Dynamic control of protein folding pathway with a polymer of tunable

represented by a Lennard-Jones potential. As reported else- where ..... (28) Miller, R.; Danko, C. A.; Fasolka, M. J.; Balazs, A. C.; Chan, H. S.; Dil...
0 downloads 0 Views 498KB Size
Download PDF
J. Phys. Chem. B 2007, 111, 12303-12309

12303

Dynamic Control of Protein Folding Pathway with a Polymer of Tunable Hydrophobicity Diannan Lu,† Jianzhong Wu,‡ and Zheng Liu*,† Department of Chemical Engineering, Tsinghua UniVersity, Beijing, 10084, and Department of Chemical & EnVironmental Engineering, UniVersity of California, RiVerside, California 92521 ReceiVed: July 30, 2007

While the knowledge of protein folding in a dilute solution is now well-advanced, little is known of the influence of surrounding conditions on the folding kinetics, in particular when the protein is in a dynamically responsive environment. Here we report a new procedure to control the pathways of protein folding by using a thermally responsive polymer that varies its hydrophobicity concomitant with the protein structural changes. The advantages of folding in a dynamic environment have been demonstrated first by Langevin dynamics simulations on the basis of coarse-grained models for both the protein and polymer and then by experiments for lysozyme refolding in the presence of poly(N-isopropylacrylamide-co-N-tert-butylacrylamide), a thermal responsive polymer that varies its hydrophobicity in response to temperature. The simulation suggests that decreasing the polymer hydrophobicity during the folding process may result in an optimized free-energy landscape that enhances both the folding yield and kinetics. The experiments affirm that an optimal folding condition can be identified when structural transitions of the protein collaborate with the polymer hydrophobicity tuned by variation of temperature.

Introduction

Model and Methods

Protein refolding, i.e., recovery of biologically active proteins from inclusion bodies, represents a major step in production of genetically engineered proteins.1-3 A main objective in the design and optimization of industrial processes is to increase the folding yield while inhibiting aggregation.4 This can be accomplished by the addition of surfactants,5-11 amphipathic molecules,12,13 polyols,14,15 or stimuli-responsive polymers16-24 that form complexes with the denatured proteins and prevent protein aggregation.16,17,25-27 Previous simulations and experiments indicate that, in the presence of additives, protein folding follows a “collapsing-rearrangement” two-step mechanism; i.e., the denatured protein first falls into a collapsed state and then folds into the native conformation.28-30 A folding additive promotes the collapse step via formation of a complex with the protein,31,32 but for efficient protein structural transition, it must be dissociated from the complex during the rearrangement step in which the protein acquires its native conformation.33 Because the collapsing and folding steps require additives of different hydrophobicity, we hypothesize that protein folding can be optimized in a dynamically responsive environment that favors both collapsing and rearrangement. In this work, we investigate protein folding in the presence of a polymer that varies its hydrophobicity during the folding process. On the basis of Langevin dynamics simulations and standard coarse-grained models for both the protein and polymer, we find that dynamic control of the polymer hydrophobicity may lead to a smoothed free-energy landscape that accelerates the rates of both protein collapsing and rearrangement. The advantage of dynamic control is also demonstrated with experiments for refolding of lyzoyme in an aqueous solution of poly(N-isopropylacrylamide-co-N-tertbutylacrylamide) (PIPTB).

Protein and Polymer Models. To simulate protein folding in the presence of polymers, we follow an off-lattice model of an all β-sheet protein proposed by Honeycutt and Thirumalai (HT),34 which is widely used to unveil the physical nature of protein both folding and aggregating.35,36 The HT model consists of 46 amino acid residues that are hydrophobic (B), hydrophilic (L), or neutral (N). All residues are treated as spherical beads of equal size, tangentially connected in a linear sequence B9N3(LB)4N3B9N3(LB)5L. The native structure of model protein is β-barrellike, as shown in Figure S-1(a) (Supporting Information). Throughout this work, all energies are reported in units of h, which is the unit energy of hydrophobic interaction, and all lengths are in units of the bond length σ, which is the equilibrium length of the connective beads in protein. The forces between nonbonded monomers are sequence-dependent and represented by a Lennard-Jones potential. As reported elsewhere,41 the Hamiltonian of the model protein includes the bond energy (Vb,1), the bond-angle energy (Vθ,1), the torsion energy (Vφ,1), and the intramolecular hydrophobic energy (VLJ,1). In the absence of polymer, the collapse temperature (Tc) and folding temperature Tf of the model protein are 0.54h/kB and 0.49h/ kB, respectively.37 The details of model protein, including the potentials, are explained in the Supporting Information. The polymer with tunable hydrophobicity is represented by a linear chain of identical spherical beads with the bond lengths and bond angles enforced by the harmonic potentials. We assume that the diameter of each bead is the same as that of the peptides and the equilibrium bond length is σ. An attractive potential is used to describe the hydrophobic interaction between nonbonded polymer segments. A dimensionless parameter 2 is used to define the degree of hydrophobicity for the polymer. A positive 2 means that there is a hydrophobic attraction between polymer segments. The hydrophobic attraction between an amino acid bead of the model protein and a polymer segment

* Corresponding author. Telephone: 86-10-6277 9876. Fax: 86-10-6277 0304. E-mail: [email protected]. † Tsinghua University ‡ University of California.

10.1021/jp076043k CCC: $37.00 © 2007 American Chemical Society Published on Web 10/03/2007

12304 J. Phys. Chem. B, Vol. 111, No. 42, 2007 is described by the interaction parameter 12, which is calculated from the Lorentz-Berthelot mixing rule; i.e., 12 ) x12. Langevin Dynamics Simulations. Simulation at constant temperature was performed by using the Andersen thermostat method, in which all particles were subjected to random collisions with “ghost” particles. The post-event velocity of each particle colliding with a “ghost” particle was chosen randomly from the Maxwell-Boltzmann distribution. Periodic boundary conditions with the minimum image convention were used to minimize the boundary effects. Each system contains one protein molecule and one polymer molecule started from a random coil configuration equilibrated at high temperature (T ) 1.5h/kB) over a period of 1 × 105τ, where

the time unit τ ≡ xmσ2/h; m was the mass of each amino acid and was the same for each polymer segment. The protein was allowed to fold by reducing the temperature to T ) 0.5h/kB, which was near the value of Tf. To obtain sufficient statistics, 1560 simulations from different initial configurations were performed for each set of simulation conditions. To confirm the validity, additionally 520 simulations from different initial configurations were launched to ensure that the quantity reported from the simulation was reproducible with an error range of 10%. The simulation is detailed in the Supporting Information. Experiments. The experimental details including the synthesis and characterization of PIPTB, refolding operation, activity assay, and fluorescence emission spectra of refolded lysozyme are provided in the Supporting Information. In brief, 1 M NIPAAm, 0.1 M NTBAAm, and 0.01 M AIBN were dissolved in 20 mL of ethyl alcohol. Polymerization was performed under nitrogen atmosphere at 60 °C for 24 h. The polymer was purified by repeated precipitation with n-hexane and characterized by FT-IR and NMR. The lowest critical solution temperature (LCST) of PIPTB was determined as 27.1 °C. Refolding of the urea-denatured lysozyme was performed by quick dilution with pH 8.2, 0.1 M Tris‚HCl containing 1 mM GSSG, 1 mM GSH, 2 M urea, and PIPTB of a given concentration. Then the solution was slowly stirred constantly and thermostated at specified temperature or over a temperature gradient. Results and Discussion

Hydrophobic Polymer Accelerates Protein Collapse but Hinders Folding. We investigate first protein folding in a dilute solution and in the presence of a polymer with fixed hydrophobicity by using Langevin dynamics simulations. The simulation results provide a useful benchmark for studying the efficiency of dynamic control. Figure 1a shows the folding trajectory for a model protein (see Methods) in terms of the radius of gyration (Rg), the structure overlap function (χ), and the intramolecular potential energy (VLJ,11) without any polymer additives. The snapshots illustrate representative configurations of the folding intermediates. As indicated in previous studies,37,38 protein folding undergoes two stages: (1) collapse of the random coil state into a molten globular state and (2) structural rearrangement leading to the native configuration. Protein collapsing is manifested by reduction of Rg below 5.0σ for the first time, occurring at approximately t2 ) 945τ, while the structural rearrangement is characterized by reduction of χ at t5 ) 2735τ. The structural transitions are also reflected in the intramolecular hydrophobic energy VLJ,11, which falls significantly once rearrangement occurs, i.e., the protein starts to fold to its native structure.

Lu et al. Figure 1b shows the folding trajectory of the same protein in the presence of a weakly hydrophobic polymer with L2 ) 30 coarse-grained segments. Similar to that shown in Figure 1a, the protein folds in two steps: collapsing at t2 ) 85τ and rearrangement at t5 ) 3140τ. The polymer accelerates protein collapsing, as evidenced in the significant reduction of the collapsing time from 945τ to 85τ, but it hinders rearrangement, as shown by extension of the rearranging time from 1790τ to 3055τ. The protein-polymer interaction energy VLJ,12 shows little change during the rearrangement step, suggesting a loose binding of the polymer with the collapsed protein. The weak interaction between the polymer and folded protein is in line with recent simulations for cageless-chaperone-assisted protein folding reported by Jewett et al.,39,40 and with experiments for protein folding assisted by weakly hydrophobic polymers.16-24 The snapshots suggest that the polymer first binds to the denatured protein (at t2 ) 85τ) and then it is released from the key protein core (at t4 ) 3070τ), resulting in a weakened hydrophobic interaction and thereby enabling protein structural transition (from t4 ) 3071τ to t5 ) 3140τ). A comparison of Figure 1a and Figure 1b suggests that a hydrophobic polymer favors protein collapsing on the one hand but hinders rearrangement to form the native conformation on the other. Thus a careful choice of the polymer hydrophobicity is pivotal in assisting protein folding. Polymer of Moderate Hydrophobicity Enhances Protein Folding Yield. Addition of a hydrophobic polymer affects not only the folding kinetics but also the recovery yield. To see such effects, we simulated protein folding with various polymers of different hydrophobicities at fixed temperature (T ) 0.50h/ kB). Here y0 and tf0 correspond to the folding yield and average folding time, respectively, in the absence of polymer at T ) 0.50h/kB. Figure 2a shows the relative folding yield (y/y0) and average folding time (tf/tf0) for different polymers. An optimal hydrophobicity of the polymer (2 ) 0.005) can be identified to improve both the folding yield and kinetics. In parallel with the simulations, we studied lysozyme refolding in a dilute solution and in an aqueous solution of poly(Nisopropylacrylamide-co-N-tert-butylacrylamide) (PIPTB) at different temperatures. Figure 2b shows the relative folding yield ((y/y0) and the relative folding time (tf/tf0). In this case, y0 and tf0 are the folding yield and average folding time of lysozyme in the absence of PIPTB at different temperatures. The results are averaged from three independent experiments. As suggested by the simulation results, an optimal temperature (T ) 30 °C), which is equivalent to an optimal hydrophobicity of PIPTB, can be identified for both the folding yield and kinetics. Protein Folding in a Dynamically Responsive Environment. Because protein collapsing prefers a polymer of strong hydrophobicity but folding into the native conformation requires only a polymer of weak hydrophobicity,38 we expect that a polymer of tunable hydrophobicity during the folding process will be ideal for promoting the folding kinetics and yield. By tuning the polymer hydrophobicity to enhance both collapsing and rearrangement, we are able to mimic the in vivo folding process in which the molecular chaperone (e.g., GroEL/GroES/ ATP) tunes its inner surface property from strongly hydrophobic to weakly hydrophobic to first capture an unfolded protein, then assist protein structural transition, and finally discharge the folded protein.41-43 Figure 3 shows the folding trajectories of the same model protein as that studied in Figures 1 and 2 in the presence of the model polymer of which the hydrophobicity was initially set at 2 ) 0.025 for the first 100τ simulation steps, then linearly

Dynamic Control of Protein Folding Pathway

J. Phys. Chem. B, Vol. 111, No. 42, 2007 12305

Figure 1. Snapshots of the intermediate states and folding trajectories of the model protein at T ) 0.50h/kB. (a) Folding in the absence of polymer. (b) Folding in the presence of a hydrophobic polymer with the concentration ratio of polymer to protein C2:C1 ) 1:1, the polymer energy parameter 2 ) 0.005, and the polymer chain length L2 ) 30.

decreased from 0.010 to 0.00 within the next 4000τ steps, and maintained at 2 ) 0.00 for an additional 1000τ steps. The protein collapses quickly from a random coil state (t1 ) 0) to a molten globular state (t2 ) 40τ) as the polymer entangles around the protein hydrophobic core via hydrophobic attractions, accompanied by a reduction of Rg,1 (lower than 5.0σ) and by a high value of χ (higher than 0.7). Here, as stated earlier, a strongly hydrophobic polymer favors protein collapsing. After 100τ simulation steps, 2 drops to 0.01, which promotes relaxation of the collapsed structure (see the snapshot at t3 ) 115τ). As the folding proceeds (from t3 and t5), the structure overlap function (χ) decreases slightly and the polymer is released from the hydrophobic core of the folding protein. Upon formation of the key folding intermediates (at t5 ) 1000τ), the protein quickly folds into the native conformation, characterized by a sharp reduction of χ and by redistribution of the polymer around the folded protein. The snapshot at t6 ) 1165τ illustrates

a representative configuration of the protein-polymer complex. As expected, we observe a strong hydrophobic interaction between the protein and polymer at the beginning stage of folding but the hydrophobic attraction decreases progressively to promote the structural rearrangement. With a decreasing hydrophobicity, the polymer facilitates protein collapsing at the early stage while it promotes structural rearrangement at the late stage. The protein collapse is accelerated by formation of protein-polymer complex, thereby suppressing the hydrophobic interactions/aggregation among proteins. On the other hand, the reduced hydrophobic interaction at the late stage of folding favors the reversible formation/dissociation of polymer-protein complexes and provides efficient kinetic paths to the native conformation. Figure 4a,b shows the simulation results for the kinetics of folding in the presence of polymer with different hydrophobicity gradients; i.e., the polymer hydrophobicity (2) falls linearly from

12306 J. Phys. Chem. B, Vol. 111, No. 42, 2007

Figure 2. Relative folding yield (y/y0) and folding time (tf/tf0) as a function of temperature (T). (a) Molecular simulation results. Here y0 and tf0 correspond to the folding yield and average folding time, respectively, in the absence of polymer at T ) 0.50h/kB. (b) Experimental data. In this case, y0 and tf0 are the folding yield and average folding time of lysozyme in the absence of PIPTB at different temperatures. The results are averaged from three independent experiments.

different initial values to 0.005 within the first 1000τ simulation steps and then is maintained at 0.005 until the end of the folding. We studied the dependences of the relative folding yield (y/y0) and folding time (tf/tf0) on the initial polymer hydrophobicity. We find that both the folding yield and kinetics vary nonmonotonically with the hydrophobicity gradients. An appropriate hydrophobicity gradient (e.g., from 2,i ) 0.050 to 2,f ) 0.005) accelerates protein folding in comparison with that at fixed hydrophobicity (2,i ) 0.050). A stronger variation of the polymer hydrophobicity (from 2,i ) 0.150 to 2,f ) 0.050), however, slows the folding rate. An optimal initial hydrophobicity (2,i ) 0.050) can be identified that results in an improved folding yield and kinetics in comparison with those in the absence of polymer and those in the presence of polymer with fixed hydrophobicity (2 ) 0.005). Qualitatively, the simulation results appear in good agreement with recent experiments indicating that, in refolding of mitochondrial malate dehydrogenase (mMDH) with cross-linked thermally responsive hydrogels, efficient folding can be achieved by a temperature swing procedure that makes the gel alternate between hydrophobic and hydrophilic states.44 The change of polymer hydrophobicity can be realized by application of a temperature gradient to a thermally responsive polymer, such as PIPTB. Above the low critical solution temperature (LCST, 27.1 °C for PIPTB), the polymer is strongly hydrophobic, but below LCST, it turns weakly hydrophobic. In our experiments, lysozyme folding in the presence of PIPTB was initiated at different temperatures that linearly decreased to 10 °C. The temperature of the water bath was automatically monitored by a temperature recorder. Parts c and d of Figure 4 show the folding kinetic curves under different temperature

Lu et al. gradients and the temperature dependences of the relative folding yield (y/y0) and the relative folding time (tf/tf0), respectively. Figure 4c indicates that lysozyme refolding is accelerated under the temperature gradient from 40 to 10 °C. Under the temperature gradient from 80 to 10 °C, the folding is suppressed first and accelerated once the temperature is lower than 70 °C. Figure 4d shows that both folding yield and folding time depend on the initial temperature or, equivalently, the initial hydrophobicity of the polymer. When the temperature drops from 40 to 10 °C during the folding process, the relative folding yield is around 1.3 times that if the temperature is fixed at 10 °C. Meanwhile, the folding time is reduced by a factor of 0.4 in comparison with that in the presence of PIPTB at 10 °C. Figure 4c,d confirms that a dynamic control of the polymer hydrophobicity can accelerate protein folding and achieve a higher folding yield, compared to those obtained in the presence of polymer under constant temperature. We have also tested the dynamic control for refolding lysozyme in a declining temperature environment, in which dextran-grafted PNIPAAm (DGP) was synthesized and used as an “artificial chaperone”. The refolding yield of lysozyme at an experimentally optimized temperature gradient is significantly higher than the counterpart obtained at a constant temperature.45 Polymer Alters the Free-Energy Landscape of Protein Folding. The kinetics of protein is determined by the free-energy landscape, which can be determined by parallel simulations (1560 in this work). Figure 5 shows the free energy of the model protein as a function of the structural overlap function χ and the radius of gyration Rg,1. Figure 5a gives the folding free-energy map without the polymer, which has a characteristic L-shaped structure reflecting the two-step folding mechanism, i.e., collapsing and rearrangement. Starting from Rg,1 > 10.0σ and χ > 0.90, the protein collapses until Rg,1 ) 4.50σ-7.50σ and χ ) 0.55-0.75 and then the protein structure rearranges to the native state until the end of the folding process. Three main basins appear on the free-energy landscape, corresponding to the random coil state, the molten globular state, and the native state, respectively. The three basins are separated by two major barriers, requiring external energy to promote protein structure transitions from the local energy minimum state to the global energy minimum state, i.e., the native conformation. Figure 5b shows the folding free-energy landscape in the presence of polymer with fixed hydrophobicity (2 ) 0.005). Again the free-energy landscape exhibits an L-shaped structure, suggesting that the two-step folding mechanism is preserved. In comparison with that shown in Figure 5a, Figure 5b has only two main basins, corresponding to the molten globular state and native state, respectively. Disappearance of the basin for the random coil conformations is due to the fact that the polymer accelerates protein collapsing. Figure 5b also shows the conformation of typical partially folded protein-polymer complex, where polymer entangles around the hydrophobic core of partially folded protein and thereby stabilizes folding intermediates. Acceleration of the protein collapsing is introduced by the lowered energy barrier between the random coil state and the collapsed state. However, the polymer hinders protein rearrangement to form the native conformation because the polymer must be released from the protein core. Indeed, the basin for the molten globular state is deepened by the polymer, indicating an intensified energy barrier between the molten globular state and the native state. As a result, the polymer promotes the protein collapse but retards the structural rearrangement.

Dynamic Control of Protein Folding Pathway

J. Phys. Chem. B, Vol. 111, No. 42, 2007 12307

Figure 3. Folding trajectory of the model protein in the presence of a polymer with tunable hydrophobicity. As in Figure 1, the concentration ratio of polymer to protein is C2:C1 ) 1:1; the polymer consists of L2 ) 30 segments. The folding was conducted for 100τ at 2 ) 0.025, then for 4000τ at a linear decrease of 2 from 0.010 to 0.00, and finally for 1000τ at 2 of 0.00.

Figure 4. (a) Protein folding yields in the presence of polymers that vary in hydrophobicity during the course of folding. Here the temperature is fixed at T ) 0.50h/kB, and the polymer concentration and chain length are the same as those in Figure 1. The different hydrophobic gradients of the polymer are created by linearly decreasing 2 from different initial values to 2,f ) 0.005 within 1000τ simulation steps, and then fixing at 2 ) 0.005 for additional 4000τ steps. (b) Relative folding yield (y/y0) and average folding time (tf/tf0) at different strengths of initial polymer hydrophobicity, 2,i. Here T ) 0.50h/kB and y0 and tf0 are the folding yield and average folding time of the model protein in the presence of polymer with fixed hydrophobicity (C2:C1 ) 1:1, 2 ) 0.005, L2 ) 30). (c) Folding yield of lysozyme in 10 mg/mL PIPTB under different temperature gradients. Folding conditions: 0.1 M Tris‚HCl, pH 8.2, 1.0 mg/mL lysozyme, 2.0 M urea, 1 mM GSSG, 1 mM GSH, and 10 mg/mL PIPTB. The gradient from high temperature to low temperature indicates the polymer from strongly hydrophobic to weakly hydrophobic as shown by the lines without symbols. (d) Relative folding yield (y/y0) and relative folding time (tf/tf0) as a function of different initial temperatures. Here y0 and tf0 are the folding yield and average folding time of model protein in the presence of 10 mg/mL polymer at 10 °C.

Figure 5c displays the folding free-energy map in the presence of polymer with tunable hydrophobicity. The folding was conducted for a duration of 100τ at 2 ) 0.025, followed by a duration of 4000τ with a linear decrease of 2 from 0.010 to 0.00, and completed for another 1000τ simulation steps at 2 ) 0.00. Again the free-energy landscape exhibits two main basins

at Rg,1 ) 4.50σ-7.0σ, χ ) 0.55-0.75, and at Rg,1 ≈ 4.0σ, χ ) 0.20-0.50, similar to those shown in Figure 5b. However, the energy barrier between these two basins becomes much shallower, which explains a favorable condition for the structural rearrangement. Figure 5c also shows two typical conformations of the protein at intermediate and folded state, respectively.

12308 J. Phys. Chem. B, Vol. 111, No. 42, 2007

Lu et al.

Figure 5. Free-energy surfaces for folding of the model protein (a) in a dilute solution, (b) in the presence of a polymer of constant hydrophobicity (2 ) 0.005) at C2:C1 ) 1:1 and L2 ) 30, and (c) in the presence of a polymer at C2:C1 ) 1:1 and L2 ) 30 but with a linearly declining hydrophobicity during the folding process; i.e., the simulation was conducted for 100τ at 2 ) 0.025, then for 4000τ at a linear decrease of 2 from 0.010 to 0.00, and finally for 1000τ at 2 ) 0.00. In all cases, the temperature is fixed at T ) 0.50h/kB. The blue regions correspond to configurations with the highest occupation probability or minimum free energy.

Formation of the partially folded protein-polymer complex accelerates protein collapsing via the lower energy barrier between the random coil state and the collapsed state, and release of the polymer with reduced hydrophobicity favors protein structural rearrangement to form the native conformation. As a result, the presence of a polymer with dynamically tunable hydrophobicity favors both the collapsing and the rearrangement and thus accelerates protein folding and enhances folding yield. Conclusion Both Langevin dynamics simulations for folding of a model protein and experimental refolding of lysozyme indicate that the protein folding can be drastically enhanced in a dynamically responsive environment that favors both collapsing and rearrangement. The change of polymer hydrophobicity during the folding process mimics chaperone-assisted protein folding in vivo where a newly synthesized protein collapses in a strongly hydrophobic environment and rearranges its structure when the chaperone reduces its interior hydrophobicity upon energy input in terms of adenosine triphosphate (ATP). In the presence of

polymer, the protein folding also follows the two-step mechanism that entails collapsing and structural rearrangement. The folding trajectory shows that a polymer of strong hydrophobicity enhances the protein collapse via assembling around the denatured protein, but it must dissociate from the complex to form the folded conformation. The free-energy landscape of the folding process illustrates that the hydrophobic polymer reduces the free-energy barrier between the random coil and molten globular states, leading to an accelerated collapsing, but if the hydrophobic interaction between the polymer and protein is too strong, the polymer may hinder the conformational rearrangement. By decreasing the hydrophobic interaction between the protein and polymer molecules progressively, both folding kinetics and yield can be enhanced significantly. Acknowledgment. This work was supported by the National Natural Science Foundation under Grant 20576061 and by the Ministry of Science and Technology through the 973 Project under Grant 2003CB716004.

Dynamic Control of Protein Folding Pathway Supporting Information Available: List of simulation details, including protein model, polymer model, and details of simulation methods, and experimental details, including materials, refolding protocols, and structure and activity analysis. In addition, some related experimental results are also included. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Georgiou, G. AIChE J. 1988, 34, 1233. (2) Buchner, J.; Rudolph, R. Bio/Technology 1991, 9, 157. (3) Guise, A. D.; West, S. M.; Chaudhuri, J. B. Mol. Biotechnol. 1996, 6, 53. (4) Bratko, D.; Cellmer, T.; Prausnitz, J. M.; Blanch, H. W. J. Am. Chem. Soc. 2006, 128, 1683. (5) Wang, J.; Lu, D. N.; Lin, Y.; Liu, Z. Biochem. Eng. J. 2005, 24, 269. (6) Xu, Q.; Keiderling, T. A. Protein Sci. 2004, 13, 2949. (7) Wetlaufer, D. B.; Xie, Y. Protein Sci. 1995, 4, 1535. (8) Zardeneta, G.; Horowitz, P. M. Anal. Biochem. 1994, 223, 1. (9) Rozema, D.; Gellman, S. H. Biochemistry 1996, 35, 15760. (10) Rozema, D.; Gellman, S. H. J. Biol. Chem. 1996, 271, 3478. (11) Rozema, D.; Gellman, S. H. J. Am. Chem. Soc. 1995, 117, 2373. (12) Karuppiah, N.; Sharma, A. Biochem. Biophys. Res. Commun. 1995, 211, 60. (13) Machida, S.; Ogawa, S.; Shi, X. H.; Takaha, T.; Fujii, K.; Hayashi, K. FEBS Lett. 2000, 486, 131. (14) Meng, F. G.; Park, Y. D.; Zhou, H. M. Int. J. Biochem. Cell Biol. 2001, 33, 701. (15) Rariy, R. V.; Klibanov, A. M. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 13520. (16) Lu, D.; Liu, Z. X.; Zhang, M. L.; Liu, Z.; Zhou, H. M. Biochem. Eng. J. 2005, 24, 55. (17) Lu, D. N.; Liu, Z. Acta Polym. Sin. 2004, 573. (18) Lu, D. N.; Zhang, K.; Liu, Z. Biochem. Eng. J. 2005, 25, 141. (19) Cui, Z. F.; Guan, Y. X.; Chen, J. L.; Yao, S. J. J. Appl. Polym. Sci. 2005, 96, 1734. (20) Cui, Z. F.; Guan, Y. X.; Yao, S. J. Chin. J. Chem. Eng. 2004, 12, 556.

J. Phys. Chem. B, Vol. 111, No. 42, 2007 12309 (21) Gil, E. S.; Hudson, S. A. Prog. Polym. Sci. 2004, 29, 1173. (22) Hirakura, T.; Nomura, Y.; Aoyama, Y.; Akiyoshi, K. Biomacromolecules 2004, 5, 1804. (23) Roy, I.; Gupta, M. N. Protein Eng. 2003, 16, 1153. (24) Roy, I.; Mondal, K.; Sharma, A.; Gupta, M. N. Biochim. Biophys. Acta: Proteins Proteomics 2005, 1747, 179. (25) Cleland, J. L.; Hedgepeth, C.; Wang, D. I. C. J. Biol. Chem. 1992, 267, 13327. (26) Cleland, J. L.; Wang, D. I. C. Abstr. Pap. Am. Chem. Soc. 1990, 200, 87. (27) Cleland, J. L.; Wang, D. I. C. Biochemistry 1990, 29, 11072. (28) Miller, R.; Danko, C. A.; Fasolka, M. J.; Balazs, A. C.; Chan, H. S.; Dill, K. A. J. Chem. Phys. 1992, 96, 768. (29) Dill, K. A. Biochemistry 1990, 29, 7133. (30) Lau, K. F.; Dill, K. A. Macromolecules 1989, 22, 3986. (31) Lu, D. N.; Liu, Z.; Liu, Z. X.; Zhang, M. L.; Ouyang, P. J. Chem. Phys. 2005, 122, Art. No. 134902. (32) Lu, D. N.; Liu, Z. J. Chem. Phys. 2005, 123, Art. No. 134903. (33) Lu, D. N.; Liu, Z.; Wu, J. Z. J. Chem. Phys. 2007, 126, Art. No. 064906. (34) Honeycutt, J. D.; Thirumalai, D. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 3526. (35) Bratko, D.; Cellmer, T.; Prausnitz, J. M.; Blanch, H. W. Biotechnol. Bioeng. 2007, 96, 1. (36) Cellmer, T.; Bratko, D.; Prausnitz, J. M.; Blanch, H. Biotechnol. Bioeng. 2005, 89, 78. (37) Lu, D. N.; Liu, Z.; Wu, J. Z. Biophys. J. 2006, 90, 3224. (38) Lu, D. N.; Liu, Z.; Wu, J. Z. J. Chem. Phys. 2007, 126, Art. No. 064906. (39) Jewett, A. I.; Baumketner, A.; Shea, J. E. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 13192. (40) Jewett, A. I.; Shea, J.-E. J. Mol. Biol. 2006, 363, 945. (41) Fenton, W. A.; Kashi, Y.; Furtak, K.; Horwich, A. L. Nature 1994, 371, 614. (42) Martin, J.; Mayhew, M.; Langer, T.; Hartl, F. U. Nature 1993, 366, 228. (43) Rye, H. S.; Burston, S. G.; Fenton, W. A.; Beechem, J. M.; Xu, Z. H.; Sigler, P. B.; Horwich, A. L. Nature 1997, 388, 792. (44) Jones, H.; Dalmaris, J.; Wright, M.; Steinke, J. H. G.; Miller, A. D. Org. Biomol. Chem. 2006, 4, 2568. (45) Lu, D. N.; Liu, Z. X.; Zhang, M. L.; Wang, X. G.; Liu, Z. Biochem. Eng. J. 2006, 27, 336.