Specific Ion–Protein Interactions Dictate Solubility Behavior of a

Jul 27, 2012 - Bioproduct Research & Development, Lilly Corporate Center, Eli Lilly and Company, Indianapolis, IN 46285. Top of Page; Abstract; Introd...
0 downloads 0 Views 964KB Size
Download PDF
Article pubs.acs.org/molecularpharmaceutics

Specific Ion−Protein Interactions Dictate Solubility Behavior of a Monoclonal Antibody at Low Salt Concentrations Le Zhang and Jifeng Zhang*,† Department of Analytical and Formulation Sciences, Amgen Inc., One Amgen Center Drive, Thousand Oaks, California 91320-1799, United States ABSTRACT: The perturbation of salt ions on the solubility of a monoclonal antibody was systematically studied at various pHs in Na2SO4, NaNO3, NaCl, NaF, MgSO4, Mg(NO3)2 and MgCl2 solutions below 350 mM. At pH 7.1, close to the pI, all of the salts increased the solubility of the antibody, following the order of SO42− > NO3− > Cl− > F− for anions and Mg2+ > Na+ for cations. At pH 5.3 where the antibody had a net positive charge, the anions initially followed the order of SO42− > NO3− > Cl− > F− for effectiveness in reducing the solubility and then switched to increasing the solubility retaining the same order. Furthermore, the antibody was more soluble in the Mg2+ salt solutions than in the corresponding Na+ salt solutions with the same anion. At pH 9.0 where the antibody had a net negative charge, an initial decrease in the protein solubility was observed in the solutions of the Mg2+ salts and NaF, but not in the rest of the Na+ salt solutions. Then, the solubility of the antibody was increased by the anions in the order of SO42− > NO3− > Cl− > F−. The above complex behavior is explained based on the ability of both cation and anion from a salt to modulate protein−protein interactions through their specific binding to the protein surface. KEYWORDS: antibody solubility, Hofmeister salts, ion−protein interactions, co-ions and counterions

■

INTRODUCTION Recombinant monoclonal antibodies are an important class of biological therapeutics because of their superior specificity, efficacy and safety profiles.1 Typically, the desired dosage is presented as a liquid formulation at high protein concentrations, i.e., more than 100 mg/mL, in prefilled syringes for convenience and compliance to the treatment regimen. This raises a practical and often overlooked fundamental question of the limits of antibody solubility. Formulating an antibody at a given solution condition beyond its solubility limit could lead to a variety of phase transitions, such as amorphous precipitation, liquid−liquid phase separation, gel formation, or particle formation.2 Understanding of antibody solubility could help predict the phase transition events for antibody solutions at high protein concentrations. Among all the excipients used in protein formulations, salts could affect protein solubility through electrostatic and hydrophobic interactions.3 But, the exact mechanisms of salt ion−protein interactions still remain to be elucidated.4 Historically, the effects of salts at high concentrations on protein solubility have been considered in terms of (direct) Hofmeister series, with anions following the order of SO42− > H2PO4− > F− > Cl− > Br− > NO3− > I− > SCN− and cations following the order of Li+ >Na+ ≈ K+ > NH4+ > Mg2+ in facilitating protein precipitation (salting out).3f,5 Recently there is growing evidence suggesting that the direct interactions of salt ions at the protein/aqueous interface contribute to the Hofmeister effect.3d,f,6 In addition, it has been shown that the © 2012 American Chemical Society

salting-out process of hydrophobic surface is mostly entropically driven while the salting-in process could be either enthalpic or entropic.7 Although protein solubility has been studied extensively at high salt concentrations,8 there is limited literature on the solubility behavior of proteins, especially antibodies, at low concentrations of different salts as the solution pH transitions from below the pI to above. In this transition, the net charge of a protein changes from net positive to neutral and then negative. It has been shown that the sign of the net charge for a protein could affect the salt-dependent solubility behavior. For example, the monovalent anions followed an inverse Hofmeister series for the positively charged lysozyme at pH conditions below its pI, with SCN− being more effective at precipitating the protein than Cl− at low salt concentrations.9 We previously reported the solubility measurement for a monoclonal antibody (measured pI = 7.2) in the monovalent salts of KF, KCl and KSCN at pH conditions near and below the pI.10 Specifically, at pH near the pI, the effectiveness of the anions in increasing the solubility of the antibody was SCN− > Cl− > F−. Surprisingly, at pH 5.3 the anions initially decreased the solubility in the order of SCN− > Cl− > F− and then began to increase solubility with the same order of effectiveness. Received: Revised: Accepted: Published: 2582

April 5, 2012 July 11, 2012 July 27, 2012 July 27, 2012 dx.doi.org/10.1021/mp300183a | Mol. Pharmaceutics 2012, 9, 2582−2590

Molecular Pharmaceutics

Article

crystalline precipitate solution was added into the wells of a Symyx 96 well plate. After the supernatant was removed, sterile water was added to each well to wash the pellet twice. Then 300 μL of solubilization medium was added to each well. The plate was centrifuged and ready for the initial concentration measurement. The plate was shaken gently on a horizontal rotator for solubilization when not used for concentration measurements. During the entire study, the samples at pH 5.3 and 7.1 were kept at room temperature. The pH 9.0 samples were at 2−8 °C during the entire duration of the experiment including centrifugation, plate shaking and concentration measurements to minimize potential chemical modifications, e.g., deamidation, of the antibody. Concentration Measurement. To determine the antibody concentration in the solution phase during solubilization, the sample plate was centrifuged and loaded onto an Agilent HPLC system. An aliquot of 1 μL solution was taken by the autosampler and injected directly for UV measurement at 280 nm. The antibody concentration of the sample was then calculated based on the area of the peak, from a standard curve generated by injecting a set of standards at different antibody concentrations. The HPLC was operated with a flow rate of 0.5 mL/min under isocratic conditions with the mobile phase of 100 mM sodium phosphate and 250 mM sodium chloride at pH 6.8. Typically within a few days, the antibody concentration in the solution phase reached equilibrium and the final antibody concentration was reported as the solubility in mg/mL. The wells were inspected under the microscope at the end of the experiment, and the solubility was only reported for those conditions still containing the crystalline precipitates. Zeta Potential Measurements. Zeta potential values of the antibody samples were measured with a Malvern Zetasizer Nano S instrument at 25 °C using a DTS 1070 zeta dip cell with a 12 mm square glass cuvette. The zeta potential was calculated by the DTS (nano, Ver.5.03) program. The samples were prepared at 1.0 mg/mL in different concentrations of NaCl and MgCl2 at pH 4.8 (20 mM acetate) and pH 9.0 (Tris 25 mM), and then measured directly on the Malvern Nano instrument. Triplicate measurements were made for each sample, and the averaged value of the zeta potentials was reported.

In this article, we extended our work to study how the solubility of this antibody was affected by Mg2+, a divalent cation in the form of the SO42−, NO3− and Cl− salts versus Na+, a monovalent cation in the forms of SO42−, NO3−, Cl− and F− salts at pH 5.3, 7.1, and 9.0. Under these solution conditions the protein has a net positive charge, is almost net charge neutral, and has a net negative charge, respectively. From the perspective of ion−protein interactions, it is anticipated that Mg2+ and SO42− would impart unique effects on the solubility of the antibody because they are strongly hydrated and have double valency in comparison to Na+ and the other anions, respectively.3b,c,11 In addition, we aim to expand our understanding of the antibody solubility in the following three areas: (i) the salt’s effect of the antibody solubility at pH close to its pI when the antibody is net charge neutral; (ii) how do cations, acting as the counterions, affect the protein solubility for the negatively charged antibody at a pH above the pI, with the anions as the co-ions; (iii) modulations of the solubility of the positively charged antibodies by the anion as counterions at the solution pH below the pI, with the cations as the co-ions. By charting the antibody solubility against the perturbation by the salts at different pH, we have discovered that the effectiveness of both cations and anions in affecting protein solubility is dependent on the solution conditions, e.g., pH, salt concentration, and salt type. Particularly, SO42− increased the solubility of the antibody even more dramatically than NO3−, rather than simply decreasing it (salting-out). We attempt to rationalize this complex behavior based on specific interactions (or binding) of ions to the antibody surface and their effects on protein−protein interactions. It is well established that changes in protein solubility can be correlated to the modulation of protein−protein interactions in solutions.12 Specifically, when protein−protein interactions become more attractive, protein solubility decreases. In addition, the binding (adsorption) of both cations and anions with the antibody surface and their effects on protein−protein interactions could be used to explain the complex and puzzling salt-specific effects on the antibody solubility.

■

MATERIALS AND METHODS Solubilization Media Preparation. Similar procedures as described previously were used for preparation of the solubilization media at pH 5.3 and 7.1.10 Briefly, the individual concentrated stock solutions of NaNO3, NaCl, NaF, Na2SO4, Mg(NO3)2, MgSO4, MgCl2 were added into the pH 5.3 and 7.1 starting solutions of 20 and 10 mM acetate and phosphate, respectively. The salts were USP grade (or higher) and purchased from Sigma Aldrich. The pH 9.0 starting solution was made by titrating the 25 mM Tris-base solution using 5 N HCl. The pH 9.0 solubilization media were then prepared by adding different amounts of the salt stock solutions to the Tris buffer. All the solubilization media were sterile and stored at 2− 8 °C before use. Unless mentioned otherwise, when salt concentration is referenced throughout this article, it is the added salt concentration, not including contributions from the buffer components. Solubility Plate Design and Preparation for Concentration Measurement. The monoclonal antibody of an IgG2 with an experimentally measured pI of 7.2 was produced at Amgen, Inc. The antibody crystalline precipitate was prepared using the procedures as previously described.10 The solubility plate design and preparation for solubility measurement were similar to our earlier work.10 Briefly, 550 μL of the antibody

■

RESULTS Antibody Solubility Data at pH 7.1 (Net Charge Neutral). Shown in Figure 1 is the antibody solubility in the different salt solutions at pH 7.1, close to its pI of 7.2, where the

Figure 1. Antibody solubility in the Na+ and Mg2+ salt solutions at pH 7.1. 2583

dx.doi.org/10.1021/mp300183a | Mol. Pharmaceutics 2012, 9, 2582−2590

Molecular Pharmaceutics

Article

antibody was expected to be net charge neutral. The addition of all seven salts resulted in an increase in protein solubility. This observation is in agreement with the previously reported salting-in behavior for this antibody in the monovalent salt solutions of KSCN, KCl and KF at pH 7.110 and is also consistent with the results obtained for small globular proteins including thermolysin13 and urate oxidase at a pH close to their pI.14 Such salting-in behavior is typical for net charge neutral proteins at low ionic strength.8d The salting-in constant for each salt (in Figure 2) was obtained by fitting the solubility data

Figure 2. Salting-in constants for the Na+ and Mg2+ salts at pH 7.1. In the inset, the antibody solubility is plotted against the Cl − concentration in order to demonstrate the salting-in efficacy difference between 1 Mg2+ and 2 Na+.

Figure 3. Antibody solubility in the solutions of (a) the Na+ salts and (b) the Mg2+ salts at pH 5.3. Shown in the inset of panel a is the zoomed view of antibody solubility below the salt concentrations of 50 mM.

against salt concentration using Setchenow plots.3a,10 The solubility data beyond the salt concentrations of 200 mM NaCl, 80 mM Na2SO4, and 300 mM NaF were not used for the calculation of salting-in constant because of the loss of the linearity, as indicated by a R2 value of less than 0.99. This might be in agreement with onset of the salting-out effect caused by the increase in the surface tension at high salt concentrations.15 The salting-in constants were 20.1, 10.0, 8.0, and 4.4 M−1 for Na2SO4, NaNO3, NaCl, and NaF, respectively, suggesting that the anions followed the ranking of SO42− > NO3− > Cl− > F− for their effectiveness at increasing the solubility of the antibody. For the divalent Mg2+ salts, the salting-in constants were 25.6 M−1, 25.1 M−1, and 25.1 M−1 for Mg(NO3)2, MgCl2, and MgSO4, respectively. Therefore, the Mg2+ salts were more effective at salting-in the antibody than the Na+ salts with the same anion. Further comparison of the salting-in constants for Na2SO4 and MgSO4 suggests that 1 Mg2+ might be more effective than 2 Na+ at increasing the solubility of the antibody. This observation is substantiated by comparison of the antibody solubility data in NaCl and MgCl2, although there are twice as many Cl− ions in MgCl2 as in NaCl at the same salt concentration. The trend becomes more apparent when the protein solubility is plotted against the concentration of Cl− (inset in Figure 2) where there are equal amounts of Cl− in both salt solutions. Antibody Solubility Data at pH 5.3 below its pI (Net Positive Charge). Figures 3a and 3b show the antibody solubility behavior in the Na+ and Mg2+ salt solutions at pH 5.3, below the pI, where the antibody should carry a net positive charge. They all exhibited nonmonotonic behavior: the antibody solubility initially decreased, reached a minimum and then rose. Anion (Counterion) Effect. As indicated by the slope of the decreased antibody solubility in the Figure 3a inset, at salt

concentrations below 10 mM the effectiveness of the anions in the Na+ salts in decreasing antibody solubility were in the order of Na2SO4 > NaNO3 > NaCl > NaF. In comparison, Na2SO4 was the most effective at increasing the solubility of the antibody at pH 7.1. Furthermore, the protein solubility minimum in the NaNO3 solution, ∼22 mg/mL, was much lower than the ∼30−32 mg/mL minimum seen in other salts. This type of behavior was also observed for lysozyme at acidic pHs in the monovalent salts studied.9b,16 After the protein solubility reached the minimum, further addition of all the salts resulted in an increase in the solubility of the antibody. In addition, the effectiveness of the anion in raising protein solubility now occurred with a trend similar to that at pH 7.1. It is interesting to note that Na2SO4, the most effective salt for reducing the protein solubility initially, became the most effective at raising the solubility. For the Mg2+ salts in Figure 3b, similar effectiveness of the anion in reducing solubility was initially observed in the order of 1 SO42− ≅ 2 NO3− ≫ 2 Cl−. Again, the protein solubility minimum was the lowest in NO3−, ∼38 mg/mL, followed by SO42− (∼43 mg/mL) and Cl− (∼48 mg/mL). After reaching the minimum, the slope of the increase in protein solubility as a function of salt concentrations was similar between the three salts. Cation (Co-Ion) Effect. The comparison of the solubility of the antibody in the MgCl2 and NaCl solutions shown in Figure 4 demonstrates that the antibody was more soluble in the MgCl2 solution. Again when the Cl− concentration was plotted as the x-axis (in Figure 4 inset), the comparison suggests that 1 Mg2+ was less effective at reducing the antibody solubility initially and then began to raise the antibody solubility slightly 2584

dx.doi.org/10.1021/mp300183a | Mol. Pharmaceutics 2012, 9, 2582−2590

Molecular Pharmaceutics

Article

solutions was similar to that seen at pH 7.1. However, there was a nonmonotonic relationship in the NaF solutions. As illustrated in Figure 5b, there was a nonmonotonic relationship at pH 9.0 between the antibody solubility and salt concentration for all Mg2+ salt solutions regardless of the anion. Once the protein solubility reached the minimum, all of the Mg2+ salts began to increase the antibody solubility. Cation (Counterion) Effect. The comparison of the antibody solubility data in the NaCl and MgCl2 solutions is shown in Figure 6. The MgCl2 was more effective in decreasing the

Figure 4. Comparison of the antibody solubility in the NaCl and MgCl2 solutions at pH 5.3. In the inset, the antibody solubility is plotted against the Cl− concentration in order to demonstrate the difference by 1 Mg2+ and 2 Na+.

more effectively than 2 Na+. Similar behavior was observed for the two pairs of NaNO3/Mg(NO3)2 and Na2SO4/MgSO4 (comparison data not shown). Antibody Solubility Data at pH 9.0 above Its pI (Net Negative Charge). Anion (Co-Ion) Effect. The antibody solubility in the Na+ and Mg2+ salts at pH 9.0 (above its pI), where the antibody was a net negatively charged protein, is shown in Figures 5a and 5b. In the NaNO3, NaCl and Na2SO4 solutions (Figure 5a), the solubility of the antibody did not change significantly at the beginning, and then started to increase monotonically afterward. The sharpness of the slope of the increase in protein solubility in the individual Na+ salt

Figure 6. Comparison of the antibody solubility in the NaCl and MgCl2 solutions at pH 9.0. In the inset, the antibody solubility is plotted against the Cl− concentration in order to demonstrate the difference between 1 Mg2+ and 2 Na+.

protein solubility at low salt concentrations ( Cl− > F− from strong to weak when binding to the antibody surface. On the other hand, the strongly hydrated divalent anions, i.e., SO42− in our study, could interact with protein surface,21 possibly through the acidic proton from the amide moiety22 and guanidinium ions on the protein surface.11,23 Recently, it was demonstrated that SO42− could interact with the positively charged BSA through positive-charged side chain more strongly than SCN− using vibrational sum frequency spectroscopy (VSFS) to measure the interfacial water structure.11 In addition, recent molecular dynamic simulations confirmed the strong propensity of SO42− to interact with the positively charged surfaces of horseradish peroxidase (HRP) and bovine pancreatic trypsin inhibitor (BPTI).24 Furthermore, the stronger binding of SO42− than Cl− to the positive-charged Arg was attributed to the slower electrophoretic mobility for Arg and tetraarginine.23b Although SO42− typically decreases the solubility of neutral hydrophobic molecules through desolvation of the polar and nonpolar portions of a molecule,3a,25 the salting-out effect is relatively weak at low concentrations.3f In summary, the ability of the anions used in our study to interact with the protein surface could follow the order of SO42− > NO3− > Cl− > F−. For the monovalent Na+ used in our study, it has been demonstrated that Na+ is typically excluded from the polar amide and nonpolar residues,6a and a net positively charged protein surface,11 in comparison to the preferential adsorption of the chaotropic anions.6a,11 In addition, some recent work26 has suggested that its interaction with the carboxylate on a protein surface could be much weaker than that of Mg2+. Furthermore, it has been demonstrated that Mg2+ could interact with not only carboxylates from proteins but also amide oxygen at the peptide group.27 Overall, the interaction of Mg2+ with a protein could be stronger than that of Na+. Proposed Mechanism for the Effect of Salt Ions on Antibody Solubility. The mechanisms are still not fully defined for how a salt, present as both cation and anion in solutions, modulates protein solubility as the concentration of salts and pH change. Based on the antibody solubility behavior and zeta-potential measurement results noted above, we propose the following qualitative biophysical mechanism. Salting-In at pH 7.1 Close to pI. At a solution pH near the pI of the protein, the antibody is net charge neutral. Both the cation and anion can access the antibody surface and work collectively to raise protein solubility monotonically as demonstrated in Figure 1 by interacting with their counterparts exposed on the protein surface. It is well-acknowledged that the preferential accumulation of solutes (enrichment near the protein surface in comparison to the bulk) at the interface between a protein and the bulk solution decreases the free energy of the protein, effectively disrupts attractive protein−protein interactions and thus results in an increase in the protein solubility.28 The surprisingly large salting-in constant for Na2SO4 suggests that SO42− may be adsorbed onto the antibody surface through strong electrostatic attractive interactions. Because Na+ in general interacts with protein fairly weakly, as mentioned above, the most likely

Figure 7. Results for the zeta-potential measurement for the antibody in the NaCl and MgCl2 solutions: (a) at pH 4.8 and (b) at 9.0. In the inset, the zeta-potential value is plotted against the Cl− concentration in order to demonstrate the difference between 1 Mg2+ and 2 Na+.

the antibody by Mg2+. In order to clearly compare the effectiveness of Mg2+ and Na+, the perturbation of the zetapotential was plotted against the Cl− concentration as shown in Figure 7a and 7b insets. The comparisons reveal that MgCl2 was less effective at decreasing the zeta-potential than NaCl when the pH is below the pI. In contrast, MgCl2 became more effective at increasing the zeta-potential at pH above the pI.

■

DISCUSSION Specific Ion−Protein Interactions. In order to understand the perturbation of antibody solubility by the selected salts under different pH conditions, it is necessary to describe the (possible) specific interaction mechanisms between salt ions and individual functional groups on the protein surface. Recent evidence suggests that specific interactions between salt ions and the protein surface, as opposed to the effect of the ions on the bulk solution property, play an essential role in affecting protein biophysical behavior in solutions.3d,f,6 In general, surface chemistry of the antibody is heterogeneous, composed of exposed charged (positive and negative), nonpolar and polar side chains as well as the peptide backbone. Salt ions may interact with the antibody at all these sites. According to the law of matching water affinities, oppositely charged ions in solutions form inner sphere ion pairs spontaneously when they have similar water affinities.3b−d,18 Specifically, weakly hydrated (or chaotropic) monovalent anions exhibit strong interactions with the positively charged, and even polar and nonpolar side chains from the amino acid residues on the protein surface (all of which are weakly hydrated) in comparison to strongly hydrated (or kosmotropic) 2586

dx.doi.org/10.1021/mp300183a | Mol. Pharmaceutics 2012, 9, 2582−2590

Molecular Pharmaceutics

Article

with the net positively charged antibody, consistent with their ranking for interacting with a protein. At pH 9.0 for the net negatively charged antibody, the effectiveness of cations followed the order of Mg2+ > Na+ for reducing the protein solubility as shown in Figure 6, matching their ranking for interacting with a protein. In stage 2 the protein solubility reaches a minimum when enough counterions are bound to achieve charge neutralization at each specific concentration. This mechanistic interpretation is in principle consistent with Tanford’s theory that the protein solubility reaches a minimum when the protein binds an equal number of counterions to satiate its surface charge.31 In other words, the antibody now can be viewed as a pseudo charge neutral complex. Then at stage 3, the pseudo charge neutral complex could be salted-in specifically by the salts, similar to the situation at pH 7.1 where the protein was net charge neutral. This is consistent with the observation that after the protein solubility minimum was reached, the further addition of any of the salts at pH 5.3, and the Mg2+ salts at pH 9.0, began to salt in the antibody in a similar manner to that observed for the neutral protein at pH 7.1. In Figure 3a, the order of the anions, SO42− > NO3− > Cl− > F−, for increasing protein solubility in stage 3, is in agreement with that at pH 7.1. Especially, Na2SO4 was the most effective at increasing the solubility of the antibody at both pH 5.3 and pH 7.1. It is difficult to rank the Mg2+ salts in stage 3 because of their subtle differences in their effectiveness at increasing the protein solubility, as suggested by their small differences in salting-in constants at pH 7.1 shown in Figure 2a. Nonetheless, the increase in solubility seen with all of the solutions supports the idea of the salting-in for a pseudo neutral-charge complex. Importantly, the above 3-stage mechanism is built based on the premise that the co-ion interacts with the antibody weakly and the interaction between the counterion and antibody dominates. This presupposition is typically true for a positively charged protein as in the Na+ and Mg2+ salt solutions at pH 5.3 because it has been demonstrated that the interactions of anions with net positively charged proteins are much stronger than those of cations,11 as also suggested in Figure 3 for this antibody. Outcomes from Co-Ion Competition during Charge Neutralization. The effect of the co-ion on the antibody solubility could be substantial as illustrated in Figure 4 and Figure 5a. The solubility data suggests that, opposite to the effect of the counterion, the binding of the co-ion with the protein surfaces tends to increase protein solubility and render the counterion less effective at reducing the protein solubility. There are two important pieces of evidence of co-ion binding and its ability to increase protein solubility effectively at pH 5.3 (in Figure 4) (but this interaction is still weaker than that between the counterion and protein): (i) the slope of the decrease in zeta-potential value and protein solubility was shallower in the MgCl2 solutions; (ii) the protein solubility minimum was higher in the Mg2+ salt compared to the Na+ with the same anion at pH 5.3, as illustrated by the Cl− salts in Figure 4. Our experimental findings are consistent with the solubility measurement for lysozyme in multivalent cation salt solutions.32 The effect of co-ions on a positively charged protein has also been demonstrated based on ternary diffusion measurements for lysozyme in both NaCl and MgCl 2 solutions.33

mechanism for Na2SO4 to salt in the antibody so dramatically is through the electrostatic interaction of sulfate with the positively charged side chains from Arg and acidic protons on the antibody surface. Moreover, the order of the anions for salting in the antibody, SO42− > NO3− > Cl− > F−, matches their ranking of interaction strength with proteins. The collective salting-in effect by both cations and anions on the antibody and effectively decreasing the attractive intermolecular interactions is supported by the fact that MgSO4 was more powerful than Na2SO4 at raising the antibody solubility per molar concentration. Our experimental finding that 1 Mg2+ was more effective than 2 Na+ at increasing the antibody solubility is consistent with the fact that Mg2+ could interact with both the amide bond and carboxylate more strongly than Na+. The larger salting-in constants of the other Mg2+ salts versus their corresponding Na+ salts are in agreement with the stronger efficacy of Mg2+ in increasing the antibody solubility (inset in Figure 2). Nonmonotonic Solubility Behavior for Charged Antibody. The nonmonotonic behavior in the antibody solubility with increasing salt concentration can be described in a 3-stage process: a decrease in stage 1, a local minimum in stage 2 and finally an increase in protein solubility in stage 3. When the solution pH moves away from the pI of a protein, the antibody acquires a net charge, albeit not evenly distributed across the surface of the protein. It is then expected that counterions may interact with the protein globally. Specifically, for a net positively charged protein at pHs below their pI, the anion from a salt is the counterion, which could bind to the protein and neutralize its net charge. On the other hand, when the solution pH is above the pI of the protein and the protein becomes net negatively charged, the cation is the counterion to carry out charge neutralization. The results of the zeta-potential measurement for the Cl− salts at pH 4.8 and MgCl2 at pH 9.0 in Figure 7 are in agreement with the charge neutralization process. The charge neutralization by counterions was also demonstrated in other protein systems, such as lysozyme by monovalent anions9b,c and BSA by multivalent cations.29 This charge neutralization due to the adsorption of counterions reveals that the sign of the net charge for a protein is one of the key elements in determining the effect of a specific salt on the solubility of the protein. At low salt concentrations studied in our experiment, the electrostatic repulsive interactions due to the presence of net charge may initially dominate other types of protein−protein interactions when a protein carries a sufficiently high surface charge density. Then, the introduction of counterions by the addition of salts neutralizes the net charge of the protein, decreases the electrostatic repulsive interactions and therefore leads to a decrease in protein solubility.3b,30 The abovedescribed mechanism is consistent with (i) the aniondependent decrease in antibody solubility for the positively charged antibody at pH 5.3 and (ii) the cation-dependent decrease in solubility for the negatively charged protein in the Mg2+ salts and NaF solutions at pH 9.0. Because of this charge neutralization process, it is then expected that the more strongly a counterion interacts with a protein, the more effectively it neutralizes the protein net charge, weakens the repulsive interactions, and decreases the protein solubility on a salt concentration basis. Specifically in stage 1 (Figure 3a inset), the effectiveness of the anions in initially reducing the protein solubility at pH 5.3 follows the trend of SO42− > NO3− > Cl− > F− for the Na+ salts to interact 2587

dx.doi.org/10.1021/mp300183a | Mol. Pharmaceutics 2012, 9, 2582−2590

Molecular Pharmaceutics

Article

Notes

For the negatively charged antibody (as indicated by the negative zeta-potential value) at pH 9.0 above the pI, Na2SO4, NaNO3 and NaCl salted in the antibody almost monotonically in Figure 5a, which is similar to the trends at pH 7.1. It suggests that the interactions between the above anions and the antibody surface are stronger than that between the Na+ and antibody. Our explanations are consistent with the outcome of recently published theoretical calculations that demonstrated strong adsorption of polarizable co-ion (weakly hydrated anions) and specific ion effects at a low and moderately negatively charged (−0.01 C/m2) hydrophobic surface.34 In our case, the estimated surface charge density for this antibody is low,2b i.e., less than −0.002 C/m2 at pH 9.0, where the coions most likely could approach and interact with the antibody. In contrast, in the NaF solution, Na+ is most likely to interact with negatively charged proteins more strongly than F−, resulting in the initial subtle drop in protein solubility. This is consistent with the fact that the strongly hydrated F− can be considered expelled from protein surfaces (especially negatively charged surfaces). To the best of our knowledge, the deviation of the solubility profile in the NaF solutions from the rest of the Na+ salts might be the first piece of convincing experimental evidence for demonstrating that Na+ interacts with a negatively charged protein more strongly than F−.

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS The authors would like to thank Dr. Linda Narhi for reviewing the manuscript and providing valuable comments.

■

CONCLUSIONS In this article, we have presented the antibody solubility behavior in different salts at the pH conditions below, near and above its pI. What we have learned might be applicable to other protein systems. Specifically, for a net charge neutral protein at a pH close to its pI, both the cations and anions present in the solution work collectively to increase protein solubility monotonically where the overall effectiveness of a salt is determined by the strength of the interaction of both ions with the protein. On the other hand, when the protein carries a sufficient net charge and the counterion interacts with the protein more strongly than the co-ion, protein solubility generally displays a salt-specific nonmonotonic transition. Mechanistically, the counterions weaken the electrostatic repulsive interactions between protein molecules, acting as a charge neutralizer, and result in decreased protein solubility. In general, the more strongly a counterion interacts with the protein, the more effectively it reduces the protein solubility. After charge neutralization is attained, the protein is salted-in by the further addition of ions, as if it was net charge neutral. Opposite to the effect of the counterion, the interaction of the co-ion with the protein surface tends to increase the protein solubility and render the counterion less effective at reducing the protein solubility. In the extreme, when the co-ion interacts with the charged protein more strongly than the counterion, only salt-dependent increase in protein solubility would occur.

■

REFERENCES

(1) (a) Carter, P. J. Potent antibody therapeutics by design. Nat. Rev. Immunol. 2006, 6 (5), 343−357. (b) Weiner, L. M.; Surana, R.; Wang, S. Monoclonal antibodies: versatile platforms for cancer immunotherapy. Nat. Rev. Immunol. 2010, 10 (5), 317−327. (c) Hansel, T. T.; Kropshofer, H.; Singer, T.; Mitchell, J. A.; George, A. J. T. The safety and side effects of monoclonal antibodies. Nat. Rev. Drug Discovery 2010, 9 (4), 325−338. (2) (a) Dumetz, A. C.; Chockla, A. M.; Kaler, E. W.; Lenhoff, A. M. Protein Phase Behavior in Aqueous Solutions: Crystallization, LiquidLiquid Phase Separation, Gels, and Aggregates. Biophys. J. 2008, 94, 570−583. (b) Mason, B. D.; Zhang-van Enk, J.; Zhang, L.; Remmele, R. L.; Zhang, J. Liquid-liquid phase separation of a monoclonal antibody and influence of Hofmiester anions. Biophys. J. 2010, 99, 3792−3800. (3) (a) Baldwin, R. L. How Hofmeister ion interactions affect protein stability. Biophys. J. 1996, 71, 2056−2063. (b) Collins, K. D. Charge density-dependent strength of hydration and biological structure. Biophys. J. 1997, 72, 65−75. (c) Collins, K. D. Ions from the Hofmeister series and osmolytes: effect on proteins in solution and in the crystallization process. Methods 2004, 34, 300−311. (d) Collins, K. D.; Neilson, G. W.; Enderby, J. E. Ions in water: characterizing the forces that control chemical processes and biological structure. Biophys. Chem. 2007, 128, 95−104. (e) Curtis, R. A.; Lue, L. A molecular approach to bioseparations: protein-protein and protein-salt intractions. Chem. Eng. Sci. 2006, 61, 907−923. (f) Zhang, Y.; Cremer, P. S. Interactions between macromolecules and ions: the Hofmiester series. Curr. Opin. Chem. Biol. 2006, 10, 658−663. (4) (a) Kunz, W. Specific ion effects in liquid, in biological systems, and interfaces. Pure Appl. Chem. 2006, 78, 1611−1617. (b) Zangi, R. Can salting-in/salting-out ions be classified as chaotropes/kosmotropes? J. Phys. Chem. B 2010, 114, 643−650. (5) Carbonnaux, C.; Ries-Kautt, M.; Ducruix, A. Relative effectiveness of various anions on the solubility of acidic Hypoderma lineatum collagenase at pH 7.2. Protein Sci. 1995, 4 (10), 2123−2128. (6) (a) Chen, X.; Yang, T.; Kataoka, S.; Cremer, P. S. Specific Ion Effects on Interfacial Water Structure near Macromolecules. J. Am. Chem. Soc. 2007, 129 (40), 12272−12279. (b) Zhang, Y.; Furyk, S.; Bergbreiter, D. E.; Cremer, P. S. Specific Ion Effects on the Water Solubility of Macromolecules: PNIPAM and the Hofmeister Series. J. Am. Chem. Soc. 2005, 127 (41), 14505−14510. (7) Zangi, R.; Hagen, M.; Berne, B. J. Effect of Ions on the Hydrophobic Interaction between Two Plates. J. Am. Chem. Soc. 2007, 129 (15), 4678−4686. (8) (a) Arakawa, T.; Bhat, R.; Timasheff, S. N. Preferential interactions determine protein solubility in three-component solutions: the magnesium chloride system. Biochemistry 1990, 29 (7), 1914−1923. (b) Arakawa, T.; Timasheff, S. N. Mechanism of protein salting in and salting out by divalent cation salts: balance between hydration and salt binding. Biochemistry 1984, 23, 5912− 5923. (c) Green, A. Studies in the physical chemistry of the proteins X: the solubility of hemoglobin in solutions of chlorides and sulfates of varying concentration. J. Biol. Chem. 1932, 95, 47−66. (d) Arakawa, T.; Timasheff, S. N. Theory of protein solubility. Methods Enzymol. 1985, 114, 49−77. (9) (a) Retailleau, P.; Ducruix, A.; Ries-Kautt, M. Importance of the nature of anions in lysozyme crystallisation correlated with protein net charge variation. Acta Crystallogr., Sect. D 2002, 58 (10), 1576−1581. (b) Ries-Kautt, M. M.; Ducruix, A. F. Relative effectiveness of various ions on the solubility and crystal growth of lysozyme. J. Biol. Chem. 1989, 264, 745−748. (c) Zhang, Y.; Cremer, P. S. The inverse and

AUTHOR INFORMATION

Corresponding Author

*Bioproduct Research & Development, Lilly Corporate Center, Eli Lilly and Company, Indianapolis, IN 46285. Phone: (317) 433-0120. Fax: (317) 276-1271. E-mail: zhang_jifeng@lilly. com. Present Address †

Bioproduct Research & Development, Lilly Corporate Center, Eli Lilly and Company, Indianapolis, IN 46285. 2588

dx.doi.org/10.1021/mp300183a | Mol. Pharmaceutics 2012, 9, 2582−2590

Molecular Pharmaceutics

Article

direct Hofmiester series for lysozyme. Proc. Natl. Acad. Sci U.S.A. 2009, 106, 15249−15253. (10) Zhang, L.; Tan, H.; Fesinmeyer, R.; Matthew, Li, C.; Catrone, D.; Le, D.; Remmele, R. L.; Zhang, J. Antibody Solubility Behavior in Monovalent Salt Solutions Reveals Specific Anion Effects at Low Ionic Strength. J. Pharm. Sci. 2011, 101, 965−977. (11) Chen, X.; Flores, S. C.; Lim, S.-M.; Zhang, Y.; Yang, T.; Kherb, J.; Cremer, P. S. Specific Anion Effects on Water Structure Adjacent to Protein Monolayers. Langmuir 2010, 26 (21), 16447−16454. (12) (a) Broide, M. L.; Tominc, T. M.; Saxowsky, M. D. Using phase transitions to investigate the effect of salts on protein interactions. Phys. Rev. E 1996, 53 (6), 6325. (b) Finet, S.; Vivarès, D.; Bonneté, F.; Tardieu, A.; Carter, C. W., Jr.; Robert, M. S. Controlling Biomolecular Crystallization by Understanding the Distinct Effects of PEGs and Salts on Solubility. In Methods in Enzymology; Academic Press: New York, 2003; Vol. 368, pp 105−129. (13) Inouye, K.; Kuzuya, K.; Tonomura, B. i. Effect of Salts on the Solubility of Thermolysin: A Remarkable Increase in the Solubility as Well as the Activity by the Addition of Salts without Aggregation or Dispersion of Thermolysin. J. Biochem. 1998, 123 (5), 847−852. (14) Giffard, M.; Colloc’h, N.; Ferté, N.; Castro, B.; Bonneté, F. o. Salting-In Effects on Urate Oxidase Crystal Design. Cryst. Growth Des. 2008, 8 (12), 4220−4226. (15) (a) Melander, W.; Horvath, C. Salt effect on hydrophobic interactions in precipitation and chromatography of proteins: an interpretation of the lyotropic serie. Arch. Biochem. Biophys. 1977, 183, 200−215. (b) Timasheff, S. N. Theory of protein solubility. Methods Enzymol. 1985, 114, 49−77. (16) (a) Retailleau, P.; Ries-Kautt, M.; Ducruix, A. No salting-in of lysozyme chloride observed at low ionic strength over a large range of pH. Biophys. J. 1997, 73, 2156−2163. (b) Guilloteau, J.; Ries-Kautt, M.; Ducruix, A. F. Variation of lysozyme solubility as a function of temperature in the presence of organic and inorganic salts. J. Cryst. Growth 1992, 122, 223−230. (17) (a) Zhang, F.; Weggler, S.; Ziller, M. J.; Ianeselli, L.; Heck, B. S.; Hildebrandt, A.; Kohlbacher, O.; Skoda, M. W. A.; Jacobs, R. M. J.; Schreiber, F. Universality of protein reentrant condensation in solution induced by multivalent metal ions. Proteins 2010, 78 (16), 3450−3457. (b) Ianeselli, L.; Zhang, F.; Skoda, M. W. A.; Jacobs, R. M. J.; Martin, R. A.; Callow, S.; Prevost, S.; Schreiber, F. Protein−Protein Interactions in Ovalbumin Solutions Studied by Small-Angle Scattering: Effect of Ionic Strength and the Chemical Nature of Cations. J. Phys. Chem. B 2010, 114 (11), 3776−3783. (18) Collins, K. D. Ion hydration: Implications for cellular function, polyelectrolytes, and protein crystallization. Biophys. Chem. 2006, 119 (3), 271−281. (19) Lund, M.; Vrbka, L.; Jungwirth, P. Specific Ion Binding to Nonpolar Surface Patches of Proteins. J. Am. Chem. Soc. 2008, 130 (35), 11582−11583. (20) Wang, J.; Dauter, M.; Alkire, R.; Joachimiak, A.; Dauter, Z. Triclinic lysozyme at 0.65 A resolution. Acta Crystallogr., Sect. D 2007, 63 (12), 1254−1268. (21) (a) Kanyo, Z. F.; Christianson, D. W. Biological recognition of phosphate and sulfate. J. Biol. Chem. 1991, 266 (7), 4264−8. (b) Chertanova, L.; Pascard, C. Statistical analysis of noncovalent interactions of anion groups in crystal structures. I. Hydrogen bonding of sulfate anions. Acta Crystallogr., Sect. B 1996, 52 (4), 677−684. (c) Flanagan, J. U.; Rossjohn, J.; Parker, M. W.; Board, P. G.; Chelvanayagam, G. Mutagenic analysis of conserved arginine residues in and around the novel sulfate binding pocket of the human Theta class glutathione transferase T2-2. Protein Sci. 1999, 8 (10), 2205− 2212. (d) Mizutani, K.; Yamashita, H.; Mikami, B.; Hirose, M. Crystal Structure at 1.9 Å Resolution of the Apoovotransferrin N-Lobe Bound by Sulfate Anions: Implications for the Domain Opening and Iron Release Mechanism. Biochemistry 2000, 39 (12), 3258−3265. (e) Verlinde, C. L. M. J.; Noble, M. E. M.; Kalk, K. H.; Groendijk, H.; Wierenga, R. K.; Hol, W. G. J. Anion binding at the active site of trypanosomal triosephosphate isomerase. Eur. J. Biochem. 1991, 198 (1), 53−57. (f) Rigden, D. J.; Walter, R. A.; Phillips, S. E. V.;

Fothergill-Gilmore, L. A. Sulphate ions observed in the 2.12 Å structure of a new crystal form of S. cerevisiae phosphoglycerate mutase provide insights into understanding the catalytic mechanism. J. Mol. Biol. 1999, 286 (5), 1507−1517. (g) Rossjohn, J.; McKinstry, W. J.; Oakley, A. J.; Verger, D.; Flanagan, J.; Chelvanayagam, G.; Tan, K.L.; Board, P. G.; Parker, M. W. Human theta class glutathione transferase: the crystal structure reveals a sulfate-binding pocket within a buried active site. Structure 1998, 6 (3), 309−322. (h) Quiocho, F. A. Atomic basis of the exquisite specificity of phosphate and sulfate transport receptors. Kidney Int. 1996, 49 (4), 943−946. (i) Voegtli, W. C.; White, D. J.; Reiter, N. J.; Rusnak, F.; Rosenzweig, A. C. Structure of the Bacteriophage λ Ser/Thr Protein Phosphatase with Sulfate Ion Bound in Two Coordination Modes. Biochemistry 2000, 39 (50), 15365−15374. (j) Chakrabarti, P. Anion Binding Sites in Protein Structures. J. Mol. Biol. 1993, 234 (2), 463−482. (k) Copley, R. R.; Barton, G. J. A Structural Analysis of Phosphate and Sulphate Binding Sites in Proteins: Estimation of Propensities for Binding and Conservation of Phosphate Binding Sites. J. Mol. Biol. 1994, 242 (4), 321−329. (22) Collins, K. D. Why continuum electrostatics theories cannot explain biological structure, polyelectrolytes or ionic strength effects in ion−protein interactions. Biophys. Chem. 2012, 167, 33−49. (23) (a) Mason, P. E.; Dempsey, C. E.; Vrbka, L.; Heyda, J.; Brady, J. W.; Jungwirth, P. Specificity of Ion−Protein Interactions: Complementary and Competitive Effects of Tetrapropylammonium, Guanidinium, Sulfate, and Chloride Ions. J. Phys. Chem. B 2009, 113 (10), 3227−3234. (b) Wernersson, E.; Heyda, J.; Kubíčková, A.; Křížek, T. s.; Coufal, P.; Jungwirth, P. Effect of Association with Sulfate on the Electrophoretic Mobility of Polyarginine and Polylysine. J. Phys. Chem. B 2010, 114 (36), 11934−11941. (c) Gokarn, Y. R.; Fesinmeyer, R. M.; Saluja, A.; Razinkov, V.; Chase, S. F.; Laue, T. M.; Brems, D. N. Effective charge measurements reveal selective and preferential accumulation of anions, but not cations, at the protein surface in dilute salt solutions. Protein Sci. 2011, 20 (3), 580−587. (24) Vrbka, L.; Jungwirth, P.; Bauduin, P.; Touraud, D.; Kunz, W. Specific Ion Effects at Protein Surfaces: A Molecular Dynamics Study of Bovine Pancreatic Trypsin Inhibitor and Horseradish Peroxidase in Selected Salt Solutions. J. Phys. Chem. B 2006, 110 (13), 7036−7043. (25) (a) Hippel, P. V.; Hleich, T. S. Ion Effects on the Solution Structure of Biological Macromolecules. Acc. Chem. Res. 1969, 2, 257− 265. (b) Nandi, P. K.; Robinson, D. R. The effects of salts on the free energy of the peptide group. J. Am. Chem. Soc. 1972, 94, 1299−1308. (c) Nandi, P. K.; Robinson, D. R. Effects of salts on the free energies of nonpolar groups in model peptides. J. Am. Chem. Soc. 1972, 94 (4), 1308−1315. (d) Robinson, D. R.; Jencks, W. P. The effect of concentrated salt solutions on the activity coefficient of acetyltetraglycine ethyl ester. J. Am. Chem. Soc. 1965, 87, 2470−2479. (26) (a) Kubíčková, A.; Křížek, T.; Coufal, P.; Vazdar, M.; Wernersson, E.; Heyda, J.; Jungwirth, P. Overcharging in Biological Systems: Reversal of Electrophoretic Mobility of Aqueous Polyaspartate by Multivalent Cations. Phys. Rev. Lett. 2012, 108 (18), 186101. (b) Kherb, J.; Flores, S. C.; Cremer, P. S. Role of carboxylate Side Chains in the Cation Hofmeister Series. J. Phys. Chem. B 2012, 116 (25), 7389−7397. (27) Harding, M. Small revisions to predicted distances around metal sites in proteins. Acta Crystallogr., Sect. D 2006, 62 (6), 678−682. (28) Timasheff, S. N. Protein Hydration, Thermodynamic Binding, and Preferential Hydration. Biochemistry 2002, 41 (46), 13473−13482. (29) Zhang, F.; Skoda, M. W. A.; Jacobs, R. M. J.; Zorn, S.; Martin, R. A.; Martin, C. M.; Clark, G. F.; Weggler, S.; Hildebrandt, A.; Kohlbacher, O.; Schreiber, F. Reentrant Condensation of Proteins in Solution Induced by Multivalent Counterions. Phys. Rev. Lett. 2008, 101 (14), 148101. (30) Riès-Kautt, M. M.; Ducruix, A. F. Crystallization of basic proteins by ion pairing. J. Cryst. Growth 1991, 110 (1−2), 20−25. (31) Tanford, C. Physical chemistry of macromolecules; Wiley: New York, 1961; pp 238−253. 2589

dx.doi.org/10.1021/mp300183a | Mol. Pharmaceutics 2012, 9, 2582−2590

Molecular Pharmaceutics

Article

(32) Benas, P.; Legrand, L.; Riess-Kautt, M. Strong and specific effects of cations on lysozyme chloride solubility. Acta Crystallogr., Sect. D 2002, 58, 1582−1587. (33) Annunziata, O.; Paduano, L.; Albright, J. G. The Effect of Salt Stoichiometry on Protein−Salt Interactions Determined by Ternary Diffusion in Aqueous Solutions. J. Phys. Chem. B 2006, 110 (32), 16139−16147. (34) Lima, E. R. A.; Bostrom, M.; Horinek, D.; Biscaia, E. C.; Kunz, W.; Tavares, F. W. Co-Ion and Ion Competition Effects: Ion Distributions Close to a Hydrophobic Solid Surface in Mixed Electrolyte Solutions. Langmuir 2008, 24 (8), 3944−3948.

2590

dx.doi.org/10.1021/mp300183a | Mol. Pharmaceutics 2012, 9, 2582−2590