Much work has been performed on understanding the effects of additives on protein thermodynamics and degradation kinetics, in particular addressing the Hofmeister series and other broad empirical phenomena. Little attention, however, has been paid to the effect of additive-additive interactions on proteins. Our group and others have recently shown that such interactions can actually govern protein events, such as aggregation. Here we use dendrimers, which have the advantage that both size and surface chemical groups can be changed and therein studied independently. Dendrimers are a relatively new and broad class of materials which have been demonstrated useful in biological and therapeutic applications, such as drug delivery, perturbing amyloid formation, etc. Guanidinium modified dendrimers pose an interesting case given that guanidinium can form multiple attractive hydrogen bonds with either a protein surface or other components in solution, such as hydrogen bond accepting counterions. Here we present a study which shows that the behavior of such macromolecule species (modified PAMAM dendrimers) is governed by intra-solvent interactions. Attractive guanidinium-anion interactions seem to cause clustering in solution, which inhibits cooperative binding to the protein surface but at the same time, significantly suppresses nonnative aggregation.
Citation: Schneider CP, Shukla D, Trout BL (2011) Effects of Solute-Solute Interactions on Protein Stability Studied Using Various Counterions and Dendrimers. PLoS ONE 6(11): e27665. https://doi.org/10.1371/journal.pone.0027665
Editor: Maria Gasset, Consejo Superior de Investigaciones Cientificas, Spain
Received: July 14, 2011; Accepted: October 21, 2011; Published: November 18, 2011
Copyright: © 2011 Schneider et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The Singapore-MIT Alliance funded this research. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Understanding how solution components interact with proteins and modulate biological processes is essential in applications such as researching methods for stabilizing protein based therapeutics and treating neurodegenerative diseases resulting from amyloid formation . Dendrimers are a relatively new class of materials with much research already devoted toward possible biomedical applications, which is quite promising –. The interaction of dendrimers with proteins is a classic example of polyvalent interactions, which is a common form of interaction between biological entities, such as the interaction between receptors and ligands, the interaction between a virus and a cell surface, etc. . Such interactions lead to a varying array of resulting effects on protein behavior. In certain cases, polycationic dendrimer molecules destabilize protein conformations ,  and enhance the formation of amorphous aggregates . However, the same dendrimer species have also been shown to dissolve amyloid fibrils and inhibit their formation . While in other cases, strong adhesive interactions with oxyanionic groups in proteins stabilize protein assemblies, such as microtubules . In several of these cases, derivative dendrimer species were produced by modifying the surface to guanidinium , , a functional group known to preferentially bind to protein surfaces .
When considering the behavior of solution additives or macromolecular species with charged surfaces, the influence of the counterion is often overlooked. The influence individual ions have on protein stability has been empirically ranked for more than a century in the well-known Hofmeister Series  and correlated with respect to preferential interactions . Furthermore, it has long been believed that the net effect of a salt is the additive effect of each ion . However, recent investigations suggest that for the particular case of guanidinium bearing compounds, the strength of attractive ion-ion interactions are the cause of varying behavior among different guanidinium salt forms –. This model explains the neutral behavior of guanidinium sulfate without the need for an unfavorable separation of charge . That is, rather than a bound guanidinium cation and an excluded sulfate anion, the attractive interaction between guanidinium and sulfate causes clustering, which interferes with the binding of guanidinium to the protein surface. More importantly, such interactions will be of particular importance for compounds containing multiple guanidinium groups, such as the previously mentioned dendrimers. If the net effect of combining ions is purely an additive effect, then exchanging the counterion should not change direct protein-dendrimer interactions or the behavior of dendrimer molecules in solution. However, if the recently revealed interactions are correct, exchanging the counterion will significantly alter the behavior of the guanidinium compounds as the effects of ion-ion interactions will be amplified.
Here, we present a study which shows that attractive guanidinium-anion interactions strongly influence the solution behavior of guanidinium modified PAMAM dendrimers. As with other similar compounds, the guanidinium chloride (GdmCl) form disrupted attractive protein-protein interactions at low concentrations but reduced thermostability, which led to enhanced aggregation. However, the aggregation suppression by the sulfate and dihydrogen phosphate salt forms was more significant and observed at all concentrations. They slowed the rate of aggregation of model proteins (α-Chymotrypsinogen A and Concanavalin A) to about 2% of the original aggregation rate at concentrations as low as 0.2 mol/L, which is around 10 times slower than when in the presence of arginine HCl or other aggregation suppressing excipients (e.g. sucrose, glycerol, etc.). Preferential interaction and computational studies of the modified generation 0 dendrimer salts in α-Chymotrypsinogen A aqueous solutions establish that attractive ion-ion interactions alter how the dendrimers interact with each other and with proteins through the formation of clusters. Such behavior is also observed for the unmodified ammonium surface but to a much lesser degree , demonstrating the specific nature of the attractive ion-ion interactions. Understanding the solute-solute interactions presented in this study gives valuable insight into the overall understanding of how ion interactions influence the behavior of macromolecular compounds with polycationic surfaces.
The most notable consequence of perturbing a protein's environment is the enhancement or inhibition of protein aggregation , . The guanidinium modified dendrimers were added to model protein solutions and incubated at an elevated temperature to determine how they influence the rate of this degradation pathway. Figure 1a shows α-chymotrypsinogen A (aCgn) monomer loss profiles, as determined by size exclusion HPLC, for solutions containing a generation 0 PAMAM dendrimer with the surface modified to GdmCl. At low dendrimer concentrations (0.05 M), the rate of monomer loss in the presence of the surface modified dendrimer is slower than when compared to the reference solution. However, this aggregation rate reduction is insignificant when compared to the rate reduction induced by a high concentration of other commonly used additives such as arginine hydrochloride (ArgHCl), which is also depicted in the figure. Furthermore, as the concentration of the surface modified dendrimer is increased, the aggregation rate reduction decreases until ultimately, the rate of aggregation is increased. At a concentration of 0.2 M, the surface modified dendrimer induces rapid aggregation, causing a 50% loss in about 10 minutes as opposed to 30 minutes for the solution containing no cosolute. These results only become worse for higher generations. As shown in Figure 1b, the rate of monomer loss for the generation 1 dendrimer with a surface modified to GdmCl indicates a large increase in the rate of aggregation at concentrations as low as 0.05 M, even though at lower concentrations the compound inhibits aggregation by a moderate amount.
For all experiments, samples were incubated at 52.5°C, initial monomer concentration, M0, was 10 mg/mL, all solutions were prepared in a 20 mM sodium citrate pH 5 buffer, and all rate loss profiles fitted to a 2nd order rate law. (a) Monomer loss profiles for solutions containing Generation 0 PAMAM dendrimers with guanidinium chloride or H2PO4 surfaces at varying concentrations. (b) Rate constant, k, for aCgn monomer loss relative to the rate constant for no additive, k0, versus additive concentration.
These results are comparable to other large compounds with surfaces modified to Gdm, which exhibit a strong interaction with proteins that results in destabilization at moderate to high concentrations , . However, a previous inquiry into various arginine salts  showed that the interaction between a Gdm functional group and a protein is strongly influenced by the counterion to the Gdm moiety. The reason for this is that hydrogen bond accepting anions will tend to form strong hydrogen bonds with the hydrogen bond donating Gdm group. Thus the interaction between the modified dendrimers and the protein can be altered by exchanging chloride with counterions such as sulfate, phosphate, citrate, acetate, etc., which are more capable of accepting hydrogen bonds .
The results shown in Figure 1a reveal that a generation 0 PAMAM dendrimer with a surface modified to guanidinium dihydrogen phosphate (H2PO4) has an ability to slow the rate of aCgn aggregation far beyond that of ArgHCl. To elaborate, with no cosolute present, half of the original amount of protein is lost within only 30 minutes. When in the presence of a high concentration of ArgHCl (0.67 M), the half-life is extended to about 2.5 hours, which is similar to other commonly used excipients. However, when in the presence of the H2PO4 form of the surface modified dendrimer at a concentration of 0.2 M, the half-life is extended to about 25 hours (see Figure S1 which shows extended data), which is a full order of magnitude longer than the solution containing ArgHCl. This superior aggregation suppression is observed at all concentrations for that salt form, as shown in Figure 1b. In that figure, the relative rate constant for aCgn monomer loss is depicted, which is the observed rate constant when a cosolute is added to the solution relative to the rate of monomer loss in a buffer only solution. The figure also reveals that aggregation suppression improves with increasing size of the dendrimer, as indicated by the monotonic improvement in aggregation suppression with each dendrimer generation.
Improved aggregation suppression is also observed when the counterion is exchanged to other hydrogen bond accepting anions. When exchanged to acetate, aggregation suppression is only improved slightly, which was anticipated given that acetate cannot form as many hydrogen bonds with Gdm as compared to other ions . When chloride is exchanged with sulfate though, the resulting compound shows nearly identical ability for inhibiting aggregation as the H2PO4 salt form. The sharp decline in the rate of aCgn aggregation prompted us to study dendrimer-aCgn interactions, both experimentally and computationally, to give greater mechanistic insight into the observed behavior. However, we first expanded the aggregation study to determine if the observations are observed for other proteins. Concanavalin A (Con A) was used as another model protein for the aggregation study, which demonstrated that the shelf-life of this protein at pH 6.5 and 37°C is extended by a factor greater than 15 when in the presence of the generation 1 dendrimer with a guanidinium sulfate surface (see Table 1), which is over 10 times longer than when in the presence of commonly used additives such as glycerol or sucrose. Table 1 depicts the factor by which the shelf-life of aCgn and Con A is extended when they are formulated with the modified dendrimers. This Shelf-Life Extension Factor was determined by comparing the length of time, t95, for a 5% loss of protein when in the presence of the compounds to the original length of time, t95,0 for a 5% loss. For comparison purposes, isotonic concentrations (as determined by VPO) of the compounds were used and Shelf-Life Extension Factor values for commonly used excipients at isotonic concentrations are shown as well. Shelf-life values of aCgn were determined at 52.5°C and at 37°C for Con A. It is clear from these results, that when utilized at a practical maximum concentration, the surface modified dendrimers, in the form of either a dihydrogen phosphate or sulfate salt, significantly improve the shelf life of these two proteins, either at high or moderate temperatures. For aCgn, the shelf life is extended by a factor between 16 and 27 when formulated with these dendrimers, which is 5 to 8 times longer than when in the presence of other aggregation suppressing additives, such as arginine HCl, sucrose, or sodium sulfate.
The results for Con A are more significant and show a much clearer relationship with the size of the dendrimers. At 37°C and pH 6.5, Con A aggregates quite rapidly. Commonly used additives can extend the shelf life, at most, by factor of only 1.5. The sulfate form of the generation 0 modified dendrimer can quadruple this to a factor of 5.9 and the sulfate form of the generation 1 modified dendrimer extends the shelf life even further, by a factor of 16.7. The phosphate form only shows a minimal improvement in the shelf life, likely due to Con A being very sensitive to ionic strength at pH 6.5 because arginine HCl, sodium chloride, and sodium phosphate all speed up Con A aggregation. The sulfate form of the dendrimers has fewer ions per mole than the phosphate form and therefore, the detrimental effect imparted on proteins that are sensitive to ionic strength is lessened when this form of the dendrimer is used.
It should be noted that these results do not extend completely to the original, unmodified dendrimer structure, which has an ammonium surface. The chloride form of the unmodified dendrimer is more destabilizing and the phosphate form is less effective at suppressing aggregation (see Figure 1b). This indicates that a Gdm surface is a necessity to produce the potent aggregation suppressing results through both protein-additive and ion-ion interactions.
The thermostability of aCgn at 1 mg/mL in the presence of the modified dendrimers was assessed by determining the denaturation midpoint temperature (Tm) from DSC scans, which is a qualitative indicator of how the conformational stability of the protein is perturbed . As shown in Table 2, the sulfate and H2PO4 salt forms of the surfaced modified generation 0 dendrimers increase Tm at a rate of 15.2 and 37.4°C*M−1, respectively, for concentrations less than 0.2 mol/L, while the chloride salt form decreases Tm at a rate of 13.9°C*M−1. One can speculate that this indicates that the sulfate and H2PO4 salt forms shift the protein folding equilibrium toward the native structure while the chloride salt form promotes unfolding. However, given that the unfolding of aCgn is irreversible, it could also indicate that the sulfate and H2PO4 salt forms reduce the rate at which aCgn aggregates during the DSC scan. The apparent thermodynamic stabilization by these forms is quite significant when compared to other conformational stabilizers (e.g. sucrose)  given that the results likely represent a combination of conformational stabilization and association suppression. The rate at which the chloride salt form lowers the melting temperature of aCgn is double that for ordinary GdmCl  and given that this dendrimer salt form inhibits aggregation at low concentrations, this shows that this surface modified dendrimer is a powerful denaturant.
MD simulations were conducted on aqueous solutions of the modified generation 0 dendrimers to quantify how ion-ion interactions may be influencing the behavior of the additives (see Table S1 for a description of the setup of each simulation). In Figure 2a, the Radial Distribution Functions (RDF) between the dendrimer and the counterions show that the sulfate and H2PO4 ions interact strongly with the dendrimer molecules, as shown by the height of the peaks relative to chloride. In Figure 2b, the RDF's between dendrimer molecules indicate that in the presence of chloride ions, dendrimer molecules do not interact with each other, however, the presence of sulfate and H2PO4 ions tends to bring dendrimer molecules together. This is further supported by MD snapshots of the simulation box (Figure 3), which show significant ion pairing in the sulfate and H2PO4 salt solutions while solutes in the chloride solution are randomly distributed. These results indicate that the Gdm group on the dendrimers can form charge-assisted hydrogen bonds with the sulfate and H2PO4 ions. The sulfate ion, which has a −2 charge on four oxygens, forms a much stronger hydrogen bond as compared to the H2PO4, which has −1 charge.
(a) RDF's between dendrimer and counterions and (b) between dendrimer molecules in different dendrimer salt solutions. The distance between the centers of mass of the dendrimers is used for calculation of the RDF's. For the counterions, the sulfur atom in sulfate, phosphorus atom in H2PO4 are utilized.
The counterion is either chloride (left), H2PO4 (middle) or sulfate (right). The dendrimer molecules are shown in Licorice style and counterions are shown as VdW spheres. The hydrogen atoms are not shown to improve the clarity.
These results also show that counterions can act as a bridge between dendrimer molecules due to attractive guanidinium-anion interactions, leading to the formation of large clusters in solution (see Figure 3). To verify and quantify this observation, the number of hydrogen bonds formed between different ion-pairs in aqueous modified dendrimer salt solutions was calculated from the simulation results (see Table 2). Sulfate and H2PO4 ions, due to the presence of multiple hydrogen bond donors and acceptors, indeed act as a bridge joining dendrimer molecules together. The number of hydrogen bonds for both salt types (∼150) is nearly an order of magnitude more than that for the chloride form (∼20), leading to numerous bridged interactions (73 to 124), which is almost nonexistent for the chloride form. These guanidinium-anion and bridged interactions have a direct impact on the number of dendrimer-protein interactions, reducing the number by nearly half when compared to the chloride form. The extent of clustering in these solutions can also be quantified in terms of the loss of the solvent-accessible area (SAA) of dendrimer molecules, as shown in Table 3. The loss of SAA due to clustering is greatest for H2PO4 (∼60%), followed by sulfate (∼40%) and chloride (20%). In the case of chloride, the loss of SAA is mainly due to the presence of counterions near the dendrimer. For sulfate and H2PO4, the dominant component to the loss of SAA is due to the overlap of dendrimer molecules. The number of H2PO4 ions is twice the number of sulfate ions per dendrimer molecule, which contributes to the higher loss of SAA as compared to sulfate.
To gain insight into how the modified dendrimer salts inhibit protein-protein interactions, preferential interaction coefficient, Γ23, values at various concentrations were determined, both experimentally via vapor pressure osmometry (VPO) measurements, and computationally via MD simulations. The experimental results for the interaction between modified generation 0 PAMAM dendrimers and aCgn are expressed in Table 2, which summarizes the polynomial fit and uncertainty of the experimental data. Theoretical preferential interaction coefficient values were computed from the MD simulation (see Figure S2, which depicts the convergence of simulated values) using the procedure outlined in our previous work ,  and the results are presented in Table 4.
At a concentration of 0.18 mol/L, the theoretical preferential interaction coefficient for the chloride salt is found to be −0.2±1, which matches well with the experimental value of −1.5±0.7. The Γ23 values for salts are a weighted average of the Γ23 values for individual ions. Γ23 for the dendrimer cation was found to have a positive value of 1, which shows that the local concentration of dendrimer molecules around the protein is higher than the bulk concentration. However, due to the negative preferential interaction value for the chloride ion (−7), the overall preferential interaction coefficient was found to be negative. The observed preferential binding of the dendrimer cation stems from the fact that the modified dendrimers can interact favorably with a variety of amino acids on the protein surface due to the presence of the Gdm group, which can form hydrogen bonds with negatively charged amino acids and the protein backbone and can also interact with aromatic amino acids via cation-π interactions. Furthermore, the dendrimer molecule can bind cooperatively with the protein surface due to multiple Gdm surface groups simultaneously interacting with the protein surface (see Figure 4, which shows a snapshot of multiple, simultaneous interactions). However, switching the counterion to either sulfate or H2PO4 inhibits the occurrence of such multiple interactions.
The guanidinium groups can hydrogen bond with negatively charged amino acids and the peptide backbone. They can also form cation-π interaction with aromatic amino acids.
Γ23 values for the sulfate (−2.7) and H2PO4 (−2.3) salt forms match well with their corresponding experimental values. The values of Γ23 for the counterions are −10 for the H2PO4 ion and −7 for the sulfate ion, which is present in half the quantity as the phosphate and chloride ions. On the basis of the observed attractive interaction between the dendrimer and these counterions, it can be argued that sulfate and H2PO4 inhibit the dendrimer molecule from binding to the protein surface. In essence, the dendrimer molecules are pulled away from the surface to interact with bulk solution components. This is verified by the individual Γ23 values for the dendrimer molecule (−3 for both salt types). These results are similar to the results of our recent work on the interaction of arginine with proteins, where the carboxylate group and various counterions limited the interaction between a protein and the Gdm group in arginine –. As mentioned before, the reduced number of hydrogen bonds between the protein and the dendrimer (see Table 2) further supports this behavior. The loss in the number of direct hydrogen bonds is compensated by the increase in the number of indirect hydrogen bonds formed between the protein and the dendrimer in which the counterion acts as a bridge.
RDF's between the four dendrimer arms and the protein surface (see Figure 5) highlight the implications of the counterions interacting with the Gdm groups. The RDF for the closest arm remains almost the same for all dendrimer salts but the RDF's for the remaining arms show a sharp decrease in peak height and increased distance from the surface of the protein for the sulfate and H2PO4 salt forms. This result further supports that for the dendrimer with a GdmCl surface, multiple arms simultaneously interact with the protein surface but for the sulfate and H2PO4 salts, only one arm can interact with the protein while the other arms face away from the surface and interact with the bulk solution. Furthermore, for the sulfate and H2PO4 salts, there are additional peaks further away from the surface for the closest dendrimer arm, which is the result of the anions acting as a bridge between the protein and the dendrimer. This interaction with the Gdm group is clearly impeding direct binding of the dendrimer to the protein surface.
The counterion is either chloride (left), H2PO4 (middle) or sulfate (right). The arms of the dendrimer are labeled 1–4 depending on their distance from the protein surface, with 1 denoting the closest arm. The distance of the central carbon atom in the guanidinium group from the protein surface is used for the calculations.
The original intent of this work was to determine if attractive protein-protein interactions could be inhibited by large additives which tend to crowd the local domain around proteins rather than being excluded to the bulk solution . It was theorized that the crowding could be promoted by counteracting repulsive steric exclusion interactions (which are significant for large additives) with the attractive interactions that occur between a protein and a denaturant. That is, creating a balance of attractive and repulsive interactions that lead to a net-neutral interaction. If such a balance occurs, the resulting compound would interfere with protein-protein interactions with little influence on conformational stability , . Given the suppression of protein aggregation at low concentrations, the results presented here show that large compounds (i.e. dendrimers) with protein-binding functional groups (i.e. GdmCl) on their surfaces disrupt protein-protein interactions due to an attractive interaction with the protein. However, for the particular case of guanidinium chloride modified PAMAM dendrimers, the net attractive interaction seems to be too strong given the conformational destabilization and enhanced aggregation at higher concentrations. The compounds can inhibit aggregation at a level comparable to other commonly used excipients but only at low concentrations.
From these results, it is obvious that volume exclusion effects (see Table 2 and Table S2, which give values for the molecular weight and partial molar volume of the modified dendrimers, which are larger by several factors than most small molecule additives) are not counteracting the preferential binding of the surface groups to the extent anticipated. Preferential binding is predicted to scale with the number of binding groups per area in accordance with the frequency of single binding interactions with the protein surface, while exclusion is known to scale with the volume of the additive . Thus, if the size of an additive increases while the density of surface groups remains constant, it was predicted that steric exclusion would dominate. However, this does not take into account the total energy of binding nor structural flexibility, which can enhance the density of surface groups. As demonstrated by “molecular glue” compounds , which also have multiple Gdm surface groups, the larger and the more flexible the compound, the stronger it binds to proteins. This indicates that the large and flexible nature of the surface modified dendrimers allows for a cooperative interaction of the multiple Gdm groups with the surface of the model protein, which we verified through MD simulations. This attractive interaction is likely stronger for the unfolded state, when more binding sites are exposed and the positive electrostatic charge on the protein surface is distributed over a larger area, enhancing the amount of preferential binding and thus denaturing the protein.
However, as demonstrated by the stabilizing effect and hydrogen bond interactions of the H2PO4 and sulfate salt forms, ion-ion interactions between the Gdm functional groups and the counterions influence how surface modified dendrimers interact with proteins, in addition to how dendrimer molecules interact with each other in solution. The RDF results show that for these alternate salt forms, the interaction between the protein and the dendrimer is determined by the frequency of single guanidinium-protein interactions (i.e. a single strong guanidinium-protein peak), rather than multiple simultaneous interactions (i.e. multiple strong guanidinium-protein peaks), a behavior more consistent with the predicted outcome of making large compounds with protein-binding groups. The ion-ion interactions described here have only recently been taken into consideration in describing the solution behavior of different guanidinium salt forms. The MD simulations performed verify that attractive ion-ion interactions bridge together multiple dendrimer molecules into clusters. This clustering leads to three important impications: (i) these clusters enhance the effective size of the additives in solution, with the size of the additive determining its ability to crowd out protein-protein interactions, (ii) these clusters are expected to reduce the mobility of the proteins in solutions due to a network of large hydrogen-bonded clusters around them (see Figure 6, which shows snapshots of aCgn in the different dendrimer solutions) which should reduce the rate of protein-protein encounters , and (iii) the formation of these clusters influences the interaction between protein and dendrimer molecules. It can be observed directly from the simulations that the cooperative binding of multiple dendrimer arms to the protein surface is inhibited by the cluster formation. This has a direct result in reducing the preferential interaction of the dendrimer molecule and eliminating the denaturing effect.
Water molecules and hydrogen atoms are not shown to improve clarity. Dendrimer molecules are shown in licorice representation and counterions are shown in van der Waals representation.
The trends discussed here are related to the impact these salts have on protein aggregation. Theoretical preferential interaction coefficient values for an inert compound the same size as the modified generation 0 dendrimer but lacking any ability to form attractive interactions shows a preferential exclusion over four times greater than the modified dendrimers (see Figure S3). It is clear from these results that even for the sulfate and H2PO4 salt forms, the surface modified dendrimers can be considered to be only slightly excluded when compared to how excluded they would be without any protein-binding surface groups. From this perspective, the preferential interaction of the surface modified dendrimers can be considered approximately net-neutral. Also considering that all of the salt forms inhibit protein aggregation at low additive concentrations, such results support the hypothesis that surface modified dendrimers are able to inhibit aggregation, in part, by slowing protein association through a disruption of protein-protein interactions. A highly excluded compound would not exhibit much of an effect on association due to a depletion of cosolute molecules in the local domain of the protein. In fact, large and highly excluded compounds often induce association due to a colloidal depletion force , . These results indicate that the counterion plays a critical role in fine tuning the attraction between protein and additive molecules, such that the extent of binding of the modified dendrimer molecule is different among the different salt forms and in certain cases, the attractive interaction between the additives and a protein is strong enough to inhibit protein-protein interactions but not strong enough to denature the protein. This is supported by Figure S4, which depicts a close-up view (within 0.6 nm) of the protein surface showing that the guanidinium chloride modified dendrimers bind to the surface of the protein with no guanidinium-chloride interaction while the guanidinium sulfate modified dendrimers crowd around the protein surface with little binding due to an interaction with sulfate. The exact contribution of inhibiting protein-protein interactions cannot be determined given that the sulfate and H2PO4 salt forms may also provide conformational stability. However, the sharp decline in the rate of aggregation at low dendrimer concentrations suggests it is a significant contribution, which is further supported by the order of magnitude improvement in the reduction of aggregation when compared to other conformational stabilizers, such as sucrose.
In conclusion, using α-Chymotrypsinogen and Concanavalin A as a model proteins, we investigated the aggregation suppressing performance of PAMAM dendrimers with surfaces modified to a variety of guanidinium salts and give a molecular level mechanistic insight into the behavior of this new class of additive. The most significant observation was that attractive additive-additive interactions dominated the behavior of the dendrimer molecules. The results presented indicate that all of the dendrimers form an attractive interaction with aCgn, leading to suppressed protein-protein interactions, which is more significant than other additives due to the size of the dendrimer molecules. The dendrimers with guanidinium chloride surfaces suppressed aggregation at low concentrations but DSC scans indicate that the additive promotes aCgn unfolding, leading to enhanced aggregation at high concentrations. Under conditions when the conformation of aCgn is not destabilized (i.e. sulfate and H2PO4 counterions), the large molecules are capable of significantly reducing the rate of aggregation at all concentrations. This stems from the behavior resulting from attractive guanidinium-anion interactions, which are lacking for the guanidinium chloride modified dendrimers. As indicated during molecular simulation snapshots, attractive guanidinium-sulfate/H2PO4 interactions cause dendrimer molecules to form clusters in solution and in return, inhibit multiple dendrimer arms from simultaneously binding to the protein, as indicated by radial distribution function plots. This reduced level of preferential binding producing a scenario in which the additive clusters solvate the surface of the protein, which reduces protein-protein interactions, without promoting unfolding. The elucidation of this particular type of additive gives insight into the behavior of PAMAM dendrimers in general, but more importantly, it demonstrates the role additive-additive interactions play in proteins stability.
Materials and Methods
Generation 0 through Generation 2 Dendritech® PAMAM Dendrimers with ethylenediamine cores, Bovine α-Chymotrypsinogen A type II (C4879), and jack bean Concanavalin A (C2010) were obtained from Sigma-Aldrich (St. Louis, MO). All other reagents were obtained from Sigma-Aldrich in the highest available grade. The concentration of aCgn and Con A were determined spectrophotometrically using extinction coefficient of 1.97 mL*mg−1 cm−1 at 282 nm and 1.37 mL*mg−1 cm−1 at 282 nm, respectively. All aCgn samples were pretreated with the enzymatic inhibitor phenylmethylsulfonyl fluoride and dialyzed against 20 mM sodium citrate, pH 5.
Dendrimer Surface Modification
The PAMAM dendrimer surface amine groups were guanylated with an excess of 1,3-Bis(tert-butoxycarbonyl)-2-methyl-2-thiopseudourea in dimethylformamide (DMF). The DMF was evaporated and the residue dissolved in diethyl ether. The product was purified by repeated precipitation with n-hexane. The BOC protecting groups were removed with 4 M HCl dissolved in dioxane and the resulting salts were washed with acetone. For alternate salt forms, the counterions were exchanged using Amberlite IRA 400 anion exchange resin loaded using the appropriate sodium salt. The purity and structure were analyzed with NMR and mass spectrometry.
The aggregation of aCgn and Con A were accelerated by incubating samples at an elevated temperature in a Bio-Rad MyCycler thermal cycler. Aggregate formation and monomer loss was monitored using an Agilent 1200 series HPLC, equipped with a Zorbax GF-250 (4.6×250 mm, 4 micron) size exclusion column and a UV-Vis detector.
Differential Scanning Calorimetry
The thermodynamic stability of 1 mg/mL solutions of aCgn in the presence of the modified dendrimers was determined by differential scanning calorimetry (Microcal VP-Differential Scanning Calorimeter) using a scan rate of 90°C/hour.
Molecular Dynamics Simulations
Molecular dynamics (MD) simulations of aqueous solutions of the guanidinium modified generation 0 dendrimer salts with and without aCgn (PDB Id: 2CGA) were performed using NAMD 2.7 , with CHARMM27  force fields and the TIP3P  water model. The force field parameters for the counterions were taken from the literature  and the force field parameters for the surface modified generation 0 dendrimer were developed using the CHARMM force field development procedure .
Preferential Interaction Coefficient
Theoretical preferential interaction coefficient (Γ23) values were calculated using a statistical mechanical method applied to and all-atom model with no adjustable parameters . Experimental values were obtained from changes in water activity as determined by vapor pressure osmometry . Please see the Text S1 for complete details of all methods utilized.
Supporting Information Text. Complete details of experimental and computational methods and results for NMR and mass spectroscopy analysis.
Convergence of preferential interaction coefficient (Γ23) of α-Chymotripsinogen A in aqueous dendrimer (GdmCl surface) salt solution. The first 10 ns of instantaneous data are not used for calculation of cumulative averages.
The influence of generation 0 PAMAM dendrimers, with surfaces modified to guanidinium, on aCgn monomer loss due to aggregation at 52.5°C. The figure depicts aCgn monomer concentration, M, versus time relative to the initial monomer concentration, M0, of 10 mg/mL, with all solutions prepared in a 20 mM sodium citrate pH 5 buffer and all profiles fitted to a 2nd order rate law. (A) Monomer loss profiles for solutions containing the guanidinium chloride salt form at varying concentrations. (B) Monomer loss profile for a solution containing the guanidinium dihydrogen phosphate salt form at a concentration of 0.2 M. The profiles for a solution containing no additive and a solution containing arginine hydrochloride at a concentration of 0.67 M are included for comparison.
Preferential Interaction Coefficient, Γμ3, values versus additive concentration. Values are for the interaction between generation 0 PAMAM dendrimers, with surfaces modified to guanidinium, and aCgn, as determined from VPO measurements. Error bars left off for clarity and curves drawn through the plots to aid the eye (see Table S2 for more detail).
Snapshots of PAMAM dendrimer and counter-ions within 0.6 nm of the protein surface. Guanidinium chloride surface (left) and Guanidinium sulfate surface (right).
Summary of additive molecular weight, partial molar volume (V), preferential interactions with aCgn as determined by VPO, and aCgn denaturation midpoint temperature increment as determined by DSC for surface modified PAMAM dendrimers.
Conceived and designed the experiments: CPS DS BLT. Performed the experiments: CPS DS. Analyzed the data: CPS DS BLT. Wrote the paper: CPS DS BLT. Performed laboratory research: CPS. Performed computational research: DS.
- 1. Frokjaer S, Otzen DE (2005) Protein drug stability: A formulation challenge. Nat Rev Drug Discovery 4: 298–306.S. FrokjaerDE Otzen2005Protein drug stability: A formulation challenge.Nat Rev Drug Discovery4298306
- 2. Lee CC, MacKay JA, Frechet JMJ, Szoka FC (2005) Designing dendrimers for biological applications. Nat Biotechnol 23: 1517–1526.CC LeeJA MacKayJMJ FrechetFC Szoka2005Designing dendrimers for biological applications.Nat Biotechnol2315171526
- 3. Gajbhiye V, Palanirajan VK, Tekade RK, Jain NK (2009) Dendrimers as therapeutic agents: a systematic review. J Pharm Pharmacol 61: 989–1003.V. GajbhiyeVK PalanirajanRK TekadeNK Jain2009Dendrimers as therapeutic agents: a systematic review.J Pharm Pharmacol619891003
- 4. Labieniec M, Watala C (2009) PAMAM dendrimers - diverse biomedical applications. Facts and unresolved questions. Cent Eur J Biol 4: 434–451.M. LabieniecC. Watala2009PAMAM dendrimers - diverse biomedical applications. Facts and unresolved questions.Cent Eur J Biol4434451
- 5. Boas U, Heegaard PMH (2004) Dendrimers in drug research. Chem Soc Rev 33: 43–63.U. BoasPMH Heegaard2004Dendrimers in drug research.Chem Soc Rev334363
- 6. Mammen M, Choi SK, Whitesides GM (1998) Polyvalent interactions in biological systems: Implications for design and use of multivalent ligands and inhibitors. Angew Chem-Int Edit 37: 2755–2794.M. MammenSK ChoiGM Whitesides1998Polyvalent interactions in biological systems: Implications for design and use of multivalent ligands and inhibitors.Angew Chem-Int Edit3727552794
- 7. Klajnert B, Stanislawska L, Bryszewska M, Palecz B (2003) Interactions between PAMAM dendrimers and bovine serum albumin. BBA-Proteins Proteomics 1648: 115–126.B. KlajnertL. StanislawskaM. BryszewskaB. Palecz2003Interactions between PAMAM dendrimers and bovine serum albumin.BBA-Proteins Proteomics1648115126
- 8. Giehm L, Christensen C, Boas U, Heegaard PAH, Otzen DE (2008) Dendrimers destabilize proteins in a generation-dependent manner involving electrostatic interactions. Biopolymers 89: 522–529.L. GiehmC. ChristensenU. BoasPAH HeegaardDE Otzen2008Dendrimers destabilize proteins in a generation-dependent manner involving electrostatic interactions.Biopolymers89522529
- 9. Heegaard PMH, Boas U, Otzen DE (2007) Dendrimer effects on peptide and protein fibrillation. Macromol Biosci 7: 1047–1059.PMH HeegaardU. BoasDE Otzen2007Dendrimer effects on peptide and protein fibrillation.Macromol Biosci710471059
- 10. Okuro K, Kinbara K, Tsumoto K, Ishii N, Aida T (2009) Molecular Glues Carrying Multiple Guanidinium Ion Pendants via an Oligoether Spacer: Stabilization of Microtubules against Depolymerization. Journal of The American Chemical Society 131: 1626–1627.K. OkuroK. KinbaraK. TsumotoN. IshiiT. Aida2009Molecular Glues Carrying Multiple Guanidinium Ion Pendants via an Oligoether Spacer: Stabilization of Microtubules against Depolymerization.Journal of The American Chemical Society13116261627
- 11. Cordes H, Boas U, Olsen P, Heegaard PMH (2007) Guanidino- and urea-modified dendrimers as potent solubilizers of misfolded prion protein aggregates under non-cytotoxic conditions. Dependence on dendrimer generation and surface charge. Biomacromolecules 8: 3578–3583.H. CordesU. BoasP. OlsenPMH Heegaard2007Guanidino- and urea-modified dendrimers as potent solubilizers of misfolded prion protein aggregates under non-cytotoxic conditions. Dependence on dendrimer generation and surface charge.Biomacromolecules835783583
- 12. Mason PE, Neilson GW, Dempsey CE, Barnes AC, Cruickshank JM (2003) The hydration structure of guanidinium and thiocyanate ions: Implications for protein stability in aqueous solution. Proc Natl Acad Sci U S A 100: 4557–4561.PE MasonGW NeilsonCE DempseyAC BarnesJM Cruickshank2003The hydration structure of guanidinium and thiocyanate ions: Implications for protein stability in aqueous solution.Proc Natl Acad Sci U S A10045574561
- 13. Hofmeister F (1888) Zur Lehre von der Wirkung der Salze. Arch Exp Pathol Pharmakol 24: 247–260.F. Hofmeister1888Zur Lehre von der Wirkung der Salze.Arch Exp Pathol Pharmakol24247260
- 14. Timasheff SN (2002) Protein-solvent preferential interactions, protein hydration, and the modulation of biochemical reactions by solvent components. Proc Natl Acad Sci U S A 99: 9721–9726.SN Timasheff2002Protein-solvent preferential interactions, protein hydration, and the modulation of biochemical reactions by solvent components.Proc Natl Acad Sci U S A9997219726
- 15. Dempsey CE, Mason PE, Brady JW, Neilson GW (2007) The Reversal by Sulfate of the Denaturant Activity of Guanidinium. J Am Chem Soc 129: 15895–15902.CE DempseyPE MasonJW BradyGW Neilson2007The Reversal by Sulfate of the Denaturant Activity of Guanidinium.J Am Chem Soc1291589515902
- 16. Mason PE, Dempsey CE, Neilson GW, Brady JW (2005) Nanometer-Scale Ion Aggregates in Aqueous Electrolyte Solutions: Guanidinium Sulfate and Guanidinium Thiocyanate. J Phys Chem B 109: 24185–24196.PE MasonCE DempseyGW NeilsonJW Brady2005Nanometer-Scale Ion Aggregates in Aqueous Electrolyte Solutions: Guanidinium Sulfate and Guanidinium Thiocyanate.J Phys Chem B1092418524196
- 17. Mason PE, Dempsey CE, Vrbka L, Heyda J, Brady JW, et al. (2009) Specificity of Ion-Protein Interactions: Complementary and Competitive Effects of Tetrapropylammonium, Guanidinium, Sulfate, and Chloride Ions. J Phys Chem B 113: 3227–3234.PE MasonCE DempseyL. VrbkaJ. HeydaJW Brady2009Specificity of Ion-Protein Interactions: Complementary and Competitive Effects of Tetrapropylammonium, Guanidinium, Sulfate, and Chloride Ions.J Phys Chem B11332273234
- 18. Shukla D, Trout BL (2010) Interaction of Arginine with Proteins and the Mechanism by Which It Inhibits Aggregation. J Phys Chem B 114: 13426–13438.D. ShuklaBL Trout2010Interaction of Arginine with Proteins and the Mechanism by Which It Inhibits Aggregation.J Phys Chem B1141342613438
- 19. Shukla D, Trout BL (2011) Preferential Interaction Coefficients of Proteins in Aqueous Arginine Solutions and Its Molecular Origins. J Phys Chem B 115: 1243–1253.D. ShuklaBL Trout2011Preferential Interaction Coefficients of Proteins in Aqueous Arginine Solutions and Its Molecular Origins.J Phys Chem B11512431253
- 20. Schneider CP, Shukla D, Trout BL (2011) Arginine and the Hofmeister Series: The Role of Ion-Ion Interactions in Protein Aggregation Suppression. J Phys Chem B 115: 7447–7458.CP SchneiderD. ShuklaBL Trout2011Arginine and the Hofmeister Series: The Role of Ion-Ion Interactions in Protein Aggregation Suppression.J Phys Chem B11574477458
- 21. Shukla D, Schneider CP, Trout BLEffects of PAMAM Dendrimer Salt Solutions on Protein Stability. J Phys Chem Lett. D. ShuklaCP SchneiderBL TroutEffects of PAMAM Dendrimer Salt Solutions on Protein Stability.J Phys Chem LettIn Press. In Press.
- 22. Deyoung LR, Fink AL, Dill KA (1993) Aggregation of Globular Proteins. Accounts Chem Res 26: 614–620.LR DeyoungAL FinkKA Dill1993Aggregation of Globular Proteins.Accounts Chem Res26614620
- 23. Fink AL (1998) Protein aggregation: folding aggregates, inclusion bodies and amyloid. Fold Des 3: R9–R23.AL Fink1998Protein aggregation: folding aggregates, inclusion bodies and amyloid.Fold Des3R9R23
- 24. Ghosh K, Dill KA (2009) Computing protein stabilities from their chain lengths. Proc Natl Acad Sci U S A 106: 10649–10654.K. GhoshKA Dill2009Computing protein stabilities from their chain lengths.Proc Natl Acad Sci U S A1061064910654
- 25. Baynes BM, Trout BL (2003) Proteins in mixed solvents: A molecular-level perspective. J Phys Chem B 107: 14058–14067.BM BaynesBL Trout2003Proteins in mixed solvents: A molecular-level perspective.J Phys Chem B1071405814067
- 26. Shukla D, Shinde C, Trout BL (2009) Molecular Computations of Preferential Interaction Coefficients of Proteins. J Phys Chem B 113: 12546–12554.D. ShuklaC. ShindeBL Trout2009Molecular Computations of Preferential Interaction Coefficients of Proteins.J Phys Chem B1131254612554
- 27. Baynes BM, Trout BL (2004) Rational design of solution additives for the prevention of protein aggregation. Biophys J 87: 1631–1639.BM BaynesBL Trout2004Rational design of solution additives for the prevention of protein aggregation.Biophys J8716311639
- 28. Baynes BM, Wang DIC, Trout BL (2005) Role of arginine in the stabilization of proteins against aggregation. Biochemistry 44: 4919–4925.BM BaynesDIC WangBL Trout2005Role of arginine in the stabilization of proteins against aggregation.Biochemistry4449194925
- 29. Minton AP (1998) Molecular crowding: Analysis of effects of high concentrations of inert cosolutes on biochemical equilibria and rates in terms of volume exclusion. Energetics Of Biological Macromolecules, Pt B. San Diego: Academic Press Inc. pp. 127–149.AP Minton1998Molecular crowding: Analysis of effects of high concentrations of inert cosolutes on biochemical equilibria and rates in terms of volume exclusion.Energetics Of Biological Macromolecules, Pt BSan DiegoAcademic Press Inc127149
- 30. Kozer N, Kuttner YY, Haran G, Schreiber G (2007) Protein-protein association in polymer solutions: From dilute to semidilute to concentrated. Biophys J 92: 2139–2149.N. KozerYY KuttnerG. HaranG. Schreiber2007Protein-protein association in polymer solutions: From dilute to semidilute to concentrated.Biophys J9221392149
- 31. Minton AP (1997) Influence of excluded volume upon macromolecular structure and associations in “crowded” media. Curr Opin Biotechnol 8: 65–69.AP Minton1997Influence of excluded volume upon macromolecular structure and associations in “crowded” media.Curr Opin Biotechnol86569
- 32. Phillips JC, Braun R, Wang W, Gumbart J, Tajkhorshid E, et al. (2005) Scalable molecular dynamics with NAMD. J Comput Chem 26: 1781–1802.JC PhillipsR. BraunW. WangJ. GumbartE. Tajkhorshid2005Scalable molecular dynamics with NAMD.J Comput Chem2617811802
- 33. Brooks BR, Bruccoleri RE, Olafson BD, States DJ, Swaminathan S, et al. (1983) Charmm - A Program For Macromolecular Energy, Minimization, And Dynamics Calculations. J Comput Chem 4: 187–217.BR BrooksRE BruccoleriBD OlafsonDJ StatesS. Swaminathan1983Charmm - A Program For Macromolecular Energy, Minimization, And Dynamics Calculations.J Comput Chem4187217
- 34. Jorgensen WL, Chandrasekhar J, Madura JD, Impey RW, Klein ML (1983) Comparison Of Simple Potential Functions For Simulating Liquid Water. J Chem Phys 79: 926–935.WL JorgensenJ. ChandrasekharJD MaduraRW ImpeyML Klein1983Comparison Of Simple Potential Functions For Simulating Liquid Water.J Chem Phys79926935
- 35. Cannon WR, Pettitt BM, McCammon JA (1994) Sulfate Anion In Water - Model Structural, Thermodynamic, And Dynamic Properties. J Phys Chem 98: 6225–6230.WR CannonBM PettittJA McCammon1994Sulfate Anion In Water - Model Structural, Thermodynamic, And Dynamic Properties.J Phys Chem9862256230
- 36. MacKerell AD (2001) Atomistic Models and Force Fields. In: Becker O, MacKerell AD, Roux B, Watanabe M, editors. Computation Biochemistry and Biophysics. New York: Marcel Dekker Inc. pp. 7–38.AD MacKerell2001Atomistic Models and Force Fields.O. BeckerAD MacKerellB. RouxM. WatanabeComputation Biochemistry and BiophysicsNew YorkMarcel Dekker Inc738
- 37. Schneider CP, Trout BL (2009) Investigation of Cosolute-Protein Preferential Interaction Coefficients: New Insight into the Mechanism by Which Arginine Inhibits Aggregation. J Phys Chem B 113: 2050–2058.CP SchneiderBL Trout2009Investigation of Cosolute-Protein Preferential Interaction Coefficients: New Insight into the Mechanism by Which Arginine Inhibits Aggregation.J Phys Chem B11320502058