pubs.acs.org/Langmuir © 2009 American Chemical Society
Protein-Solid Interactions: Important Role of Solvent, Ions, Temperature, and Buffer in Protein Binding to r-Zr(IV) Phosphate Michael R. Duff, Jr. and Challa V. Kumar* Department of Chemistry, University of Connecticut, Storrs, Connecticut 06269-3060 Received May 27, 2009. Revised Manuscript Received June 27, 2009 The interaction of proteins with a solid surface involves a complex set of interactions, and elucidating the details of these interactions is essential in the rational design of solid surfaces for applications in biosensors, biocatalysis, and biomedical applications. We examined the enthalpy changes accompanying the binding of met-hemoglobin, met-myoglobin, and lysozyme to layered alpha-Zr(IV)phosphate (20 mM NaPipes, 1 mM TBA, pH 7.2, 298 K) by titration calorimetry, under specific conditions. The corresponding binding enthalpies for the three proteins are -24.2 ( 2.2, -10.6 ( 2, and 6.2 ( 0.2 kcal/mol, respectively. The binding enthalpy depended on the charge of the protein where the binding of positively charged proteins to the negatively charged solid surface was endothermic while the binding of negatively charged proteins to the negatively charged solid was exothermic. These observations are contrary to a simple electrostatic model where binding to the oppositely charged surface is expected to be exothermic. The binding enthalpy depended on the net charge on the protein, ionic strength of the medium, the type of buffer ions present, and temperature. The temperature dependence studies of binding enthalpies resulted in the estimation of heat capacity changes accompanying the binding. The heat capacity changes observed with Hb, Mb, and lysozyme are 1.4 ( 0.3, 0.89 ( 0.2, and 0.74 ( 0.1 kcal/(mol 3 K), respectively, and these values depended on the net charge of the protein. The enthalpy changes also depended linearly on the enthalpy of ionization of the buffer, and the numbers of protons released per protein estimated from this data are 12.6 ( 2, 6.0 ( 1.2, and 1.2 ( 0.5 for Hb, Mb, and lysozyme, respectively. Binding enthalpies, independent of buffer ionization, are also estimated from these data. Entropy changes are related to the loss in the degrees of freedom when the protein binds to the solid and the displacement of solvent molecules/protons/ions from the protein-solid interface. Proton coupled protein binding is one of the major processes in these systems, which is novel, and the binding enthalpies can be predicted from the net charge of the protein, enthalpy of buffer ionization, ionic strength, and temperature.
Introduction Solid-bound enzymes perform key functions in biosensors, biocatalysis, biomedical devices, certain implants, and protein arrays,1 and controlling the bound enzyme behavior is essential for these applications. For example, retention of native-like structure, long shelf life, and a high degree of biological activity of the bound enzyme are highly desired. Controlling these attributes in a desired manner continues to be a challenge, and there are no specific rules or guidelines as to how this can be achieved for a given enzyme and a solid support.2 One hypothesis is that the behavior of the bound enzyme is controlled by its interactions with the underlying solid support at the molecular level, and stabilization of the native state while severely destabilizing the denatured state (Chart 1) increases the overall thermodynamic stability of the bound enzyme.2 Enhanced stability of the enzyme is expected to increase its shelf life, operational lifetime, and degree of retention of its structure/activity, all of which are desired. However, our understanding of protein-solid interactions is still primitive, even though several cases of enhanced stability and activity of the bound enzymes are reported.2 Our approach to gain insight into this problem has been to systematically evaluate the magnitudes of specific interactions of (1) Fessner, W. D. Biocatalysts - From Discovery to Application. Springer: Berlin, 1999. (2) Mudhivarthi, V. K.; Bhambhani, A.; Kumar, C. V. Dalton Trans. 2007, 5483–5497. (3) (a) Chaudhari, A.; Thota, J.; Kumar, C. V. Microporous Mesoporous Mater. 2004, 75, 281–291. (b) Kumar, C. V.; Chaudhari, A. J. Am. Chem. Soc. 2000, 122, 830– 837.
Langmuir 2009, 25(21), 12635–12643
enzymes with selected supports by calorimetric and spectroscopic methods and correlate the thermodynamic parameters with the intrinsic properties of the corresponding enzymes.3 We also recognize that the solvent, electrolytes, and buffer also play a critical role in the binding process and influence the bound enzyme behavior. In this context, we attempted to systematically measure the thermodynamic parameters for the binding of a small set of enzymes to an inorganic solid support, under a controlled set of conditions, by isothermal titration calorimetry (ITC). Zirconium phosphate, R-Zr(IV)(HPO4)2 3 H2O (abbreviated as R-ZrP, Chart 2), has been chosen for these studies because it is chemically and topologically homogeneous, rigid, inert, and stable ( 0) while similarly charged partners be endothermic (ΔH < 0) (Table 1). Note that binding of positively charged lysozyme to negatively charged R-ZrP is unfavorable with respect to the enthalpy change. Therefore, a simple electrostatic model cannot account for the sign of ΔH, and hence, the solvent, cations and buffer ions may play an important role in the binding process. The above ITC data with Hb were fitted using a single-site binding model, and the binding parameters have been extracted from these fits (Figure 1B, red line). Data from multiple titrations (same buffer) were analyzed and averaged over several measurements, and these indicated binding parameters for Hb to be Kb = 2.4 ( 0.3 106 M-1, ΔH = -24 ( 2 kcal/mol, ΔS = -48 ( 7.5 cal/(mol K), and binding site size (1/n) of ∼450 phosphates per protein (Table 1). The Kb of Hb from ITC matched well from that obtained by centrifugation studies.3 Binding of all three proteins have been examined under similar conditions for the purposes of a quantitative comparison. The ITC data for Mb and lysozyme were also analyzed quantitatively, and Mb data has been fitted with a single-site binding model. The corresponding binding parameters are collected in Table 1. The binding constants obtained for Mb from the above data are higher than those determined from centrifugation studies3b (0.2 106 M-1 from centrifugation data versus 12638 DOI: 10.1021/la901901k
3.1 106 M-1 for ITC). This difference is partly due to the fact that the buffers used in the two experiments are different. In centrifugation studies, 10 mM K2HPO4 pH 7.2 was used, while the ITC experiment was done with 20 mM NaPipes, 1 mM TBA pH 7.2. Differing ionic strengths and/or buffer compositions could be some of the reasons for the disagreements between the two experiments. The best fits for lysozyme data required a two-site model, and one of the two binding constants of lysozyme agreed well with that from the centrifugation method. Current studies show that the ITC method is more sensitive than the centrifugation method, because it can distinguish between the two binding modes of lysozyme, while the centrifugation method was not sensitive enough to detect the two distinct binding modes. Note that the best fits to the ITC data also produced estimates of the corresponding ΔH, ΔS, and ΔG values for the binding, which are collected in Table 1. The ΔH values were allowed to float to obtain the best fits, and the data show that binding of Hb and Mb is enthalpy driven while lysozyme binding is entirely entropy driven. This small set of proteins also demonstrates that the binding free energy is not related to the overall charge of the protein and that binding is driven by multiple interactions. Since the value of ΔH obtained from the above analysis was model dependent, we also measured ΔH by a model independent method. Direct Measures of Binding Enthalpies. To avoid uncertainties associated with the use of a binding model in estimating the ΔH values from the ITC data, we estimated ΔH directly, using a model independent method. This was done by adding small portions of protein solution to moderately concentrated suspensions of R-ZrP, while monitoring the heat released or absorbed with each addition. After subtracting the corresponding heats of dilution and averaging the data over multiple measurements, the heats of binding have been estimated by the model-independent method and the corresponding data collected in Table 2. In the case of Hb binding to R-ZrP, the ΔH values obtained from the full titration (-24 ( 2 kcal/mol) and the modelindependent method (-24.2 ( 2.2 kcal/mol) agreed very well, while in the case of Mb, the two values differed to some extent (-14 ( 1 kcal/mol and -10.6 ( 2 kcal/mol, respectively), but within our experimental error. For lysozyme, the binding enthalpy obtained from the model-independent method (6.2 ( 0.2 kcal/mol) differed substantially from that determined by the full titration Langmuir 2009, 25(21), 12635–12643
Duff and Kumar
Article
Table 2. Model-Independent ΔH Values for the Binding of Proteins to r-ZrP at 298 K (20 mM NaPipes, 1 mM TBA pH 7.2) and Some of the Important Characteristics of the Proteins protein property
Hb
Mb
lysozyme
buffer ΔH (kcal/mol) pI charge at pH 7 hydrodynamic radius (A˚) mol. mass (g/mol) no. of residues
Pipes -24.2 ( 2.2 6.5 -8 32 64 500 427
Pipes -10.6 ( 2 6.8 -3.9 20.4 16 500 153
Pipes 6.2 ( 0.2 10.9 þ7 19.8 14 300 129
Figure 3. Effect of ionic strength on the molar binding enthalpies of Hb (100 μM, black circles), Mb (100 μM, red squares), and lysozyme (100 μM, blue triangles) with R-ZrP (1 mM) in NaPipes (5-40 mM Pipes, 1 mM TBA pH 7.2, at 298 K). In some cases, the error bars are too small to be visible.
Figure 2. Plots of molar binding enthalpies of Hb (100 μM, black circles), Mb (100 μM, red squares), and lysozyme (100 μM, blue triangles) with R-ZrP (1 mM), as a function of temperature (20 mM NaPipes, 1 mM TBA, pH 7.2). Error bars are as indicated, while some are too small to be visible in the plots.
method (2.3 ( 0.9 and 59 ( 38 kcal/mol). This latter outcome is understandable because the model independent method averages the individual contributions of multiple binding modes in a way consistent with their respective contributions, at low loadings. For example, even though the high affinity mode dominates at low protein loadings, the observed enthalpy will be a weighted average of multiple modes prevalent under those conditions. We used the direct method to measure ΔH in all subsequent studies, because the model independent method is accurate and it gives an appropriate average when multiple binding modes are involved. This method is also free of assumptions used to analyze the binding data, and it does not require the whole titration to be performed. Temperature Dependence of Binding Enthalpies. Useful information regarding the binding process can be obtained by measuring the heat capacity changes (ΔCp) accompanying the binding event, and ΔCp can be accurately estimated by measuring ΔH as a function of temperature. Therefore, we measured the binding enthalpies as a function of temperature from 278 to 318 K by the model independent method (1 mM R-ZrP, 100 μM protein, 20 mM Pipes buffer, 1 mM TBA). The binding enthalpies for the three proteins are plotted as a function of temperature (K), and all three plots indicated linear increases in enthalpy, with positive slopes (Figure 2). The data show that, in all three cases, the change in heat capacity accompanying the protein binding is positive, that is, the bound state has a higher heat capacity than the free components, and that the magnitude of ΔH decreases with increase in temperature. The binding enthalpy remains exothermic for Hb over this narrow range of temperatures, but those of Mb and lysozyme Langmuir 2009, 25(21), 12635–12643
undergo sign reversal. Linear least-squares fit to the enthalpy data according to eq 2 indicated ΔCp values of 1.4 ( 0.3, 0.89 ( 0.2, and 0.74 ( 0.1 kcal/(mol 3 K) for Hb, Mb, and lysozyme, respectively. Ionic Strength Dependence of the Binding Enthalpy. The role of electrostatics in the binding of proteins to R-ZrP was examined by measuring ΔH as a function of ionic strength. Since the proteins as well as the solid are electrostatically charged, ionic interactions are expected to contribute to the ΔH term. Increased ionic strength is expected to weaken these electrostatic interactions and, in turn, reduce the binding affinity/ enthalpy. Binding enthalpies were measured as a function of [Naþ] (540 mM Pipes buffer, pH 7.2, at constant protein concentration, temperature, and R-ZrP concentrations, at 298 K) by the model independent method. At sodium ion concentrations >50 mM, ΔH approached zero, for all the three proteins, while at low ion concentrations, Hb binding was strongly exothermic (Figure 3). Lysozyme, on the other hand, indicated a stronger endothermic binding at low ionic strengths. Linear fits to the data using the polyelectrolyte theory21 indicated slopes of 20.8 ( 2.9, 3.4 ( 0.6, and -3.9 ( 1.4 kcal/mol per ln([Naþ]) for Hb, Mb, and lysozyme, respectively. In this narrow range of ionic strengths, electrostatic interactions contributed substantially to binding enthalpy of Hb, while the electrostatic contributions are much smaller for Mb and lysozyme. This trend is surprising but indicated the stronger control of ΔH by negative charges when compared to the positive charges. Contributions of Buffer Protonation/Ionization to the Binding Enthalpy. Binding of a protein to a charged surface can alter the pKa values of ionizable functional groups at the protein-solid interface, and protons may be absorbed or released during the binding.18 Therefore, enthalpy of binding may depend on the enthalpy of protonation/ionization of the buffer, and the net number of protons absorbed or released during the binding has been estimated. This was done by measuring ΔH as a function of ionization enthalpies of selected buffers and the number of protons released have been calculated by using eq 3.20 While the ionization enthalpies of phosphate and acetate buffers are small, those of HEPES and MOPS are 4-5 times as (21) Manning, G. S. J. Chem. Phys. 1969, 51, 924–33.
DOI: 10.1021/la901901k
12639
Article
Duff and Kumar
Figure 4. Plot of binding enthalpy as a function of buffer ionization enthalpies. Binding enthalpies of Hb (100 μM, black circles), Mb (100 μM, red squares), and lysozyme (100 μM, blue triangles) to R-ZrP (1 mM) in 10 mM phosphate (closed circles), 20 mM Pipes, 90 mM HEPES, and 60 mM MOPS, and all contained 1 mM TBA pH 7.2, at 298 K. In some cases, the error bars are too small to be visible. We estimate that 12.6 ( 2, 6.0 ( 1.2, and 1.2 ( 0.5 protons are absorbed for each molecule of Hb, Mb, and lysozyme binding, respectively. The observed enthalpy change is smaller when buffer ionization enthalpy is greater.
high, at the same pH, temperature, and ionic strength. Therefore, these buffers provided a simple method to count the net number of protons that are released or absorbed when proteins bind to R-ZrP by simply measuring the binding enthalpies in phosphate, PIPES, HEPES, and MOPS buffers. The concentration of buffer used was adjusted to keep the concentration of sodium ions constant (30 mM) because enthalpies depended on sodium ion concentration (see above). Binding enthalpies of the three proteins, as measured by the model independent method (at constant concentration of sodium ions, pH, at 298 K) were plotted as a function of the enthalpy of buffer ionization. The binding enthalpies decreased in magnitude with increase in the buffer ionization enthalpy (Figure 4), and linear fits to the ΔH data using eq 3 indicated positive slopes with 12.6 ( 2, 6.0 ( 1.2, and 1.2 ( 0.5 protons absorbed for Hb, Mb, and lysozyme binding, respectively. Therefore, the data clearly show that protein binding is intimately connected to proton release from the protein-solid interface. The intrinsic binding enthalpies (buffer-independent enthalpy), determined from the y-intercept of the lines of Figure 4, are -70 ( 11, -25 ( 5, and -1.3 ( 0.5 kcal/mol for Hb, Mb, and lysozyme, respectively.
Discussion Elucidating how binding energetics (ΔH, ΔS, ΔG) are related to the molecular properties of the bound protein is important for a clearer understanding of the binding process. It could provide a rational approach to designing solid surfaces for optimal binding of enzymes and proteins. Controlling the binding enthalpies, for example, either by designing an appropriate solid surface or by adjusting the solution conditions, is important, because large negative enthalpy of binding may cause local heating and may cause partial denaturation of the bound protein. Similarly, if distortion of the protein structure improves binding and releases more heat by making additional contacts with the solid, then bound protein structure is more likely to be distorted and the resulting biocatalyst will be less stable/active. Therefore, binding enthalpies could have substantial effect on bound protein behavior, and they need to be 12640 DOI: 10.1021/la901901k
controlled by a rational approach. We systematically examined the binding enthalpies of three proteins under specific conditions, and ΔH data are discussed in terms of the intrinsic properties of the protein as well as the physical characteristics of the solution conditions. Binding Parameters. The ITC data for the binding of the three proteins show that binding of Hb and Mb to the solid can be analyzed using a single-binding site model while that of lysozyme requires a two-site model. All three proteins are known to intercalate in the galleries of the solid,3,9 and hence, the major binding mode for all three proteins is intercalation (Chart 3). Protein binding on the outside of the stacks of the solid is unlikely, because washing of the samples with buffer, after intercalation, did not remove the bound protein. Therefore, two binding modes of lysozyme are perhaps due to the intercalation of monomer vs dimer. Two strong pieces of evidence support this model. One is that lysozyme is known to aggregate at high protein concentrations22 and high loadings on the solid promote protein aggregation. The second is that the differential scanning calorimetric studies of lysozyme bound to R-ZrP indicated enzyme-enzyme interactions at high loadings but not at low loadings.6 Thus, two types of bound enzyme molecules can be identified in the case of lysozyme, thus justifying the use of a two-site model. All three proteins indicated moderate to high affinities (Kb = 2-150 106 M-1) for R-ZrP (Table 1), and the higher the positive charge on the protein, the greater the affinity. However, the observation that binding of negatively charged proteins (Hb, Mb) is exothermic while that of positively charged protein (lysozyme) is endothermic suggests that a simple electrostatic model cannot explain the binding thermodynamics. Therefore, we examined how the binding energetics, enthalpies, for example, depended on temperature, ionic strength, and the nature of the buffer ions to elucidate their contributions to the binding energetics. Binding Enthalpies. Binding enthalpies are estimated using two different methods; one is based on the titration method (model-dependent) and the other a model-independent, direct method. Values from both methods agreed reasonably well in the cases of Hb and Mb, whose binding can be described by the single-site model while differences are evident in the case of lysozyme (two-site model). Since the direct method requires no assumptions regarding the binding process, we decided to measure binding enthalpies, under controlled conditions, using this second approach. Data show that the binding enthalpies of negatively charged Hb (pI = 6.5) and Mb (pI = 6.8) and positively charged lysozyme (pI = 10.9) correlated strongly with the net charge on the protein (at pH 7.2, Figure 5). But there is no correlation between ΔH and the number of basic or acidic residues. The slope of the linear fit to the data in Figure 5 shows that ΔH is 1.8 kcal/mol per unit charge on the protein, and the sign of ΔH is opposite to that of the net charge on the protein. These observations can be explained by invoking the role of counterions in the binding process (Chart 5). Binding of the positively charged proteins to negatively charged R-ZrP would result in the release of counterions from the protein-solid interface (Chart 5, top row), and this process will be endothermic due to the release of ions into the solution. Binding of negatively charged proteins to R-ZrP, on the other hand, would require the binding of counterions at the proteinsolid interface for the neutralization of excess negative charge (Chart 5, bottom row). In support of the above model, the binding (22) Nesmelova, I. V.; Fedotov, V. D. Biochim. Biophys. Acta 1998, 1383, 311– 316.
Langmuir 2009, 25(21), 12635–12643
Duff and Kumar
Article Chart 5. Protein Binding to a Charged Solid Is Coupled to the Release or Binding of Counter Ionsa
a For example, binding of a positively charged protein to negatively charged R-ZrP is expected to release cations and anions from the interface where the protein serves as a counter ion to the solid (top, endothermic process). Binding of a negatively charged protein to negatively charged solid, on the other hand, requires the binding of counter ions at the interface to achieve neutralization of excess negative charge (bottom, exothermic process).
the sum of heat capacities of the corresponding individual components. ΔC p surfacearea ¼ ½0:32ðΔASAnon-polar Þ -0:14ðΔASApolar Þ
Figure 5. Strong correlation of binding enthalpies with net charge on the protein (298 K, 20 mM Pipes, pH 7.2, some error bars are too small to be visible in the plot).
of negatively charged proteins onto the surface of negatively charged sulfonated polystyrene was suggested to incorporate cations at the interface to eliminate electrostatic repulsion between the protein and the sorbent surface.23 Thus, binding or the release of counterions could control the enthalpy change to a significant extent, and hence, ΔH is expected to depend on the net charge on the protein. Therefore, we propose that the linear relationship between ΔH and protein charge, as well as the sign of ΔH, can be explained by invoking the role of counterions in protein binding process, as illustrated in Chart 5. Heat Capacity Changes. Solvent is also expected to play a vital role in the binding energetics, and this can be deduced from heat capacity changes. Plots of ΔH vs temperature for the three proteins were linear (Figure 2), and ΔH vs T plots indicated positive ΔCp values of 1.4 ( 0.3, 0.89 ( 0.2, and 0.74 ( 0.1 kcal/(mol 3 K) for Hb, Mb, and lysozyme, respectively. Thus, the heat capacities of the protein/solid complexes are higher than (23) Norde, W.; Lyklema, J. J. Colloid Interface Sci. 1978, 66, 285–294.
Langmuir 2009, 25(21), 12635–12643
ð4Þ
The observed ΔCp is often related to change in solvent-accessible surface area of the hydrophobic groups (ΔASAnonpolar) as well as those of polar groups (ΔASApolar) by eq 4.24 Positive ΔCp values are explained by the exposure of nonpolar groups to the solvent and/or burial of charged groups at the protein-solid interface. For example, binding of osteoprotein and statherin to hydroxyapatite indicated small but positive ΔCp values.25 To test this hypothesis, we plotted ΔCp as a function of specific properties of our proteins, and the ΔCp values correlated very well (R2 = 0.996) with the total number of positively charged groups on the protein surface (lysines and arginines)26 (Figure 6). Slope of the linear plot (ΔΔCp) indicated 0.09 kcal/(mol K) per positively charged surface residue. ΔCp did not correlate with the net charge on the protein, the total number of polar groups (acidic þ basic) on the surface, the number of surface hydrophobic groups, or the total surface area of the protein. We conclude from these analyses that positively charged side chains on the protein surface control heat capacity changes, and protein binding buries polar groups at the protein-solid interface. This conclusion is also consistent with our model in Chart 5, since such burial requires the release or absorption of cations from the interface. These conclusions will be further tested with a larger set of proteins, in future studies. Ionic Strength Dependence. Since charged residues are involved in the binding process, contributions of electrostatic interactions were factored out by measuring ΔH as a function of ionic strength. Magnitudes of binding enthalpies decreased with (24) Spolar, R.; Livingston, J. Biochem. 1992, 31, 3947–3955. (25) Goobes, G.; Goobes, R.; Shaw, W. J.; Gibson, J. M.; Long, J. R.; Raghunathan, V.; Schueler-Furman, O.; Popham, J. M.; Baker, D.; Campbell, C. T.; Stayton, P. S.; Drobny, G. P. Magn. Reson. Chem. 2007, 45, S32–S47. (26) (a) Perutz, M. F.; Fermi, G.; Poyart, C.; Pagnier, J.; Kister, J. J. Mol. Biol. 1993, 233, 536–545. (b) Maurus, R.; Overall, C. M.; Bogumil, R.; Luo, Y.; Mauk, A. G.; Smith, M.; Brayer, G. D. Biochim. Biophys. Acta 1997, 1341, 1–13. (c) Wang, J.; Dauter, M.; Alkire, R.; Joachimiak, A.; Dauter, Z. Acta Crystallogr., Sect. D. 2007, 63, 1254–1268.
DOI: 10.1021/la901901k
12641
Article
Figure 6. Correlation of ΔCp for protein binding to R-ZrP (R2 = 0.996) with the total number of basic residues (lysines þ arginines) on the protein surface. The slope indicated ΔΔCp of 0.09 kcal/(mol 3 K) per basic residue.
increasing ionic strength, which indicates that the electrostatic interactions are weakened, and the slope of the semilog plot (Figure 3) was the greatest (20.8 kcal/mol/ln([Naþ])) for Hb and the smallest (-3.9 kcal/mol/ln([Naþ]) for lysozyme (Figure 3). Such a decrease in enthalpy change with ionic strength was noted for the binding of a variety of DNA-binding proteins to DNA,27,28 and the decrease in the magnitudes of the enthalpies is consistent with the increased screening of the electrostatic interactions at higher ionic strengths. Slopes of the plots in Figure 3, ionic strength dependence of ΔH, scale linearly with the number of surface basic groups (lysines and arginines) but not well with the net charge (Suppl. Figure 3, Supporting Information) or the number of polar surface groups. Electrostatic contributions to ΔH/ln[Naþ] are controlled by the number of positively charged residues, and all three enthalpy plots in Figure 3 intersected near 70 mM ionic strength; however, the significance of this intersection is not clear. The ΔH1M values for Hb, Mb, and lysozyme, obtained by the extrapolation of these plots to 1 M sodium ion concentration, are 56, 5, -10 kcal/mol, respectively. There are significant differences between the intrinsic binding enthalpies of the three proteins, at 1 M NaCl, and enthalpy changes that are independent of ionic strength that are now established. These represent all other contributions to the binding enthalpy. Proton-Coupled Protein Binding. Binding of a charged protein to a charged surface alters the pKa of ionizable groups at the protein-solid interface, and hence, protein binding is expected to be coupled to proton release or absorption. Therefore, buffer ionization can also contribute to measured ΔH. These contributions were factored out by measuring ΔH as a function of ionization enthalpies of four different buffers (Figure 4, measured at constant ionic strength, temperature, and pH). In the cases of Hb and Mb, the observed ΔH decreased with increasing buffer ionization enthalpies, and the data clearly show that protein binding is coupled to proton absorption or release during the binding event. Indeed, when Hb or Mb bind to the solid, we observed that the solution pH is altered, and pH needs to be readjusted to the desired value. In the case of lysozyme, however, ΔH is nearly independent of the buffer used, and hence, the proton release/absorption is negligible. (27) Oda, M.; Furukawa, K.; Ogata, K.; Sarai, A.; Nakamura, H. J. Mol. Biol. 1998, 276, 571–590. (28) Milev, S.; Bosshard, H. R.; Jelesarov, I. Biochemistry 2005, 44, 285–293.
12642 DOI: 10.1021/la901901k
Duff and Kumar
The buffer-dependence data can be readily explained by the requirement that binding of negatively charged proteins to negatively charged surfaces requires charge neutralization at the interface, and this can be achieved either by the absorption of counterions (Chart 5) or protonation of polar groups at the protein-solid interface or by a suitable combination. Protonation at the interface would result in the deprotonation of the buffer ions and contributes to the observed enthalpy changes. In case of lysozyme, which is positively charged, such protonation is not required, and hence, the enthalpy change does not depend on the buffer enthalpy of ionization. In support of the above explanation, the binding of somatostatin peptide analogues (pKa of terminal amine is 7.2 and pH of the buffer was 7.4)29 and β-amyloid peptides to negatively charged lipid membranes was suggested to be coupled to proton release/absorption.30 Overall, the ΔH data show that ions, protons, solvent, and polar groups play a significant role in controlling binding enthalpies. Other weak contributors such as ZrP-ZrP interactions are expected to be minor, as the layers are held together by van der Waals interactions, and the plates are exfoliated prior to protein binding. Future studies will address how other thermodynamic properties of these systems are controlled by intrinsic properties of proteins.
Conclusions The thermodynamic characterization of protein binding to the layered inorganic solid, R-ZrP, is reported here. Thermodynamics of protein binding invlove ion release, proton absorption/release, dehydration/hydration of surface polar groups, and electrostatic interactions. The number of charged groups on the surface of the protein have the greatest correlation with protein binding thermodynamics. Binding enthalpies scale linearly with the net charge on the protein, while the heat capacity changes revealed the key role of polar groups on the protein surface. Observed binding enthalpies are also related to the ionization enthalpies of buffers used, and therefore, care needs to be exercised to this detail. The protein tries to eliminate electrostatic repulsion on binding to a charged surface by absorbing/releasing counterions and/or protons. Burial of the polar groups at the protein-solid interface and consequent release of solvent from these groups results in increased heat capacity of the system. Thus, a more detailed understanding of the interactions that take place upon protein binding to R-ZrP is obtained from the current studies, and these show that a simple electrostatic model (interaction of point charges) alone does not account for the observed energetics; additional factors such as ion/proton release, driven by electrostatic interactions, also need to be taken into account. The sign of ΔH depends on the sign of net charge on the protein. Heat capacity changes are dominated by the displacement of solvent from polar groups from the protein-solid interface. Current findings thus indicate the strong, active role of protein-solid interface on the binding energetics, and these conclusions need to be tested further, in future studies. By careful studies of the associated processes responsible for the binding of proteins to solid surfaces, better methods can be designed to facilitate or inhibit protein binding to solid surfaces, as desired. Improved understanding of protein binding may also aid in the design of rational methods to stabilize proteins on solid surfaces. Overall, this paper is an initial step in elucidating the details of protein-solid interactions, which is a complex process. (29) Seelig, J.; Nebel, S.; Ganz, P.; Bruns, C. Biochemistry 1993, 32, 9714–9721. (30) Terzi, E.; Holzemann, G.; Seelig, J. Biochemistry 1995, 252, 633–642.
Langmuir 2009, 25(21), 12635–12643
Duff and Kumar
In the future, we hope to provide a clearer understanding of these interactions. Acknowledgment. We thank Mr. V.K. Mudhivarthi for providing R-ZrP samples for the current studies. Financial support from the National Science Foundation is gratefully acknowledged (DMR-0300631and DMR-0604815).
Langmuir 2009, 25(21), 12635–12643
Article
Supporting Information Available: The powder XRD patterns for samples intercalated with proteins (Table 1) and the ITC titration data for Mb and lysozyme binding to R-ZrP (Suppl. Figure 1 and 2) and the lack of correlation of ΔH/ln[Na] vs net charge on the protein (Suppl. Figure 3). This material is available free of charge via the Internet at http://pubs.acs.org.
DOI: 10.1021/la901901k
12643
