and atVHP60 were acquired at 25 °C on the Bruker Avance III HD spectrometers operating at magnet strength values of 850 MHz, 600 and 500 MHz (1H frequency) and equipped with helium-cooled (TCI) or room-temperature (SmartProbe) probes (University of British Columbia, Vancouver, BC, Canada; Western Washington University, Bellingham, WA; Boston University School of Medicine, Boston, MA, respectively). All spectra were processed using NMRPipe22,23 and analyzed using NMRViewJ (One Moon Scientific). NMR Resonance Assignment for atVHP76. The backbone NMR resonance assignments (1H, 15N, and 13C) were produced by the standard approach24,25 from a combined use of the following 2D and 3D heteronuclear data sets: 15N HSQC, HNCACB, CBCA(CO)NH, HNCO, HN(CA)CO, HNCA, HCA(CO)N, HCAN, and HCA(CO)N. The nonaromatic side chain resonance (1H, 15N, and 13C) assignments were produced from 15N-HSQC, HBHA(CO)NH, H(CCO)NH, and CC(CO)NH. Aromatic side chain 1H assignments were derived from 13C-HSQC, (Hβ)Cβ(CγCδ)Hδ, (Hβ)Cβ(CγCδCε)Hε, (Hβ)Cβ(CγCarom)Harom, and 15N-TOCSY. All samples were referenced internally with water 1H resonance at 4.78 ppm relative to DSS at 25 °C. The peak height values for the “15N-

(Branson sonifer, 50% duty cycle). DNase I (1 μg/mL) was added to the lysed cells and incubated at room temperature for 30 min. Lysed cells solutions were then pelleted at 36000g for 30 min on a Sorvall Lynx 4000 centrifuge (Thermo Scientific) to pellet insoluble cell debris. The supernatant was then filtered through successive 5 μM and 0.45 μM filters. Nickelnitrilotriacetic acid (Ni-NTA) Superflow resin (QIAGEN) was equilibrated with lysis buffer, and then the lysate was added and incubated with the resin for 1 h at 4 °C with shaking. The solution was added to a gravity flow Ni-NTA column and after elution of the flow-through, the column was rinsed with four column volumes of wash buffer (50 mM sodium phosphate, 300 mM NaCl, 20 mM Imidazole, pH 8.0). (His)6-tagged atVHP76 or atVHP60 was then eluted with elution buffer (50 mM sodium phosphate, 300 mM NaCl, 250 mM imidazole, pH 8.0) in fractions. All fractions were analyzed for protein presence with 15% sodium dodecyl sulfate poly acrylamide gels (SDS-PAGE). For the final purification, elution fractions that contained (His)6-tagged atVHP76 or atVHP60 were loaded onto a HiLoadTM 16/60 Superdex 75 preparatory grade size exclusion chromatography column (120 mL, GE Healthcare). The column was pre-equilibrated with PIPES buffer (20 mM PIPES, 50 mM NaCl, pH 6.8) and attached to an Ä KTAprime Plus liquid chromatography system (GE Healthcare). The column was run at 1 mL/min, and the major peak was collected. The identity of the purified protein fragments was confirmed with mass spectroscopy (Advion expressionL CMS mass spectrometer interfaced with a Dionex Ultimate 3000 HPLC system). Analytical Size-Exclusion Chromatography. Analytical high performance size exclusion chromatography (HP-SEC) was performed using a Superdex 75 GL 10/300 column connected to an Ä KTA Purifier HPLC system. Data analysis was performed using the Unicorn software package (version 5.31). Samples of atVHP76 were injected at a range of concentrations 0.15−1.2 mM and eluted from the column at a flow rate of 0.5 mL/min in 20 mM PIPES and 50 mM NaCl, pH 6.8. A gel filtration standard spanning a molecular mass range of 1350−670000 Da (Bio-Rad) was prepared via manufacturer instructions and used for the construction of calibration curves (log of the molecular weight vs average distribution coefficient). F-Actin Binding Assay. An actin binding protein spindown assay Biochem Kit: Muscle Actin (Cytoskeleton Inc.) was employed with atVHP76 or atVHP60 as per the manufacturer’s instructions until the pellet dissolution stage. Lyophilized rabbit skeletal actin was resuspended to a concentration of 1 mg/mL with general actin buffer (5 mM Tris-HCl, 0.2 mM CaCl2, pH 8.0) and allowed to equilibrate at 4 °C for 30 min before the addition of actin polymerization buffer (100 mM Tris-HCl, 50 mM KCl, 2 mM MgCl2, 1 mM ATP, pH 7.5) for F-actin stock (21 μM) generation. The mixture then was equilibrated for 1 h at room temperature for F-actin polymerization. Lyophilized bovine serum albumin (BSA) and α-actinin were used as negative and positive controls for F-actin binding respectively and were resuspended to the stock concentrations of 3.4 and 1 mg/mL respectively. A clarification step for atVHP76 or atVHP60 and BSA was performed through centrifugation at 150000g on a Sorvall MX 150 micro-ultracentrifuge (Thermo Scientific) for 1 h at 4 °C. Headpiece samples at total concentration of 2, 5, 10, 20, 30, 40, 50, 60, 120 μM were then incubated with F-actin stock (actin concentration 16.8 μM after dilution) for 1 h at room temperature before centrifugation at C

DOI: 10.1021/acs.biochem.7b00856 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

relaxation rates of backbone amide nitrogens in 0.9 mM atVHP76 were measured using standard pulse sequences32 on a Bruker 500 MHz spectrometer with the Avance III console and equipped with a cryoprobe (Boston University School of Medicine, Boston, MA). The {1H−15N}-NOE ratio values were measured from pairs of spectra recorded with (NOE) and without (control) proton saturation during the recycle delay. The saturation period was 5 s in the NOE experiment, and the recycle delay was 5 s in the control experiment. Each NOE experiment was repeated three times. Relaxation delays were the following for the R1 experiment (0.01, 0.07, 0.14, 0.23, 0.35, 0.77, 1.00 s) and for the R2 experiment (0.004, 0.038, 0.078, 0.126, 0.184, 0.260 s). The recycle delay of 2.5 s and 1024 × 80 complex points were employed. Relaxation rates were analyzed via NMRViewJ.22,23 Estimates for the R1, R2 rates, and 1H−15N NOE ratio values were performed with tensor2 software.33 Free Molecular Dynamics Simulations (fMD). Molecular dynamics package AMBER was utilized at every stage of the simulation.34 System Initialization. The starting structure was the final solution NMR structure. Each structure was parametrized with the AMBER ffSB14 force field.34 Each system was neutralized with three Cl− ions using Joung and Cheatham parameters35 with CPPTRAJ36 from AmberTools.34 3058 explicit solvent molecules were added using the TIP3P water model37 in a truncated octahedral box. System Minimization and Equilibration. Initially, the protein atoms were held fixed. To eliminate van der Waals clashes, the solvent molecules and ions were subjected to 1000 steps of the steepest descent minimization. Subsequently, 1000 steps of conjugate gradient minimization, with a force constant of 25 (kcal/mol)/Å2 applied to the solute molecule. Next, the whole system was minimized with 1000 steps of the steepest descent followed by 1500 steps of conjugate gradient minimization without restraints. Heating was completed over 10 ps at constant volume from 100 to 300 K with weak positional restraints of 25 (kcal/mol)/Å2) applied to the protein molecule. A 2 fs time step was used and a weak coupling thermostat were used to control temperature, with a collision frequency of 1.0 ps−1. A version of the SHAKE algorithm38 was used to constrain bonds involving hydrogen atoms, with a tolerance of 0.002. A 10 Å cutoff was used for nonbonded interactions and particle mesh Ewald (PME) was used to handle long-range electrostatics.39 A five step minimization protocol followed heating in which the restraints applied to the protein were gradually reduced from 5.0 to 0.5 (kcal/mol)/Å2. At each step, minimization was carried out with 1000 steps of the steepest descent followed by 1500 steps of conjugate gradient minimization. A Berendsen coupling constant of 0.2 ps was used. Finally, 100 ps of fMD was run at 300 K with no restraints and constant pressure to relax the density of water. Production Molecular Dynamics. Production fMD simulations were run for 1.1 μs using graphics processing code (GPU) code with the PMEMD.cuda implementation of SANDER from Amber14 on an NVIDIA GTX 970 GPU. Simulations were held at constant pressure (1 atm) periodic boundaries and 300 K using Berendsen coupling constants of 5.0 ps. Long range interactions were calculated with PME. A 2 fs integration time step was used. Simulation coordinates were recorded every 1 ps. Trajectories analysis was performed with CPPTRAJ, where the structural stability of the simulation was examined in terms of RMSD vs simulation time. RMSD figures

HSQC intensity versus concentration” studies were estimated via 2D line-shape modeling by nlinLS facility of NMRPipe package.26 Solution NMR Structure Calculation. The following experiments were collected for distance restraints: 3D 15Nedited NOESY (mixing time 250 ms), 3D 13C-edited NOESY aliphatic (mixing time 100 ms), and 3D 13C-edited NOESY aromatic (mixing time 100 ms). Peak visualization and volume measurements were performed with NMRViewJ.22,23 The solution structure of V4HP63 was determined using CYANA 2.127 based on the NMR-based distance and dihedral angle restraints. Distance restraints were derived from heteronuclear 3D and proton 2D NOESY data recorded in 90% H2O - 10% 2 H2O. The dihedral angle restraints were produced prior to the CYANA simulations with PREDITOR28 utilizing the backbone chemical shift values of 1H, 15N, 13Cα, 13CO, and 12Cβ. Initially, CYANA simulations utilized only the 3D NOESY (15N NOESY, 13C NOESY aliphatic, and 13C NOESY aromatic) sets to assign automatically during the iterative simulations. Each simulation was conducted according to a standard CYANA protocol (noeassign macro) iteratively determining the solution structure while concurrently assigning the maximally high number of NOE cross-peaks. The conversion from the NOE peak volumes to the interproton distance restraints was performed via the default automated mechanism in CYANA software. The simulations started with 1000 test conformers and employed 10 000 steps of torsion angle dynamic steps. The 10 conformers with the lowest penalty (NOE/dihedral violations) function were retained for analysis. Every simulation was performed with more than one seed value for the random number generation, and the common assignments from the simulations were used as an assigned set of NOEs for the next simulations. After no more of the 3D NOESY peaks were assigned, the peaks/volumes from the 2D 1 H NOESY were added and assigned by the same iterative process during CYANA runs, while all the previously obtained 3D NOE assignments were retained. Every 2D or 3D NOE peak generated distance restraints violated by >0.5 Å at the end of a run was manually unassigned for the next simulation after investigation of the peak validity. The final ensemble of 10 structures with the lowest penalty energy were submitted to the PDB (entry 5VNT) and BMRB (entry 30289) repositories. Structure Representation and Modeling. The structure models for presentation were generated with Chimera,29 and solvent accessibility area values were calculated with DSSP software.30 Homology modeling of the headpiece domains in plant villins 1, 2, 3, and 5 was performed via Phyre2 online server.31 Solution NMR Investigation of atVHP76 Sensitivity to Ions. A 1.0 mM sample of 15N-atVHP76 was prepared by standard procedure, and buffer exchanged through an Amicon Ultra-15 centrifugal filter device with a 3 kDa mw cut off (Millipore) to lower the NaCl concentration from 50 mM to 145 μM. A 15N-HSQC spectrum was taken at each of the following concentrations of NaCl: 145 μM, 3.6, 50, 91, 100, 200, and 300 mM. All spectra were analyzed by overlaying the data sets to determine if any peaks show changes. The sample was buffer exchanged again via a filter device to the standard 50 mM NaCl buffer and then was equilibrated with first 100 mM then 200 mM KCl. A 15N-HSQC spectrum was taken after each equilibration step and analyzed by overlaying the data sets. Solution NMR Dynamics Measurements. Longitudinal, R1, transverse, R2, and heteronuclear NOE 1H−15N, σNH, D

DOI: 10.1021/acs.biochem.7b00856 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

atVHP76 and was thus named V4HP63. Residue numbering within V4HP63 was maintained as in atVHP76 (Figure 1). The solution NMR structure of V4HP63, the 63-residue Cterminal, the structured region of atVHP76 (Figure 3), was

were generated, and running averages were calculated using the XMGRACE program (http://plasma-gate.weizmann.ac.il/ Grace/). Principal component analysis was carried out using CPPTRAJ. Only Cα backbone atoms were considered in the analysis. Global rotational and translational movements were removed by subjecting each frame of the trajectory to a RMS-fit against the overall average coordinates. Porcupine plots were generated using the normal mode wizard plugin for VMD.40,41

■

RESULTS Sample Design. Two C-terminal fragments from A. thaliana villin 4 were expressed as recombinant protein fragments and characterized in vitro for structure and function: atVHP76 (76 residues) and atVHP60 (60 residues) (Figure 1). The C-terminal sequence corresponding to the folded, 63residue headpiece V4HP63 was deposited to PDB/BMRB repositories (Figure 1). The same 63-residue sequence of V4HP63 was used for the in silico molecular dynamics simulations. Oligomeric State of atVHP76: Monomer. The atVHP76 fragment was expressed in E. coli with an N-terminal 6xHis tag and purified by affinity and size-exclusion chromatography. The His tag was not cleaved. High performance size-exclusion chromatography (HPLC) data indicate that atVHP76 is a monomer at ambient temperature within the range of concentrations tested (0.15, 0.30, 0.60, and 1.2 mM) (Supplementary Figure 2). In parallel, the intensities of the 15 N-HSQC spectral lines show a dependence on the sample concentration (0.1, 0.3, 0.9 mM) and number of scans (324, 36, 4 respectively) consistent with the monomeric state (Supplementary Figure 3). No changes in the 15N-HSQC spectra (e.g., peak appearance/disappearance or peak shifting) were observed between the three concentrations tested. These data are consistent with a monomeric construct that does not aggregate at concentration at or under 1.0 mM. The C-terminal 63-Residue Stretch of Villin 4 (V4HP63) Shows a Canonical “Villin Headpiece” Fold. The 15N-HSQC spectrum of atVHP76 contains a set of welldispersed cross-peaks characteristic of a mostly folded polypeptide (Figure 2). NMR resonance assignment and

Figure 3. (A) The 10-structure ensemble of solution NMR structures of V4HP63; The centerpiece of the hydrophobic cap (W70) and the side chains of the aromatic core (F53, F57, and F64) of the C-terminal subdomain are shown in “sticks”, (B) Secondary structure and subdomains: N-terminal subdomain (cyan), C-terminal subdomain (orange); the salt bridge residues, E45 and K76, are shown in “sticks”; the helices are labeled with underlined letters: A, B, C and D; the structure is flipped horizontally vs panel A by ∼180°; (C, D, E) the hydrophobic core (opaque white CPK) and the loops (colored backbone) of V4HP63, VHP67, and DHP68 respectively. The hydrophobic core residues were those least exposed to the solvent with combined exposure of 9.0° average RMSD from the mean for the 10 best models all backbone atoms (residues 23−82) all heavy atoms (residues 23−82)

625 155 156 174 140 10 215 59 0 10 0 0.18 Å 0.67 Å

W70 side chain. Similarly to the chicken villin headpiece (HP67), which is stabilized by an intersubdomain salt bridge between E39 and K76,6,42 a buried salt bridge, E45-K76, exists between the N- and C- terminal subdomains of V4HP63 (Figure 3B). As in HP67, the distance between E45 Oε and K76 Hζ atoms ranges within 2.0−3.5 Å for 8 models out of our 10 best structures. The other two models show an alternative E45 side chain orientation, with the E45 Oε and K76 Hζ distance ranging within 4.1−5.5 Å. The solvent accessibility of the salt bridge residues as measured by DSSP software30 equals 6 Å2, significantly less than 21 Å2 found by the same software in HP67. As in HP67 and DHP68,6,43 the hydrophobic core of V4HP63 involves residues from both the N- and C- terminal subdomains (Figure 3C−E). These residues were least exposed to the solvent which together account for 0.58) and disordered sequence for residues 10−19 (