Single Molecule Force Spectroscopy and ... - ACS Publications

Sep 26, 2017 - Single Molecule Force Spectroscopy and Molecular Dynamics Simulations as a Combined Platform for Probing Protein Face-Specific Binding ...
0 downloads 18 Views 3MB Size
Article pubs.acs.org/Langmuir

Single Molecule Force Spectroscopy and Molecular Dynamics Simulations as a Combined Platform for Probing Protein FaceSpecific Binding Kartik Srinivasan,†,‡ Suvrajit Banerjee,†,‡ Siddharth Parimal,†,‡ Lars Sejergaard,†,‡ Ronen Berkovich,§ Blanca Barquera,‡,∥ Shekhar Garde,†,‡ and Steven M. Cramer*,†,‡ †

Howard P. Isermann Department of Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York 12180, United States ‡ Center for Biotechnology and Interdisciplinary Studies, Rensselaer Polytechnic Institute, Troy, New York 12180, United States § Department of Chemical Engineering, Ben-Gurion University, Beer-Sheva 84105, Israel ∥ Department of Biology, Rensselaer Polytechnic Institute, Troy, New York 12180, United States S Supporting Information *

ABSTRACT: Biomolecular interactions frequently occur in orientation-specific manner. For example, prior nuclear magnetic resonance spectroscopy experiments in our lab have suggested the presence of a group of strongly binding residues on a particular face of the protein ubiquitin for interactions with Capto MMC multimodal ligands (“Capto” ligands) (Srinivasan, K.; et al. Langmuir 2014, 30 (44), 13205−13216). We present a clear confirmation of those studies by performing single-molecule force spectroscopy (SMFS) measurements of unbinding complemented with molecular dynamics (MD) calculations of the adsorption free energy of ubiquitin in two distinct orientations with selfassembled monolayers (SAMs) functionalized with “Capto” ligands. These orientations were maintained in the SMFS experiments by tethering ubiquitin mutants to SAM surfaces through strategically located cysteines, thus exposing the desired faces of the protein. Analogous orientations were maintained in MD simulations using suitable constraining methods. Remarkably, despite differences between the finer details of experimental and simulation methodologies, they confirm a clear preference for the previously hypothesized binding face of ubiquitin. Furthermore, MD simulations provided significant insights into the mechanism of protein binding onto this multimodal surface. Because SMFS and MD simulations both directly probe protein−surface interactions, this work establishes a key link between experiments and simulations at molecular scale through the determination of protein face-specific binding energetics. Our approach may have direct applications in biophysical systems where face- or orientation-specific interactions are important, such as biomaterials, sensors, and biomanufacturing.



INTRODUCTION

changes of proteins adsorbed onto various surfaces using molecular dynamics (MD) simulations.18−21 Recently, the combination of SMFS and MD has been used to measure the energetics of peptides binding to lipid bilayers22 and hydroxyapatite23 surfaces. The presence of a wide range of chemical functionalities on protein surfaces often leads to specific surface patches or “faces” of a particular protein being highly favored for adsorption onto different surfaces.24−26 Here we focus on studying face-specific interactions of the protein ubiquitin with a surface presenting multimodal (MM) chromatographic ligands. MM ligands are capable of highly selective separations of proteins by offering

The study of protein−surface interactions is of cardinal interest in the fields of bioprocessing,1,2 biomaterials,3 and drug delivery.4 In these applications, proteins adsorb onto various surfaces via water-mediated interactions. Such protein−surface interactions have been investigated using a range of experimental techniques including solid-state nuclear magnetic resonance (NMR),5,6 1H NMR in concert with circular dichroism (CD),7 and hydrogen−deuterium exchange mass spectroscopy.8,9 In addition, single-molecule force spectroscopy (SMFS) has been used by several investigators to examine systems that include specific interactions between immobilized biomolecules.10−17 Furthermore, the exponential rise in computational power has made it possible to gain fundamental insights into the nature of orientational and conformational © XXXX American Chemical Society

Received: August 24, 2017 Published: September 26, 2017 A

DOI: 10.1021/acs.langmuir.7b03011 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

detector controlled by a Millennium chromatography software manager. PCR-Based Site-Directed Mutagenesis. Ubiquitin mutants (S20C and A46C) were generated using the wild-type ubiquitin gene (cloned into the pET 15b vector) as a template. To create the cysteine point mutations on ubiquitin, primers were ordered from Invitrogen (Carlsbad, CA). Site-directed mutagenesis was carried out using the QuikChange Lightning Site-Directed Mutagenesis Kit. The final mutagenesis product was used to transform Escherichia coli XL10-Gold ultracompetent cells, which were plated onto LB (Luria− Bertani, Miller) agar plates containing ampicillin (100 μg/mL) and incubated at 37 °C. Five clones were isolated and grown in liquid LB medium containing ampicillin (100 μg/mL). Plasmid was isolated using a Promega Wizard Plus SV Miniprep kit. The plasmid DNA was sequenced to confirm the introduction of the desired mutation. E. coli BL21(DE3) cells were then transformed with the mutated plasmid. Expression and Purification of Cysteine Mutants of Ubiquitin. The two ubiquitin mutants, UBQ S20C and UBQ A46C, were expressed using the BL21(DE3) strain of E. coli in LB media and purified using methods similar to those previously described for the wild-type protein.45 The cysteine mutants had a high propensity to form dimers through the formation of disulfide linkages. Therefore, the only differences in the purification methods were that the cationexchange chromatography and size-exclusion chromatography steps were carried out in the presence of 10 mM β-mercaptoethanol as a reducing agent. All other expression and purification steps remained unaltered. The final protein concentration was determined by spectrophotometric analysis at 280 nm with a molar extinction coefficient of 1490 M−1 cm−1.46 Preparation of MM Ligand-Functionalized AFM Tips. Gold-coated AFM tips were rinsed with Milli-Q water followed by ethanol and then immersed into an ethanol solution containing the thiol-terminated “Capto ligand” at a concentration of 1 mM for 24 h. The “Capto ligand” linker was synthesized and purified according to previously described procedures.44 Upon removal from solution, the tips were rinsed with ethanol, followed by Milli-Q water, dried with a gentle stream of nitrogen, and immediately transferred to the AFM instrument for force measurements. Preparation of Immobilized Ubiquitin Surfaces. Gold-coated coverslips were washed with Milli-Q water followed by ethanol and then dried with a stream of nitrogen. The substrates were then treated with UV/ozone for 30 min. The surfaces were then treated with a 5:1:1 mixture of Milli-Q water, ammonia (25% (v/v)), and hydrogen peroxide (30% (v/v)) for 10 min. The surfaces were then rinsed thoroughly with Milli-Q water, followed by ethanol. The surfaces were then dried under a stream of nitrogen and immersed in an ethanol solution containing a mixture of thiol-terminated linkers, EG6N and EG6OH, at a total concentration of 1 mM in compositions of 0.03 or 0% EG6N (mole basis) for 24 h. The substrates were then rinsed thoroughly with ethanol and Milli-Q water and dried with a stream of nitrogen. The surfaces were then treated with the SSMCC hetero bifunctional cross-linker (1 mg/mL solution in PBS buffer, pH 7.4) for 1 h to make them thiol reactive. The surfaces were then rinsed with Milli-Q water and ethanol and dried with nitrogen. Finally, the maleimide-activated surfaces were reacted with either UBQ S20C or UBQ A46C (150 μM solution in PBS buffer, pH 7.4) for 3 h. The surfaces were then washed sequentially with PBS buffer (pH 7.4), Milli-Q water, 10 mM acetate buffer (pH 5.0) containing 0.02 (w/w)% sodium azide, and Milli-Q water. The surfaces were then stored in 10 mM acetate buffer (pH 5.0) containing 0.02 (w/w)% sodium azide until the force measurements were performed. X-ray Photoelectron Spectroscopy. X-ray photoelectron spectroscopy (XPS) was employed to determine the atomic composition of the immobilized protein and MM ligand-functionalized surfaces. A PHI 5400 XPS instrument equipped with a Mg Kα probe beam was used to obtain the XPS spectra. As shown in Table S1, the major peaks of interest were C (1s), O (1s), N (1s), S (2p), and Au (4f). Spectra were obtained over a surface area of 1 μm2. Survey scans with a pass energy of 89.45 eV were first performed to identify elements present on the

different types of interactions (e.g., electrostatic, hydrophobic, and dispersion) within a single chromatography system.27−33 This unique ability has provided an important new dimension of separations that is having a profound impact on biomanufacturing. To gain fundamental insights into the complex mechanism of binding of MM ligands to protein surfaces, our lab has previously reported on a range of experimental and theoretical studies.34−38 Recently, we examined the binding of the protein ubiquitin to MM ligands immobilized on self-assembled monolayers (SAMs)39−43 coated on gold nanoparticles44 by employing isothermal titration calorimetry (ITC) and NMR spectroscopy. Importantly, the NMR experiments revealed a “preferred” binding face of ubiquitin for the MM surface. Here we present a combination of SMFS and molecular simulation studies to probe the energetics of these protein surface interactions and to obtain molecular level insights into the adsorption process. For the SMFS experiments, the protein was strategically immobilized on to SAM surfaces using appropriate cysteine mutants to either expose or sterically shield the “preferred” face. For the MD simulations, the umbrella sampling method was employed using suitable constraining methods to enable a comparison with the SMFS results. Our results demonstrate that a combination of SMFS and MD simulations presents a approach to determine facespecific binding energetics and to obtain molecular level insights into protein binding phenomena.



MATERIALS AND METHODS

Single-Molecule Force Spectroscopy. Materials. Sephadex G50 Superfine media was packed into a BioRad glass chromatography column (5 cm × 16 cm). An SP Sepharose FF HiLoad 16/10 column was obtained from GE Healthcare. A Bondapak C18 Prep column (7.8 mm × 300 mm) was obtained from Waters. Hexaethylene glycol thiols terminated with N-hydroxysuccinimide ester or hydroxyl (EG6OH) or amine groups (EG6N) were purchased from Prochimia (Poland). Acetonitrile was purchased from Acros Organics. Ethanol (anhydrous, 200 proof) was purchased from EMD Millipore. Iron(III) chloride, ammonium chloride, potassium phosphate, potassium sulfate, calcium chloride, tris (hydroxymethyl) aminomethane, hydrochloric acid, protease cocktail inhibitor, deoxyribonuclease (DNase), sodium chloride, sodium azide, acetic acid, sodium acetate, β-mercaptoethanol, hydrogen peroxide, and sodium hydroxide were purchased from Sigma-Aldrich. Magnesium chloride (hexahydrate) was purchased from Mallinckrodt Baker. Ampicillin and IPTG were obtained from Gold Biotechnology. Sulfosuccinimidyl-4-(N-maleimidomethyl) cyclohexane-1-carboxylate (SSMCC) was purchased from Fisher Scientific. Immobilized tris (2-carboxyethyl) phosphine disulfide (TCEP) disulfide reducing gel was purchased from Fisher Scientific. Centriprep centrifugal filter devices were purchased from Millipore. N-benzoyl lysine was purchased from Chem-Impex International. The atomic force microscope (AFM) tips employed in this work (gold-coated tips, triangular shaped, nominal spring constant of 0.06 N/m, and radius of curvature of 30 nm) were purchased from Bruker. Gold-coated coverslips (coated with a 2 nm layer of titanium and a 10 nm layer of gold) were purchased from Platypus Technologies. A QuikChange Lightning Site-Directed Mutagenesis Kit was obtained from Stratagene (Agilent Technologies, Santa Clara, CA). A Wizard Plus SV Miniprep kit was purchased from Promega (Madison, WI). Equipment. Purification of ubiquitin mutants was performed using an Ä kta Explorer 100 apparatuys equipped with Unicorn Software 3.1 from GE Healthcare. Reverse-phase chromatography was performed using a Waters high-performance liquid chromatography system consisting of a 600 multisolvent delivery system, a 717 Waters Intelligent Sample Processor auto injector, and a 996 photodiode array B

DOI: 10.1021/acs.langmuir.7b03011 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

isobaric ensemble at a temperature of 298 K and a pressure of 1 atm using the Nosé−Hoover thermostat61,62 (time constant for coupling was 0.5 ps) and Parrinello−Rahman barostat63 (time constant was 1.0 ps), respectively. An anisotropic barostat was employed to allow independent relaxation in the z direction. The Particle Mesh Ewald method64 was used to calculate electrostatic interactions. Real space cutoff and cutoff for Lennard-Jones interactions were both taken to be 1 nm. A time step of 2 fs was used. Configurations were saved every 1 ps for analysis. MD Simulation of SAM, Water, and Ions Only. An MD simulation was performed using systems that contained the “Capto” SAM surface, water, and counterions to obtain density profiles of water, counterions, and SAM heavy atoms. This system was first equilibrated at the temperature and pressure mentioned above for 4 ns, followed by a production run of 8 ns. Four nanoseconds was chosen as equilibration time because water and ion density profiles remained unchanged after this period of time had elapsed. These density profiles are shown in Figure S5B. After evaluation of water density profile, the z* axis was defined that originates at the position where the value of normalized density of water was equal to half of its bulk value and extends into the bulk (see Figure S5B). Subsequent PMF profiles are reported with respect to z* = 0 as origin. Four nanoseconds was also chosen to be the equilibration time for each window simulation in the ubiquitin“Capto” SAM umbrella sampling simulations discussed later. Umbrella Sampling Simulations of Two Faces of Ubiquitin Binding to “Capto” and −OH SAMs. Two sets of umbrella sampling simulations were performed at the aforementioned temperature and pressure to compute the protein-“Capto” SAM potential of mean force (PMF) for each face of the protein (“preferred” and “opposite” faces). Figure S5C shows the starting structure for a window simulation employed in ubiquitin-“Capto” SAM umbrella sampling. To compute the potential of mean force for each face of the protein, about 30 window simulations were performed with a harmonic potential being applied between the protein and the SAM in each of these simulations; that is, the perpendicular distance between the plane formed by the ninth carbon atoms in all SAM strands and the protein ubiquitin was constrained. Details of parameters used (window spacing and force constants) for biasing potentials in umbrella sampling simulations are given in Table S2. Three C−α atoms on the protein face opposite to that facing the SAM were position-restrained in the x−y plane in each set of umbrella sampling simulations. This prevented the protein from changing its orientation with respect to the SAM, thus maintaining a desired face exposed to the SAM throughout these simulations. The PMF was calculated using the weighted histogram analysis method (WHAM).65 In each window, the system was equilibrated for 4 ns, followed by 8 ns of production run. Error analysis was performed using a Bayesian bootstrapping method.66 A similar simulation and analysis procedure was followed to perform two more sets of umbrella sampling simulations to evaluate the potential of mean force between each face of the protein and the purely hydrophilic −OH surface.

surface, followed by acquisition of element-specific spectra with a pass energy of 1 eV. Force Measurements. Force measurements were performed using an Asylum MFP3D Atomic Force Microscope (Asylum Research, Santa Barbara, CA). Force measurements between the functionalized tips and immobilized ubiquitin substrates were performed in the presence of 10 mM acetate buffer (pH 5.0, 0.02 (w/w)% sodium azide). To account for any surface heterogeneity, a routine was employed to randomly sample points across the substrate surface to generate a statistically significant measurement. A constant approach velocity of 1000 nm s−1 was used for every retract velocity studied. A relative force trigger of 200 pN was used on all the force−distance cycles to limit probe and surface damage upon contact. The functionalized tip was allowed to dwell for 0.9 s before pulling away. Cantilever spring constants were calibrated using the thermal noise method and ranged from 15 to 54 pN/nm.47,48 Force−distance curves were collected and analyzed using IGOR PRO 5.0 (WaveMetrics, Lake Oswego, OR). Finally, MATLAB was employed to evaluate distribution functions of minimum force (protein−surface detachment force) from the individual force−distance curves by employing a set of custom-written procedures. MD Simulations. Simulation Details and Parameterization. The length and breadth of the periodic cuboidal simulation box (that is, dimensions of box in x and y directions, respectively) were 7 and 6 nm, respectively. The starting height of the box (that is, dimension of box in the z direction) was 15 nm, which reduced slightly upon equilibration as the barostat employed here allowed independent relaxation in this direction only. The SAMs were constructed as described in refs 49 and 50. As described in those references, each SAM contains two leaflets, each containing 196 C10 molecules. Two C10 strands are attached to either side of a base sulfur atom and point roughly in opposite directions (see Figure S5A). The sulfur atoms were position-restrained to locations corresponding to those on gold 111 lattice51 using a harmonic constraint with a sufficiently large force constant (40 000 kJ/mol.nm2), thus producing a double SAM slab on the x−y plane in the 3D periodic system (see Figure S5A). The alkane tails of the SAM strands were represented using the OPLS united atom representation.52 Two types of head groups were used: the Capto MMC ligand headgroup and hydroxyl headgroup. Parameters for Capto MMC ligand were assigned using parameters of constituent groups from the all-atom Assisted Model Building with Energy Refinement (AMBER) Parm-94 force field.53 This type of parametrization has been done previously in our group36,54 and was shown to produce agreement with NMR spectroscopy experiments in terms of identifying binding regions of these ligands dispersed freely in solution to proteins. The hydroxyl head groups were modeled by using the all-atom AMBER Parm-94 force field. Two types of surfaces were constructed: (i) A “Capto” SAM surface, comprising 49 Capto MMC and 147 hydroxyl head groups on each SAM slab, was constructed. The Capto MMC ligands were evenly distributed on the surface, producing a surface density of 1.17 ligands/nm2. The hydroxyl head groups provided a hydrophilic background for presenting the Capto head groups. (ii) To compare the ubiquitin-“Capto” SAM affinities of the two protein faces to a control system, a homogeneous hydrophilic surface was also constructed. This surface comprised 196 −OH head groups on each SAM slab. Water molecules near such −OH surfaces were previously shown to exhibit bulk-like behavior,55 and hence these surfaces were ideal for serving as a comparison with the “Capto” SAM surface, which had specialized modes of water-mediated interaction. The protein ubiquitin (PDB ID: 1D3Z) was modeled using Amber94 force-field.53 PROPKA56 was used to assign charges to the protein at pH 5.0. All ionizable residues were charged at this pH, giving rise to a net +1 charge to the protein. For the protein Capto SAM system, 97 Na+ counterions were added to the system to maintain charge neutrality, whereas, for the protein −OH SAM system, one Cl− ion was required to neutralize the system. Water molecules were represented using the extended simple point charge (SPC/E) model.57 All-atom molecular dynamics simulations were performed using GROMACS.58−60 All simulations were conducted in the isothermal−



RESULTS AND DISCUSSION Single-Molecule Force Spectroscopy. Immobilization of Ubiquitin Mutants and Multimodal Ligands. Figure S1A in the Supporting Information shows the “preferred” binding face and the location of residues with dissociation constants (KD) calculated from the NMR experiments reported previously.44 The intent of the current study was to determine the energetics of unbinding of this “preferred” binding face as well as the “opposite” side of ubiquitin to a multimodal surface using both SMFS and molecular simulations. The targeted probing of a particular face of ubiquitin required constraining the protein in a specific orientation on the substrate. Accordingly, sitedirected mutagenesis was employed to create cysteine mutations on two faces of the protein surface, as shown by the residues highlighted in red in Figure S1B. After expression and purification, ubiquitin mutants UBQ S20C and UBQ A46C were covalently immobilized at a controlled density onto a gold C

DOI: 10.1021/acs.langmuir.7b03011 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 1. Schematic representation of various components of single-molecule force spectroscopy experimental setup. (A) Exposed “preferred” binding face (cyan) by functionalization of UBQ S20C and (B) the steric shielding of the “preferred” face by functionalization of UBQ A46C, that is, exposure of the “opposite” face (black). Cysteine mutation on protein surface is denoted in red. The protein is immobilized to a SAM having EG6 linkers. The remaining SAM strands terminate in a hydroxyl group, thus providing a hydrophilic background to these proteins. (C) Detail of functionalized AFM cantilever surface presenting “Capto” multimodal ligands. (D) Schematic diagram of the Capto MMC headgroup. It is capable of interacting with the protein via hydrophobic interactions and π−π and π-cation interactions via a phenyl group, hydrogen bonding by an amide linkage, electrostatic interactions by a carboxylic acid group, and hydrophobic interactions by an aliphatic tail.

to confirm the immobilization chemistry. XPS data are also presented in Table S1 for a “Capto” ligand-functionalized AFM tip surface where the nitrogen content was roughly two times that of the sulfur content, which was expected based on the ligand chemistry. Prior to obtaining meaningful data from the force spectroscopy experiments, it was important to first conduct measurements with varying ligand densities of EG6N. The protein densities on the substrates were diluted in a sequential manner by controlling the EG6N ligand density in the underlying SAM surface. Force measurements were then conducted on these surfaces functionalized over a range of protein densities to establish conditions that resulted in single-molecule binding events. At very low ligand densities no binding events were observed. As the ligand density was increased, the point where binding events were first observed was taken as the density corresponding to single molecule events. This corresponded to a SAM containing a mixed monolayer composed of 0.03% EG6N and 99.97% EG6OH. Using previously established data for SAM formation on gold, the surface density of ubiquitin corresponding to this mixed monolayer composition was calculated to be roughly 1 protein per 800 nm2.67 Importantly, the structural similarity of the EG6N and the EG6OH linkers prevents the formation of clusters of amine head groups in the SAM prior to immobilization of the protein, which is expected to result in a relatively homogeneous distribution of the immobilized proteins.42 Estimation of Face-Specific Free Energy of Unbinding. SMFS was then employed to probe the energetics of ubiquitin binding to the “Capto” SAM functionalized on the AFM probe tip. “Capto” ligand-coated AFM tips were retracted from the protein immobilized surfaces at finite velocities (see schematic in Figure 2A) to obtain force−distance retraction curves, and a representative force−distance curve for one such binding event is shown in Figure 2B. 2000 such force−distance measurements were performed at each retraction velocity, and five different velocities were employed for three different surfaces (immobilized UBQ S20C, immobilized UBQ A46C, and the control with no protein immobilized). For each force−distance curve, the minimum corresponds to the point where the protein detached from the probe surface and the force value at this

substrate functionalized with SAMs having EG6 linkers, as described in the Materials and Methods section. In brief, SAMs were prepared on gold substrates from thiols containing hexaethylene glycol that were terminated with either hydroxyl (EG6OH) or primary amine (EG6N) groups. The amine groups of EG6N were then acylated with the bifunctional linker sulfosuccinimidyl-4-(N-maleimidomethyl) cyclohexane-1-carboxylate (SSMCC). The maleimide group of SSMCC was then reacted with the thiol groups on the cysteine mutants of ubiquitin, as shown in Figure S2. Similar chemistry has been previously employed where peptides were immobilized at controlled densities.16 As can be seen in Figure 1, this facespecific immobilization technique was expected to result in presentation of the “preferred” face for the S20C mutant (Figure 1A) and shielding of this face for the A46C mutant (Figure 1B). To perform the force measurements with these immobilized ubiquitin mutants, it was necessary to functionalize the gold-coated tip of an AFM cantilever with a SAM presenting the MM ligand (Figure 1C). This ligand, referred to as “Capto” ligand, is representative of an industrially important MM chromatographic ligand, which is rapidly gaining importance in the bioprocessing industry (Capto MMC from GE Healthcare).44 Figure 1D shows this ligand, which is capable of interacting with protein surfaces through hydrophobic interactions via its phenyl and alkyl groups, electrostatic interactions via a carboxylic acid group, and hydrogen bonding via an amide linkage. Surface Characterization. To evaluate the chemistry of these surfaces, X-ray photoelectron spectroscopy was carried out to provide an elemental analysis at different stages of immobilization for the immobilized protein surfaces and the final functionalized AFM tip (Table S1 in the Supporting Information). The XPS data confirmed the formation of SAMs and the subsequent immobilization of ubiquitin onto these substrates. The nitrogen content increased to the same extent for both ubiquitin mutants following the protein immobilization step, confirming that the proteins were covalently immobilized at similar surface densities. It is important to note that the data generated in Table S1 were for surfaces that had a mixed monolayer composed of 4% EG6N and 96% EG6OH. This ratio was selected to enable characterization and D

DOI: 10.1021/acs.langmuir.7b03011 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 3. Determination of “face”-specific binding strength of ubiquitin with the MM ligand surface. Mean unbinding force values (denoted by symbols) calculated from repeated force−distance measurements are plotted against retract rates for experiments with the exposed “preferred” face of ubiquitin (UBQ S20C), the “opposite” face of ubiquitin (UBQ A46C), and in the absence of ubiquitin (control). Curves are a fit to the two-state theoretical model,68,69 from which the value of the free energy of unbinding, ΔGU, is calculated.

significant difference in the unbinding free energy between the two mutants. The mutant exposing the “preferred” face (UBQ S20C) produced a ΔGU = 131.5 ± 16.5 kJ mol−1 that was much higher than the value of 29.5 ± 2.4 kJ mol−1 obtained for the mutant with the “opposite” face exposed (UBQ A46C) (Figure S4). The “opposite” face showed binding strength close to that for the control surface having no proteins immobilized (see Figure S4). Although it cannot be claimed in this analysis that the interactions of the protein only with the ligand-coated tip have been deconvoluted out from nonspecific interactions emanating from the hydrophilic background of the proteinimmobilized SAM, it can be observed clearly that the trend in equilibrium forces (and unbinding free energies) confirms the previously obtained preference of Capto ligands for the “preferred” face of ubiquitin via nuclear magnetic resonance spectroscopy.44 Additionally, it should be noted that the manifestation of nonspecific interactions here is likely to occur in the tail of the unbinding forces distributions for the protein (for example, the force distribution shown in Figure 2C) corresponding to the unbinding events in the control experiments. To re-emphasize our main findings by employing SMFS: (i) The unbinding forces (and equilibrium free energies) are considerably more favorable in the “preferred” face experiments compared with the “opposite” face, demonstrating the functionality of the “preferred” face-Capto ligand interaction; (ii) the opposite face and control surface have very close ΔGU, indicating their interactions being dominated by nonspecific interactions. In the next section, molecular dynamics simulations were employed further to recheck such preferences. Molecular Dynamics Simulations. Our main focus in this section was to confirm the preference of “Capto” ligand for a particular face on ubiquitin and to complement experimental single-molecule force measurements by calculating free energies of binding of two faces of ubiquitin to the “Capto” SAM surface (Figure 4A, with detail in Figure 4B). This surface was constructed by functionalizing some of the SAM surface headgroups with the “Capto” chemistry, with the remaining headgroups being −OH, providing a hydrophilic background (see Figure 4A and Figure S5A).49,50

Figure 2. Single-molecule force spectroscopy experiments. (A) Schematic representation of single-molecule force spectroscopy experiment on the specific face of ubiquitin. The red residue shows the point of covalent attachment to the surface via cysteine mutation. (B) Representative force−distance curve showing a binding event between the “Capto” SAM on the AFM tip and functionalized protein with “preferred” face exposed (UBQ S20C). (C) Histogram of detachment force measured from repeated force−distance curves for interaction between the MM ligand functionalized tip (spring constant 53.46 pN nm−1) and immobilized UBQ S20C at a retract velocity of 800 nm s−1.

minimum is the detachment force. From the cumulative distribution functions (CDFs) of the detachment forces, the mean force value was calculated for each species at a given retract rate. (See the Supporting Information text for more details and Figure S3 for representative CDFs.) Figure 3 shows the mean nonequilibrium detachment force ⟨F⟩ plotted versus retract rate for the “preferred” face of ubiquitin (UBQ S20C), the “opposite” face of ubiquitin (UBQ A46C), and in the absence of ubiquitin (control). As can be seen in Figure 3, the data indicate that the “preferred” face of ubiquitin (UBQ S20C) exhibited a higher mean nonequilibrium detachment force ⟨F⟩ at all retract velocities. Furthermore, the data in Figure 3 indicate that the force values for the mutant with the “opposite” face exposed (UBQ A46C) were similar to the values obtained with the control surface over a range of retract velocities. We then estimated the tip surface unbinding free energy, ΔGU, by fitting a theoretical model reported by Friddle et al. (see the Supporting Information text for details) that estimates equilibrium free energies from the nonequilibrium mean unbinding forces.68,69 As can be seen in Figure S4, these results demonstrated a E

DOI: 10.1021/acs.langmuir.7b03011 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 4. Schematic representation of various components of molecular dynamics simulations setup. (A) MD simulation snapshot of system having SAM, water, and ions. The z* axis originates from the SAM−water interface and extends into bulk (refer to Figure S5 for details of origin of the z* axis). SAM is shown in gray sticks, “Capto” ligand head groups in tan sticks, hydroxyl head groups in red (oxygen) and white (hydrogen) spheres, ions in blue spheres, and water in line representations. (B) Snapshot showing SAM strands having “Capto” ligand head groups only. Phenyl moiety is shown in yellow, amide linkage in orange, carboxylic acid group in red, and aliphatic tail in green sticks. (C) Snapshot of “preferred” face of ubiquitin (cyan surface) interacting with “Capto” SAM. Water molecules are not shown for clarity. (D) Snapshot of “opposite” face of ubiquitin (black surface) interacting with “Capto” SAM. Similar SAM surfaces were also constructed having only −OH head groups (not shown) to study interactions of the above two faces of ubiquitin with this reference hydrophilic surface.

Figure 5. Molecular dynamics simulations-based investigation of protein−SAM binding (color coding for parts B, C, E and F: SAM shown in gray sticks, −OH head groups in red and white spheres, “Capto” head groups in tan sticks, sodium counterions in orange spheres, protein backbone as cyan cartoon, positively charged side chains involved in binding in blue, negatively charged in red, hydrophobic in green, hydrogen bonding in magenta, water molecules within 0.4 nm of protein and surface shown in lines representation). (A) −OH SAM-Ubiquitin PMFs for “preferred” and “opposite” faces interacting with this surface. (B) Representative structure of ubiquitin’s “preferred” face bound to the −OH SAM. (C) Representative structure of ubiquitin’s “opposite” face bound to the −OH SAM. (D) “Capto” SAM-Ubiquitin PMFs for “preferred” and “opposite” faces interacting with this surface. (E) Representative structure of ubiquitin’s “opposite” face bound to the “Capto” SAM. (F) Representative structures of ubiquitin’s “preferred” face bound to the “Capto” SAM at different separation distances.

both the “preferred” and “opposite” faces of ubiquitin (Figure 4C,D, respectively) interacting with the “Capto” surface as well as with a reference hydrophilic −OH surface. SMFS provides an estimate of the (reversible) work required to pull ubiquitin from the SAM surface to the bulk water phase. The statistical mechanical equivalent of this quantity is

We performed simulations of reference systems containing only water, ions, and the SAM surface to calibrate density profiles (see Figure S5B) and also performed umbrella sampling simulations containing proteins that enabled the calculations of face-specific interactions of the protein with this “Capto” SAM. These umbrella sampling simulations focused on F

DOI: 10.1021/acs.langmuir.7b03011 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

were subsequently used to estimate equilibrium adsorption free energies. It is quite remarkable that despite these distinct differences the trends in the free energy are similar in both systems, with the previously hypothesized preferred face of ubiquitin exhibiting significantly stronger binding than the opposite face. The features in the PMFs were further examined to shed light on the mechanism of binding of the two faces of ubiquitin to the “Capto” and −OH SAMs. For the −OH surface, both the “preferred” and “opposite” faces interacted primarily via hydrogen bonding (Figure 5B,C, respectively). For the “Capto” SAM surface, a single minimum was observed for the “opposite” face at z* = 1.75 nm (Figure 5D), where a favorable interaction occurred between K63 and the carboxylic acid moiety on the “Capto” SAM. Interestingly, this interaction was favorable despite an abundance of negatively charged residues on this face (E16, E18, and E64), which could contribute to electrostatic repulsion from the SAM surface. The interaction of the “preferred” face with the “Capto” SAM was more complex (Figure 5F). At position 4 on this profile the protein did not directly contact the SAM, although there was a weak favorable interaction at this distance (z* = 3.35 nm). At position 5 (z* = 2.55 nm), the protein became anchored to the SAM via the electrostatic interactions of R74. As the protein further approached the SAM (position 6 at z* = 1.8), an additional electrostatic contact was observed with K11 along with a group of three hydrophobic residues, which were separated from the SAM by a layer of water. Finally, following desolvation, another minimum was observed at position 7, where these hydrophobic residues were in direct contact with the SAM. These results illustrate how the interplay of favorable electrostatic and hydrophobic interactions can result in the significantly stronger multimodal interactions that were observed for the “preferred” face in the single-molecule force spectroscopy experiments.

represented by the difference in excess chemical potentials of the protein in the vicinity of the surface and the bulk (depicted here as ΔGads). In a system where the protein molecule is present at a certain concentration, this difference can be obtained by the knowledge of equilibrium number densities in a “vicinal” region near the surface and in the bulk. Alternatively, it can also be calculated from the detailed potential of mean force, w(z*), for a protein as it is moved along the direction, z*, perpendicular to the surface using eq 1 (see Supporting Information text section “Origin of Equation 1 in Main Text” for more details) ⎧ w(z*) ⎫ ⎬ ΔGads = − kT ln exp⎨− ⎩ kT ⎭

(1)

where k is the Boltzmann constant and T is absolute temperature. The “vicinal” region chosen here stretches from z* = zmin * to z* = (zmin * + 1) nanometers, where zmin * , in each PMF is the point on z* below which data is not available (that is, PMF is infinitely large). In other words, z*min is where the protein was in contact with the surface such that it could not penetrate into the surface any further. Because the PMF is reported with respect to the protein center of mass, this lower bound is nonzero. The upper limit was chosen as (z*min + 1) nanometers because extending the “vicinal” domain further than this led to no appreciable change in the value of ΔGads (