Stability of Tethered Proteins - Langmuir (ACS Publications)

Mar 31, 2009 - Gaurav Anand , Fuming Zhang , Robert J. Linhardt , and Georges Belfort ... Sumanth N. Jamadagni , Shekhar Garde and Georges Belfort...
0 downloads 0 Views 1MB Size
Download PDF
pubs.acs.org/Langmuir © 2009 American Chemical Society

Stability of Tethered Proteins Gaurav Anand,† Sumit Sharma,‡ Sanat K. Kumar,*,‡ and Georges Belfort*,† †

Howard P. Isermann Department of Chemical and Biological Engineering and Center for Biotechnology and Interdisciplinary Studies, Rensselaer Polytechnic Institute, Troy, New York 12180-3590, and ‡Department of Chemical Engineering, Columbia University, New York, New York 10027 Received November 13, 2008. Revised Manuscript Received February 11, 2009

The stability of tethered globular proteins under denaturing conditions was interrogated with a hydrophobic surface, since conventional structural methods like circular dichroism (CD) and fluorescence or infrared spectroscopy could not be used because of the presence of an opaque solid substrate and extremely low surface concentrations. For free protein in solution, CD spectra gave well-known unfolding denaturing curves for lysozyme (LYS) and ribonuclease A (RNase A). The unfolding process for covalently tethered LYS and RNase A was followed, with multimolecular force spectroscopy (using an atomic force microscope in force-mode), via the adhesion energy between a functionalized self-assembled monolayer (CH3-SAM) probe and the protein molecules covalently bound to a carboxylic SAM on a gold-coated glass coverslip. The adhesion energy passed through a maximum for the tethered proteins during excursions with temperature or chemical denaturants. The initial rise in adhesion energy on increasing the temperature or GuHCl concentration was due to increasing exposure of the unfolded hydrophobic core of the proteins to the CH3-SAM tip, while the decrease in adhesion energy at high temperature or large concentrations of denaturant is attributed to interprotein association with nearest neighbors. Attempts to recover their folded state upon cooling (or reducing GuHCl concentration) were unsuccessful. Also, dilution of surface-tethered LYS reduced the aggregation with nearest neighbors about 6-fold. These results are in qualitative agreement with Monte Carlo simulations on a simple two-letter lattice protein model, especially for low concentrations of grafted proteins.

Introduction The adsorption of proteins to surfaces is a topic that has received considerable interest in the past few years.1,2 Recent reports have focused on searching for protein-resistant surfaces,3-5 conformational changes induced by different surface chemistries,6 and on the effect of surfaces on protein aggregation (amyloid fibrillation).7,8 Little, to our knowledge, has been reported on the structural behavior of tethered proteins at interfaces, let alone their denaturing profiles and stability and how these properties compare with proteins in the bulk phase. This is somewhat surprising since, in vivo, many proteins are attached to biological surfaces such as the cell membrane, chaperones during folding, cornea, bones, and arteries. A complicating factor is the difficulty in measuring the conformational behavior of surface-associated proteins due to their low concentrations. Techniques such as fluorescence and attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy have difficulty at such low interfacial concentrations, especially for monolayer coverage.3 The goals of the research were to determine the unfolding behavior of tethered proteins, to compare the *Corresponding author. E-mail: [email protected] (G.B.); sk2794@ columbia.edu (S.K.K.). (1) Malmsten, M. Biolpolymers at Interfaces; Surfactant Science Series; Marcel Dekker, Inc: New York, 2003; p 110. (2) Calonder, C; Tie, Y; Van Tassel, P. R. Proc. Natl. Acad. Sci. U.S.A. 2001, 98 (19), 10664–10669. (3) Sethuraman, A; Vedantham, G; Imoto, T; Przybycien, T; Belfort, G. Proteins 2004, 56(4), 669–678. (4) Kane, R. S.; Deschatelets, P; Whitesides, G. M. Langmuir 2003, 19(6), 2388–2391. (5) Ostuni, E; Chapman, R. G.; Liang, M. N.; Meluleni, G; Pier, G; Ingber, D. E.; Whitesides, G. M. Langmuir 2001, 17(20), 6336–6343. (6) Sethuraman, A; Belfort, G. Biophys. J. 2005, 88(2), 1322–1333. (7) Sluzky, V; Tamada, J. A.; Klibanov, A. M.; Langer, R. Proc. Natl. Acad. Sci. U.S.A. 1991, 88(21), 9377–9381. (8) Nayak, A; Dutta, A. K.; Belfort, G. Biochem. Biophys. Res. Commun. 2008, 369(2), 303–307.

4998

DOI: 10.1021/la803771d

denaturation behavior of proteins in solution with those tethered to a surface, and to develop an experimental measurement technique for measuring the destabilization of tethered proteins. Consequently, we present a method here, multimolecular force spectroscopy (MMFS) that allows one to probe the surfaceassociated protein layers. The method is similar to chemical force spectroscopy (CFS). In MMFS, an atomic force microscope (AFM) in force-mode, i.e., with a 10 μm diameter borosilicate sphere attached to a cantilever tip, is used. Two model globular proteins, hen egg lysozyme (LYS) and ribonuclease A (RNase A), were chosen because they exist as monomers, have a conserved hydrophobic core, and have been extensively studied. In order to compare the denaturing properties of natively folded proteins in bulk solution with those tethered to a surface, and to probe the connection between misfolded proteins and adhesion, we purposely denatured covalently bound proteins using temperature or denaturant excursions. Then, the adhesion (binding) energy between the tethered proteins (on a gold-coated glass coverslip) and a layer of CH3-terminated self-assembled alkanethiol monolayer (CH3-SAM) (on the sphere attached to the cantilever) was measured. To help explain the thermally or chemically induced denaturation of LYS and RNase A, Monte Carlo (MC) simulations on model lattice peptides tethered to a surface were undertaken. Two letter lattice models proposed by Dill9 and tethered to a “hydrophilic” surface were used. These are then contacted with a hydrophobic surface to simulate the laboratory experiments that used a CH3-SAM cantilever tip, and the number of protein-hydrophobic surface contacts as a function of temperature was estimated. We use this equilibrium measure as an indicator of the adhesion energy measured in the experiments. As is well understood in the (9) Dill, K. A. Biochemistry 1990, 29(31), 7133–7155.

Published on Web 3/31/2009

Langmuir 2009, 25(9), 4998–5005

Anand et al.

Article

polymer community, this approach ignores any viscoelastic contributions to adhesion. This is a serious shortcoming. Nevertheless, this quantity’s temperature dependence agrees qualitatively with experiments, and the protein’s adhesion energy is maximized in the vicinity of the unfolding transition for the tethered protein. However, qualitative differences exist between the experiments and the theory at high temperatures, and possible reasons for these differences are discussed below.

Experimental Section Materials. All materials and reagents were used as received. Glass coverslips (0.20 mm, Corning, New York) and AFM cantilevers were coated with 15 nm of titanium (Ti, 99.999% International Advanced Materials, Spring Valley, NY) followed by 50 nm of gold (99.999%, International Advanced Materials) using the electron beam evaporator under a pressure of less than 10-6 Torr. Hen egg LYS, N-hydroxysuccinimide (NHS), water-soluble 1-ethyl-3-(3-dimethylaminopropyl) hydrochloride carbodiimide (EDC) were purchased from Sigma-Aldrich Chemicals, St. Louis, MO. Bovine pancreatic RNase A was purchased from Worthington biochemical corporation, Lakewood, NJ. All the solutions were filtered using 0.22 μm poly (vinylidine difluoride) (PVDF) membranes (Millipore corporation, Bedford, MA). Methods. QCM-D. A quartz crystal microbalance with :: dissipation (QCM-D) (D300 System, Q-Sense AB, Goteborg, Sweden) was used to follow the amount of protein grafted per unit area and the dissipation with time. QCM-D is an ultrasensitive weighing device which can detect adsorbed mass to the resolution of less than 1 ng/cm2. The device consists of a thin disk of quartz with metal electrodes coated on the both faces of the disk. The crystals used in our study were gold-coated to facilitate SAM formation and subsequent chemical functionalizations. The crystal is connected to an external circuit that induces the crystal to oscillate in a shear mode at its resonant frequency, f. The lateral amplitude of vibrating crystal is 1-2 nm. The mass adsorbed or desorbed from the gold surface induces a frequency shift, Δf, which is a function of change in mass, Δm. Also, any mass that adsorbs to the surface will oscillate with the same lateral displacement and frequency as the underlying crystal. If the adsorbed film is elastic, it oscillates in phase with the crystal, and there is no energy loss. If however, the film is inelastic (of soft matter like proteins, cells, lipid bilayers), energy is dissipated via shear waves. The dissipation factor D, is defined as D ¼

EDissipated 2πEStored

ð1Þ

where EDissipated is the energy dissipated during one oscillation period, and EStored is the energy stored during the oscillation. In contrast to rigid films, the viscoelastic properties of soft matter give rise to energy dissipation, i.e., ΔD > 0. For adsorbed mass with no slip, rigid attachment, and small mass as compared to the crystal mass, Sauerbrey10 derived a simple relationship between the adsorbed mass Δm and the change in frequency Δf. Δm ¼ -C

Δfn n

ð2Þ

where C = 17.7 ng 3 cm-2 3 Hz1-, n is the overtone number, n = 3,5,7, and f is the frequency of the overtone. Gold-coated AT-cut quartz crystals with fundamental frequency of ∼5 MHz were cleaned by immersion in a 1:1:5 mixture of H2O2 (30%), NH3 (25%), and distilled water at 60 C for 20 min. The cleaned crystals were then washed with a distilled water/ethanol mixture (10) Sauerbrey, G. Z. Phys. A: Hadrons Nucl. 1959, 155(2), 206–222.

Langmuir 2009, 25(9), 4998–5005

(50:50) and then dried with nitrogen gas. The crystals were then exposed to UV-ozone for 10 min and were thoroughly rinsed with ethanol and dried under nitrogen before use. Crystals were soaked in 2 mM solution of HS(CH2)11COOH in ethanol for 12 h, rinsed with ethanol and then dried under nitrogen. The terminal carboxylic groups were then equilibrated with deionized water for 15 min followed by soaking into an equal volume mixture of 0.1 M NHS and 0.4 M water-soluble carbodiimide EDC for 30 min. The crystals were then washed with deionized water in triplicate and installed inside the QCM chamber. Then, LYS and RNase A were chemically grafted onto the carboxy-SAM layer on the coverslips by the NHS chemistry. Four separate resonant frequencies (overtones, n = 1, 3, 5, and 7) were used to detect the oscillation of the shear wave through the crystal at 5, 15, 25, and 35 MHz, respectively.11 The data from the seventh overtone is reported as it has the minimum noise. Atomic Force Microscopy. The “molecular puller” was a onedimensional (1-D) AFM (MEFP-1 Asylum Research, Santa Barbara, CA). Borosilicate glass spheres of 10 μm diameters were attached to the AFM cantilever tips (Si3N4 Novascan, Ames, IA). The cantilever tip was modified with a gold-coated 10 μm diameter borosilicate sphere and was coated with a monolayer of HS(CH2)11CH3 (Sigma-Aldrich Chemicals, St. Louis, MO) SAM. The spring constant (0.06 N/m) of each cantilever was recalibrated before measuring the adhesion forces using a two step procedure. First, the slope of the contact region during force-distance measurements was used to calculate the sensitivity of the lever in nanometers per volt, and then a “thermal tune” was performed to determine resonant frequency of the cantilever. An algorithm in IGOR (Wavemetrics Inc., Portland, OR) computed the spring constant using the Equi-partition theorem.12 Using the Deriaguin approximation to convert adhesion forces, Fa, into energy, Ea, of interaction, between two flat surfaces (large sphere of radius, R, and flat substrate), the measured forces, Fa, were normalized by the radius, R (5 μm), of the silica sphere, such that Ea = Fa/R.13 Here the loading rates are kept constant for all runs. They were 1 mN/m, and the time period under constant compliance was 1 s. Protein Tethering. For alkanethiol SAM assembly, goldcoated coverslips and AFM probes were respectively soaked in a 2 mM solution of HS(CH2)11COOH (COOH-SAM, SigmaAldrich Chemicals, St. Louis, MO) and HS(CH2)11CH3 (CH3SAM, Sigma-Aldrich Chemicals) in ethanol for 12 h, rinsed with ethanol, and then dried under nitrogen. Then, LYS and RNase A were chemically grafted onto the carboxy-SAM layer on the coverslips by the NHS chemistry.14 Figure 1 is a schematic of the experimental set up and shows the interaction between the probe (CH3-SAM) (top) and the tethered protein (bottom). Protein Denaturation and MMFS. Intermolecular adhesion energy measurements were generated between CH3-SAM surface on the cantilever tip and the protein covalently immobilized to SAM-COOH on the glass coverslip. All the force measurements were conducted in 10 mM PBS buffer at pH 7.4. At pH 7.4, the carboxylic acid group was ionized and negatively charged (pKa 5.5). Control experiments were performed without grafted protein, and the adhesion between the CH3-SAM surface and the SAM-COO- surface was negligible (data not shown). This result is in agreement with the previous studies by Sethuraman et al.15 To chemically perturb the protein structure, protein grafted substrates were soaked in phosphate-buffered saline (PBS buffer) at varying concentrations of GuHCl (Sigma-Aldrich (11) Dutta, A. K.; Belfort, G. Langmuir 2007, 23(6), 3088–3094. (12) Hutter, J. L.; Bechhoefer, J. Rev. Sci. Instrum. 1993, 64(7), 1868–1873. (13) Derjaguin, B. V. Kolloid Z. 1934, 69, 155–164. (14) Lahiri, J; Isaacs, L; Tien, J; Whitesides, G. M. Anal. Chem. 1999, 71(4), 777–790. (15) Sethuraman, A; Han, M; Kane, R. S.; Belfort, G. Langmuir 2004, 20(18), 7779–7788.

DOI: 10.1021/la803771d

4999

Article

Anand et al.

Figure 1. Schematic of the MMFS experiment using an AFM. Two surfaces interact in the AFM; the top surface is a microsphere of radius R =5 μm functionalized with CH3-SAM attached to the AFM cantilever, and the bottom surface is a coverslip onto which a layer of covalently attached protein (LYS or RNase A) to COOHSAM using NHS chemistry is fixed. The lysines on the exterior of protein molecules, shown by black patches, chemically bond with the COOH-SAM by EDC/NHS coupling chemistry. Gray patches show the conserved hydrophobic groups within the core (figure not to scale). Chemicals) (0-6 M) for 12 h in a refrigerator (4 C). The substrates were thoroughly rinsed with clean PBS in triplicate to wash off any salt before taking the adhesion measurements. For the thermal experiments, 0.1 mg/mL LYS and RNase A in 25 mM sodium phosphate buffer were grafted onto COOHSAM-gold substrates and heated to different temperatures for 30 min. Then, they were rapidly cooled to room temperature, and the adhesion measurements were immediately obtained with CH3-SAM on the gold coated borosilicate sphere, which in turn was attached to the AFM cantilever tip. Circular Dichroism (CD). To follow denaturation of bulk protein solutions, CD experiments were performed. For thermal and chemical denaturation experiments, LYS and RNase A were dissolved in 25 mM sodium phosphate buffer at a concentration of 0.5 mg/mL. A quartz cuvette of path length 1 mm (Hellma, Inc., Denmark) was used to obtain the far-UV spectrum (190-260 nm) of different samples. Loss of R-helix was recorded with increasing temperature or GuHCl concentration by measuring the relative molar ellipticity at λ = 222 nm.16 To compare the tethered results (AFM) with those in solution (CD), excursions in both temperature and denaturant were pursued. To chemically perturb the protein structure in bulk solution, varying concentrations of GuHCl (0-6 M) in PBS were added to the protein solution, and the aliquots were stored for 12 h in a refrigerator (4 C) before obtaining the CD spectra. Standard thermal denaturation curves of each protein in solution were obtained, i.e., individual samples were heated to a certain temperature for 30 min and then rapidly cooled down to room temperature before obtaining CD spectra. All the AFM and CD measurements were conducted at 22 ( 0.1 C. Temperatures as high as 90 C were used to denature proteins tethered on the (16) Madison, V; Schellman, J. Biopolymers 1972, 11(5), 1041–1076.

5000

DOI: 10.1021/la803771d

surface and in the bulk solution. In the AFM experiments such high temperatures lead to various complications, e.g., thermal frequency shifts, change in cantilever spring constant, changes in deflection sensitivity, enormous amount of noise, and may cause substantial evaporation of the buffer. Therefore the coated glass coverslips were exposed to different temperatures in a water bath and then were rapidly cooled to room temperature before taking the force measurements using the AFM. Although the CD instrument is capable of automatically ramping the temperature up to 95 C, we still heated the samples separately and quenched to room temperature before taking the CD measurement. The melting temperature of LYS and RNase A did not change as a consequence of rapid quenching as compared with unquenched heating.17 This allowed comparison of the surface denaturation (monitored by MMFS) with the denaturation in bulk (monitored by CD). Unfolding Reversibility. To probe the reversibility of unfolding and to explain the decrease in adhesion above Tmax and [GuHCl]max, two different experiments at two extreme temperatures and two extreme denaturant concentrations were chosen, respectively, on either side of the maximum values from the adhesion curves. After tethering LYS to the COOH-SAM as described above, the swatch was immersed separately in (i) PBS and (ii) 6 M GuHCl in PBS for 12 h at 4 C (on either side of [GuHCl]max), then washed with PBS to remove the GuHCl, and the adhesion energy was measured. The two samples were stored at 4 C for 12 h in PBS and allowed to slowly refold; they were then washed with PBS at room temperature (22 ( 0.1 C), and again the adhesion energy was measured. It has been reported that LYS refolds in bulk solution in about 4 h, but we chose 12 h because, in our case, the LYS molecules were tethered to the surface and therefore have diminished entropy and degrees of freedom to refold.18 Next, the samples were exposed to their maximum, i.e., to [GuHCl]max = 4 M in PBS for 12 h at 4 C (i.e., at the maximum point of [GuHCl]max), and adhesion energy was again recorded after washing with PBS buffer at room temperature. In a second set of experiments, after tethering LYS to the COOH-SAM as described above, the swatch was immersed in PBS at 22 and 90 C (on either side of Tmax,) for 30 min, and the adhesion energies were measured between the CH3-SAM probe and the tethered proteins on each swatch. Then, both samples were stored at 4 C for 12 h in PBS and allowed to slowly refold. Samples were then immersed in PBS at room temperature (22 ( 0.1 C) and again the adhesion energy was measured. Next, the samples were reheated to their maximum, i.e., to Tmax = 72 C (LYS) for 30 min, cooled rapidly to room temperature through fast immersion in PBS (22 ( 0.1 C), and adhesion energy was again recorded. Dilution Experiment. The goal in this experiment was to test whether interaction with nearest neighbors would be diminished when the surface concentration of tethered proteins was reduced by 50%. The thermal denaturation of diluted surface-tethered LYS was measured. LYS concentration on the surface was reduced by 50% using poly(ethylene glycol) (PEG)-terminated thiol as a diluent. The protocol was similar to that of Lahiri, et al., (1999).14 See Supporting Information for details. Simulations. MC simulations of lattice proteins tethered to an athermal or noninteracting surface were performed. The simulations were aimed at understanding how tethered proteins interact with a hydrophobic surface, such as a SAMcovered AFM tip with a hydrophobic sphere, and with neighboring tethered proteins. The equilibrium simulations cannot capture the contribution of viscoelastic effects to the adhesion energy when an AFM tip is retracted from the surface. (17) Pfeil W. Protein Stability and Folding: A Collection of Thermodynamic Data, Supplement 1; Springer: New York, 2001. (18) Gao, Y-G; Guan, Y-X; Yao, S-J; Cho, M-G. Korean J. Chem. Eng. 2002, 19 (5), 871–875.

Langmuir 2009, 25(9), 4998–5005

Anand et al.

Article

Figure 2. Chemical immobilization of LYS (a,b) and RNase A (c,d) as monitored with QCM-D in sodium phosphate buffer at pH 8.5 on a carboxyl SAM on gold activated by EDC and NHS. Protein solution was introduced at time t = 5 min, and unadsorbed solution was washed off at time t = 65 min with the background PBS buffer in which the proteins were dissolved. Mass of grafted (a) LYS and (c) RNase A. Change in energy dissipation for grafted (b) LYS and (d) RNase A. This phenomenon may play an important role in the experiments, and forms the basis of a subsequent study. In the lattice protein model introduced by Dill,9 each bead of the lattice protein represents an amino acid. The protein model has two kinds of beads: hydrophobic and polar. In addition to the constraint of single occupancy of a lattice site, which insures the hard core repulsion between beads, the hydrophobic beads were assumed to have nearest-neighbor attractions with each other of magnitude ε, while the interaction energy between two polar beads or between a polar and a hydrophobic bead was set to zero. All the empty sites on the lattice correspond to solvent. Lattice proteins were tethered at random amino acids in the sequence and at random locations on an athermal surface. The athermal or noninteracting surface represented the COOHSAM on which the proteins were tethered during the laboratory experiments. After tethering, the proteins were quenched to a low temperature with interprotein interactions set to zero to obtain folded-like globular structures. (Temperature is defined in reduced units, T* = kBT/ε, where kB is Boltzmann’s constant, and T is the temperature.) Above the tethered proteins, a hydrophobic probe was placed, which represented the AFM sphere covered with CH3-SAM in the laboratory experiments. The magnitude of interaction between a hydrophobic protein unit and the surface is equal to ε. Sequential canonical ensemble MC simulations were then performed, wherein each subsequent simulation was at a slightly higher temperature than the previous one. The MC simulations were conducted using the standard Metropolis method.19 At each temperature, 108 MC equilibration cycles were performed, followed by 108 production cycles. The adhesion energy with the top hydrophobic probe was calculated from the number of hydrophobic contacts between the tethered proteins and the probe surface. Three protein models;a 64 mer,20 a 124 mer,21 and a 248 mer lattice (19) Allen M. P., Tildesley D. J. Computer Simulation of Liquids; Oxford University Press, Inc: New York, 1987. (20) Yue, K; Dill, K. A. Proc. Natl. Acad. Sci. U.S.A. 1995, 92(1), 146–150. (21) Lattman, E. E.; Fiebig, K. M.; Dill, K. A. Biochemistry 1994, 33(20), 6158–6166.

Langmuir 2009, 25(9), 4998–5005

proteins;were studied. The 248 mer protein consisted of two 124 mer proteins joined end to end. A simulation of 2  108 MC cycles took approximately 24 h of CPU time on a Xeon dual core CPU for four tethered 64-mer proteins.

Results Protein Tethering. The tethering reaction of proteins (LYS, RNase A) to the gold-coated and COOH-SAM-covered coverslips was followed by QCM-D. Changes in mass, using the Sauerbrey equation to convert frequency change to mass,10 and dissipation for both LYS and RNase A are shown in Figure 2. Before introducing protein solutions, the activated crystals were stabilized with background buffer, 10 mM PBS, in which the proteins were dissolved, for more than an hour. After the crystals were saturated with buffer and there was no further drop in frequency, the data acquisition was restarted. In Figure 2, at time t = 0, the horizontal lines in the frequency and the dissipation plots correspond to the baseline obtained after the crystal equilibrated with the buffer. After 5 min of data acquisition, protein solutions were introduced at a concentration of 1 μM in 10 mM PBS buffer. The sudden increase in mass and dissipation after about 5 min is due to the immobilization of protein molecules through lysine residues by displacement of the NHS group. The reaction proceeded for 60 min and then the nonspecifically adsorbed protein was removed by flushing the sensor with 10 mM PBS buffer. As can be seen from the figure, the covalent grafting of LYS and RNase A onto the COOHSAMs appeared to reach a maximum grafted packing density of approximately 200 ( 20 ng/cm2 after washing at 60 min. Also, there was a decrease in mass and dissipation after the wash at ∼60 min due to the removal of loosely bound protein. Voros 22 showed that the change in dissipation was less for rigid adsorbed films as compared with more flexible films. This is because the (22) Voros, J. Biophys. J. 2004, 87(1), 553–561.

DOI: 10.1021/la803771d

5001

Article

Anand et al.

using CD222 for proteins in solution and normalized adhesion energy, Ea = Ea/Emax for tethered proteins on a surface. (a) The fraction of helix from CD222, as a function of varying [GuHCl] in a solution of 25 mM sodium phosphate buffer at pH 7.4 and 22 ( 0.1 C, [LYS] = 0.5 mg/mL and (b) with a normalized adhesion energy constant, Emax = 0.88 ( 0.04 mN/m, for covalently bound LYS with CH3-SAM, as a function of varying [GuHCl] in 10 mM PBS buffer at pH 7.4. (c) The fraction of helix from CD222, with varying [GuHCl] in a solution of 25 mM sodium phosphate buffer at pH 7.4 and 22 ( 0.1 C, [LYS] = 0.5 mg/mL and (d) with a normalized adhesion energy constant, Emax = 0.40 ( 0.02 for covalently bound RNase A with CH3-SAM, as a function of varying [GuHCl] in 10 mM PBS buffer at pH 7.4.

Figure 4. Thermal denaturation of LYS (a,b) and RNase A (c,d) using CD222 for proteins in solution and normalized adhesion energy, Ea = Ea/Emax for tethered proteins on a surface. (a) The fraction of helix from CD222, as a function of varying T in a solution of 25 mM sodium phosphate buffer at pH 7.4 and 22 ( 0.1 C, [LYS] = 0.5 mg/mL and (b) with a normalized adhesion energy constant, Emax = 0.95 ( 0.06 mN/m of covalently bound LYS with CH3-SAM, as a function of varying T in 10 mM PBS buffer at pH 7.4. (c) Thermal denaturation of RNase A. The fraction of helix from CD222, with varying T in a solution of 25 mM sodium phosphate buffer at pH 7.4 and 22 ( 0.1 C, [LYS] = 0.5 mg/mL and (d) with a normalized adhesion energy constant, Emax = 0.76 ( 0.03 mN/m, of covalently bound RNase A with CH3-SAM, as a function of varying T in 10 mM PBS buffer at pH 7.4.

rigid or less flexible adsorbed molecules vibrate in phase with the crystal and therefore dissipate less energy as compared with flexible molecules, which vibrate off-phase with the vibrating crystal and therefore dissipate more energy. Thus, the smaller dissipation for the RNase A layer suggests that LYS molecules are more flexible than RNase A molecules (Figure 2b,d). Adhesion Measurements with Chemical Denaturation. The data in Figure 3 compare the structural stability (using AFM for adhesion and CD for loss of helix at 222 cm-1) of tethered LYS (Figure 3a,b) and RNase A (Figure 3c,d) with that in solution during exposure to an increasing concentration, C (M), of a destabilizing osmolyte, guanidiium hydrochloride (GuHCl) at pH 7.4 and 22 ( 0.1 C. The literature values of Cm, defined as the molarity of chemical denaturant at which 50% of the protein structure is lost, for LYS is 3.62 M and RNase A is 3.0 M.17 From Figure 3a,c, the proteins in solution appear to retain 50% of their native structure, Cm, up to 4 M (LYS) and 3 M (RNase A) GuHCl. For the tethered proteins using MMFS (Figure 3b,d), we conjecture that the increase in normalized adhesion energy is due to the exposure of the conserved hydrophobic core of LYS and RNase A to the hydrophobic probe when the protein molecules start to unfold in the presence of increasing concentrations of GuHCl. The reason for the decline in normalized binding energy above 4 M (LYS) and 3 M (RNase A) GuHCl concentration has not been previously reported. We speculate that protein-protein interactions dominated once the unfolding process commenced on the surface. This question is further addressed below. The data points and errors for Figure 3b,d were obtained from the mean and variance of a Gaussian profile fitted to a histogram of adhesion energy

measurements between the CH3-SAM probe and the tethered protein at different concentrations of denaturant (see raw data in Figure 1S in the Supporting Information). Adhesion Measurements with Thermal Denaturation. The data in Figure 4 compare the structural stability (using AFM for adhesion and CD for loss of helix) of covalently bound LYS (Figure 4a,b) and RNase A (Figure 4c,d), with that in solution during exposure to increasing temperature, T, in PBS at pH 7.4. The melting points for LYS and RNase A, Tm, defined as the temperature at which the protein loses 50% of its native structure in solution have been reported as ∼73 C17 and ∼61 C,17 respectively, which is close to those observed here in solution at 78 and 65 C (Figure 4a,c). Since the measurement method required that the samples be removed from the test temperature and immediately cooled to room temperature for CD and AFM measurements, this suggests that LYS and RNase A retained their structural states during exposure to thermal stress followed by a rapid cooling. Both LYS and RNase A lose R-helix content steadily with increasing temperatures. As with the chemical denaturation experiments, the two tethered proteins exhibit normalized adhesion energy maxima during thermal denaturation (Figure 4b,d). After reaching the maximum normalized adhesion energy, the adhesion declined with further thermal increments, probably due to association of neighboring protein molecules on the surface. This is addressed below, when we test for reversible unfolding. Unfolding Reversibility. To probe the reversibility of unfolding and to explain the decrease in adhesion energy above Tmax and [GuHCl]max for one of the proteins (LYS), two different experiments were undertaken. Details of these experiments

Figure 3. Chemical denaturation of LYS (a,b) and RNase A (c,d)

5002

DOI: 10.1021/la803771d

Langmuir 2009, 25(9), 4998–5005

Anand et al.

Figure 5. Unfolding of surface-tethered LYS for (a) chemical and (b) thermal denaturation (Figures 3b and 4b) are included here to help align the new data in panels c-f with the master unfolding curves (a, b). The results for the control experiments are shown in panels c and d for chemical and thermal denaturation, respectively. (c) Step 1: After grafting, the LYS-tethered coverslip (sample) was soaked in PBS buffer for 12 h at 4 C and adhesion was measured (solid); Step 2: Sample was again soaked in PBS buffer for 12 h at 4 C and adhesion was again measured (striped); Step 3: Sample were soaked in 4 M GuHCl for 12 h at 4 C, and the adhesion was measured (checkered). (d) Step 1: After grafting, adhesion was measured at 22 C (solid); Step 2: Sample was again soaked in PBS buffer for 12 h at 4 C, and adhesion was again measured (striped); Step 3: Sample was exposed to 72 C for 30 min, and the adhesion was measured (checkered). (e) Step 1: After grafting, the swatch was soaked in 6 M GuHCl for 12 h at 4 C, and adhesion was measured (solid); Step 2: Sample was soaked in PBS buffer to facilitate refolding of LYS for 12 h at 4 C, and adhesion was measured (striped); Step 3: Sample was soaked in 4 M GuHCl for 12 h at 4 C, and the adhesion was measured (checkered). (f) Step 1: After grafting, the sample was exposed to 90 C for 30 min, and adhesion was measured (solid); Step 2: Sample was soaked in PBS buffer to facilitate refolding of LYS for 12 h at 4 C, and adhesion was then measured (striped); Step 3: Sample was again exposed to 72 C for 30 min, and the adhesion was measured (checkered).

are given in the Experimental Section. The goal was to first expose LYS to a very high denaturant condition ([GuHCl] and T) for a period (12 h and 30 min, respectively) as defined by the data in Figures 3b and 4b (Step 1). Then sufficient time (12 h) was allowed for refolding (Step 2), and finally the samples were placed at the maxima conditions ([GuHCl]max = 4 M and Tmax = 72 C) for a period (12 h at 4 C, and 30 min, respectively) (Step 3). After each step listed above, the adhesion energy between the tethered LYS and the CH3-SAM probe was measured. In Figure 5, the results are referred to as (a), (c), and (e) (for Steps 1-3 for GuHCl addition, respectively) and as (b), (d), and (f) (for Steps 1-3 for T-change, respectively). Figure panels 3b and 4b are redrawn as Figure 5a and 5b to facilitate better understanding of the experiments and results. Comparing the results in Figure 5 within error, we conclude the following. Both chemical and temperature denaturant excursions have similar effects on the behavior of tethered LYS. An increase in denaturant concentration or temperature to their Langmuir 2009, 25(9), 4998–5005

Article

respective maxima (Figure 5a,b) resulted in the expected high adhesion energy (Figure 5c,d). However, exposure of tethered LYS to very high denaturant concentrations (6M) or temperature (90 C) for 12 h and 30 min, respectively, did not allow LYS to attain maximum adhesion value (at [GuHCl] = 3 M and T = 72 C), even when allowed to refold in PBS buffer for 12 h at 4 C (Figure 5e,f). Dilution. Although the results in Figure 5 back the “interaction-between-neighbors” hypothesis during denaturation, for further support, LYS was tethered to the surface at 50% of the previous concentration (Figure 4b) and interdispersed with covalently attached PEG moieties during denaturation with increasing temperature. Thus, the lower availability of nearest neighbors should result in both a lower peak due to less protein and a lower slope after the peak due to fewer interactions (Figure S2). Again, we observed a peak at ∼72 C. However, the peak was broader, and the fall in adhesion energy beyond the peak was much shallower than that observed before, i.e., there was a 6-fold decrease in negative slope between the two cases. We interpret this to mean that there were significantly less proteinprotein interactions. Lowering the concentration further was not feasible with our method, as the peak was on the order of the error and could not be observed. Taken together, the unfolding irreversibility and the dilution experiments strongly support the “interaction-between-neighbors” hypothesis during denaturation. MC Simulations of Lattice Proteins. To explain the temperature dependence of adhesion energy with a hydrophobic surface from tethered proteins, MC simulations of tethered proteins were performed. The simulation system and technique has been described above. The equilibrium adhesion energy with the hydrophobic surface was defined as the average number of hydrophobic beads-surface contacts. Figure 6 shows the adhesion energy with the hydrophobic surface as a function of temperature for the three model proteins. The important features to observe are, first, that a peak in the adhesion energy is observed for all proteins and the profiles are asymmetric. There is a sharp increase in the adhesion energy with the hydrophobic surface at the temperature when the proteins unfold, but the drop in the adhesion energy at higher temperatures is less sharp, especially compared to experiments. However, the drop in adhesion energy at high temperatures is sharper for longer model proteins. In Figure 7, the radius of gyration square ÆR2gæ of the 64-mer tethered proteins is compared with that in the bulk. The ÆR2gæ of tethered proteins shows a sharp increase at the temperature where the maximum in adhesion energy is observed (T* = 0.35), indicating that the proteins unfold at that temperature. This behavior of ÆR2gæ of tethered proteins shows that they are less stable than those in the bulk, as observed in laboratory experiments. The simulation result can be understood as follows. At low temperatures, the proteins maintain their globular structure and do not interact with the hydrophobic probe surface. As the temperature is increased, the proteins unfold and start interacting with the hydrophobic probe surface leading to a jump in adhesion energy. On further increase in temperature the proteins lose some hydrophobic contacts with the hydrophobic surface to gain entropy and also the interprotein interactions increase when the proteins unfold. While our simulations qualitatively capture the experimental trends, they are quantitatively inaccurate. Two possible reasons include: First, the simulations measured the equilibrium number of hydrophobic monomersurface contacts and use this quantity as a surrogate for the DOI: 10.1021/la803771d

5003

Article

Anand et al.

the peak in adhesion energy becomes sharper for longer chains. This result also could help resolve the differences between the experiments and the simulations. Regardless of these differences this simple theory accounts qualitatively for the peak that is observed in the experiments. We have covered a larger parameter space in the simulations than that described above to explain the experimental results, but none of them have given a very satisfactory explanation. We have briefly discussed these in the appendix to the paper.

Figure 6. Adhesion energy in reduced units for different protein lengths as determined using MC simulations. 0 represents the 64-mer protein, O represents the 124-mer protein, and 3 represents the 248-mer protein.

Figure 7. A comparison of radius of gyration squared, ÆR2gæ, of 64-mer protein when tethered, O and when in bulk, 0.

adhesion energy or the thermodynamic work of adhesion. Previous work in polymer adhesion has similarly shown that the thermodynamic work of adhesion underestimates the experimentally measured work of adhesion by orders of magnitude, with this difference being attributed to viscoelastic effects.23 Since such viscoelastic effects are probably only relevant when the chains are unfolded, this, could potentially explain the discrepancy observed at high temperature. Second, our implicit solvent models, which do not account for water, thus, miss any changes in the magnitude of the hydrophobic effect with temperature. Since these aspects may be expected to play a large role at higher temperature, again, this might help rationalize the temperature discrepancy at high values. A related discrepancy is that our model proteins do not undergo intermolecular association, as has been deduced from the experiments. Preliminary results obtained by us, indeed suggest that (23) Baljon, A. R. C.; Robbins, M. O. Science 1996, 271(5248), 482–484.

5004

DOI: 10.1021/la803771d

Conclusions Proteins fold differently when in free solution and when confined by an external barrier.24 The folding and unfolding behavior of proteins confined in small volumes such as in the chaperone cavity25 the ribosome tunnel,26 on in polymeric pores during synthetic membrane filtration27 is of great interest. Here, we are concerned with protein unfolding when confined to a flat surface through tethering. Globular protein molecules have predominantly hydrophobic residues within their core. As the GuHCl concentration increased to ∼4 M or the temperature approached ∼72 C in separate denaturation studies, both LYS and RNase A started to unfold. Possibly, their inner hydrophobic core became exposed to the approaching hydrophobic AFM tip (CH3-SAM). The sudden jump in adhesion energy at these critical conditions was likely due to the strong hydrophobic interaction between the hydrophobic probe and the increasingly exposed hydrophobic protein core. It appears that when tethered-globular protein molecules started to unfold on the surface, individual hydrophobic moieties became accessible to the hydrophobic AFM cantilever tip and to nearest neighbors. This induced attraction due to hydrophobic interactions and resulted in a jump in the adhesion energy between the probe and the tethered proteins. As this process proceeded with greater exposure of hydrophobic residues with increasing denaturant (GuHCl) or temperature, proximal tethered proteins began to associate resulting in a decrease in exposure of hydrophobic residues to the solvent. This resulted in a lowering of the adhesion force with the CH3-SAM probe as observed here (Figures 3b,d and 4b,d). Both the unfolding irreversibility and the dilution experiments strongly supported the “interaction-between-neighbors” hypothesis during denaturation. We demonstrated that MMFS using AFM is a direct method to monitor unfolding of proteins tethered to solid substrates. When the denaturant concentration or temperature was raised above 4 M and ∼72 C, respectively, LYS quickly lost its tertiary structure, inducing protein-protein association through hydrophobic interactions and solvent exclusion. MC simulations to study the dependence of thermodynamic work of adhesion on temperature of tethered proteins on a hydrophobic AFM tip qualitatively agree with the results of the experiments. The simulations showed a maximum in the adhesion energy at a temperature when the proteins unfolded, but they miss the viscoelastic effects and the influence of water. The adhesion energy measured in the simulations was the interaction energy of tethered proteins with a flat hydrophobic surface, similar to the (24) Mittal, J; Best, R. B. Proc. Natl. Acad. Sci. U.S.A. 2008, 105(51), 20233–20238. (25) Chan, H. S.; Dill, K. A. Proteins 1996, 24(3), 345–351. (26) Nissen, P; Hansen, J; Ban, N; Moore, P. B.; Steitz, T. A. Science 2000, 289 (5481), 920–930. (27) Belfort G, Zydney, A. L. Interactions of proteins with polymeric synthetic membranes. In Biopolymers at Interfaces, 2nd ed.; Malmsten, M, Ed.; Marcel Dekker, Inc: New York, 2003.

Langmuir 2009, 25(9), 4998–5005

Anand et al.

Article

interaction energy measured between the hydrophobic SAM and the tethered proteins in the experiments.

(Rensselaer Polytechnic Institute) for useful suggestions and critical discussions.

Acknowledgment. We acknowledge the support of U.S. Department of Energy, DOE (DE-FG02-90ER14114 and DOE DE-FG02-07ER46429) and the National Science Foundation (Grant No. CTS-94-00610). We thank Amit K. Dutta

Supporting Information Available: Raw data, adhesion measurements of LYS on a surface with reduced surface density, and additional simulations. This material is available free of charge via the Internet at http://pubs.acs.org.

Langmuir 2009, 25(9), 4998–5005

DOI: 10.1021/la803771d

5005