236 In Vitro PROTEIN DEPOSITION [16] disulfide bonds), aggregation processes often compete with correct folding. Aggregation has attracted far less attention than protein folding. To date, little is known about the nature of these aggregation processes competing with correct folding or about the structure of the aggregates. Despite this lack of fundamental knowledge, powerful methods are available to inhibit aggregation side reactions in protein refolding. More detailed analyses of the refolding processes competing with correct folding will improve protein folding further. Acknowledgments This work was supported by the Federal German Ministry for Bildung Wissenschaft, Forschung, und Technologie (BMBF). We thank Susanne Richter for critically reading the manuscript and Hauke Lilie for the artwork. [ 16] Inhibition of Stress-Induced Aggregation of Protein Therapeutics By JOHN F. CARPENTER, BRENT S. KENDRICK, BYEONG S. CHANG, MARK C. MANNING, and THEODORE W. RANDOLPH Introduction There are numerous unique, critical applications for proteins in human healthcare. Dozens of protein therapeutics are already on the market or are in clinical trials. 1 However, even the most promising and effective protein therapeutic will not be beneficial to human health if its stability cannot be maintained during packaging, shipping, storage, and administra- tion. Instability can lead to protein aggregation, which is a major problem in the biopharmaceutics field. The aggregates are usually nonnative in structure and may remain soluble and/or precipitate from solution. Typi- cally, the aggregated molecules cannot be dissociated without the addition of a strong chaotropic agent (e.g., several molar guanidine hydrochloride). In some cases, the aggregates have nonnative intermolecular covalent link- ages, e.g., disulfide bonds. In addition to reducing efficacy, aggregates in parenterally administered proteins can cause adverse patient reactions such as immune response, 1 y. j. Wang and R. Pearlman, in "Pharmaceutical Biotechnology," Vol. 5. Plenum Press, New York, 1993. Copyright © 1999 by Academic Press All rights of reproduction in any form reserved. METHODS IN ENZYMOLOGY, VOL. 309 0076-6879/99 $30.00
`
`Ex. 2029-0001
`
`

`
`[161 INHIBITION OF PROTEIN AGGREGATION 237 sensitization, or even anaphylactic shock. 2'3 Therefore, if even a small per- centage of the protein molecules are aggregated, a product can be rendered unacceptable. Thus, what can be simply a nuisance--to be removed by centrifugation--to the protein chemist doing basic laboratory research can completely derail the product development process for a biotechnology company. Unless costly human clinical trials are conducted to prove that a product containing aggregates is both safe and effective, it is essential that aggregate formation is completely prevented during all stages of product processing, shipping, and storage. This goal is difficult to achieve because the free energy barrier between native and unfolded states is only on the order of 50 kJ/mol. 4 Furthermore, it appears that proteins can form nonnative aggregates from protein molecules that are only slightly perturbed from the native structure (e.g., molten globules). These protein molecules can even be a part of the ensemble of species that encompass the native conformation. Thus, even minor stresses (e.g., subdenaturing concentrations of chaotropic agents, small changes in pH, or temperatures well below the melting point of the protein) can foster protein aggregation. Aggregation is greatly favored if the protein is at relatively high concentrations (e.g., >10 mg/ml), which is often the case for protein therapeutic products. The situation becomes even more complex if one considers that, in addition to resistance to acute stresses encountered during product processing, therapeutic proteins must maintain stability during storage for up to 18-24 months. Even under conditions greatly favoring the native state (e.g., optimal pH, buffer, and ionic strength at 30°), there can be slow accumulation of aggregated molecules, which can be significant on the time scale for the shelf-life of pharmaceutical products. This article describes the types of stresses and conditions that are rou- tinely found to cause aggregation of purified therapeutic proteins. Stresses encountered during processing include short-term exposure to high temper- ature during pasteurization of aqueous solutions, freeze-thawing, freeze- drying, and exposure to denaturing interfaces due to agitation, filtration, air bubble entrainment during filling, and so on. It also considers the effects of long-term storage on protein stability in aqueous solutions and dried solids. Each section describes how the rational choice of stabilizing additives (Table I) can be used to inhibit protein aggregation. These choices are based on a clear understanding of the mechanisms by which different addi- tives succeed or fail as protein stabilizers under different conditions. The mechanisms are considered in detail and are illustrated by selected exam- 2 R. E. Ratner, T. M. Phillips, and M. Steiner, Diabetes 39, 728 (1990). 3 C. A. Thornton and M. Ballow, Arch. Neurol. 50, 135 (1993). 4 C. N. Pace, Trends Biochern. Sci. 15, 41 (1990).
`
`Ex. 2029-0002
`
`

`
`238 In Vitro PROTEIN DEPOSITION [ 16] TABLE I CLASSES OF PROTEIN STABILIZERS EFFECTIVE AT INHIBITING STRESS-INDUCED AGGREGATION a Type of stress condition Effective stabilizers Short-term thermal stress Sugars; polyols; salting-out salts; methylamines; polymers Same Same; surfactants Disaccharides; surfactants Disaccharides; carbohydrate polymers in combi- nation with disaccharides; suffactants Surfactants Long-term storage in aqueous solution Freezing/freeze-thawing stress Dehydration stress Long-term storage in dried solid Interracial stresses a See text for example compounds and their mechanisms of action and for definitions of stresses. pies, Finally, the article briefly describes a model that allows the calculation of degree of expansion of the native state needed to form an aggregate- fostering species in aqueous solution and the utility of infrared spectroscopy to characterize the structure of proteins in precipitates. Short-Term Thermal Stress in Aqueous Solution It is well recognized that aggregation can be triggered by the exposure of proteins to elevated temperatures in aqueous solution. Although this stress may seem simple to avoid, there are certain pharmaceutical processes that must employ heat treatments. Regulatory agencies require that any protein derived from blood or circulatory organs must undergo some type of viral inactivation. The most widely accepted method is exposure of aqueous solutions to 60 ° for 10 hr. However, many protein products exhibit some level of thermally induced aggregation during heat treatment. In general, thermally induced aggregation appears to follow a simple scheme. In a homogeneous protein preparation, perturbation of the native protein structure during heating can foster sufficient unfolding to promote aggregation. Alternatively, a population of protein molecules may contain a fraction that is especially sensitive to thermal stress. The fraction may be a misfolded form of the active protein, a chemically modified (e.g., proteolytically clipped) version of the active protein or a less stable protein impurity. With either type of sample, during heating the unfolded protein molecules then associate, often with native protein as well as themselves, forming soluble, nonnative aggregates. Finally, soluble aggregates accumu- late and eventually associate into insoluble protein precipitates.
`
`Ex. 2029-0003
`
`

`
`[I 6] INHIBITION OF PROTEIN AGGREGATION 239 Protein aggregation during thermal treatment can be minimized by increasing the free energy of protein unfolding (i.e,, increase the melting point of the native conformation), which is often accomplished by em- ploying stabilizing additives. The most potent stabilization can usually be obtained if a specific ligand, which binds strongly to the native state (e.g., an enzyme substrate or cofactor) and interacts minimally with the denatured state, can be identified. 5 These additives are highly effective at relatively low bulk concentration, in stoichiometric ratios with the protein. Also, nonspecific stabilizers can be used alone or in combination with specific stabilizers. 6-1° Low molecular, nonspecific compounds usually only confer useful stabilization at concentrations of several hundred millimolar or higher. These stabilizers include sugars (e.g., sucrose), salting-out salts (e.g., ammonium sulfate), polyols (e.g., glycerol), and certain methylamines (e.g., trimethylamine N-oxide). High molecular weight polymers such as polyeth- ylene glycol (PEG), gelatin, and hydroxyethyl starch are also effective stabilizers, often at concentrations above a few percent (w/v). However, low molecular polyethylene glycol (e.g., <3 kDa) or polyvinylpyrrolidone should be avoided because they can reduce the melting temperature of the native state and increase protein aggregation. 7 Finally, combinations of stabilizers from one or more of these classes can be used to enhance the inhibition of thermally induced protein unfolding and the resultant pro- tein aggregation. 8 Mechanism for Thermodynamic Stabilization by Additives The effects of both specific and nonspecific stabilizers can be explained by a single, straightforward thermodynamic mechanism. It is important to emphasize that the following arguments are applicable to many different reversible equilibria between protein states or species; even those that would be encompassed within the native state ensemble, but only have small differences in surface area (see later). For simplicity, we will consider a two-state model, in which there is an equilibrium between native and denatured states of the protein (N "~ D). At room temperature and in nonperturbing solvent environments, the na- 5 j. Brandt and L. O. Andersson, Int. J. Peptide Protein Res. 8, 33 (1976). 6 W. R. Porter, H. Staack, K. Brandt, and M. C. Manning, Thromb. Res. 71, 265 (1993). 7 M. Vrkljan, M. E. Powers, T. M. Foster, J. Henkin, W. R. Porter, J. F. Carpenter, and M. C. Manning, Pharm. Res. 11, 1004 (1994). 8 T. M. Foster, J. J. Dorrnish, U. Narahari, J. D. Meyer, M. Vrkljan, J. Henkin, W. R. Porter, J. F. Carpenter, and M. C. Manning, Int. J. Pharm. 134, 193 (1996). 9 L. Bjerring-Jensen, J. Dam, and B. Teisner, Vox Sang. 67, 125 (1994). 10 p. K. Ng and M. B. Dobkin, Thromb. Res. 39, 439 (1985).
`
`Ex. 2029-0004
`
`

`
`240 In Vitro PROTEIN DEPOSITION [ 16] tive state is favored because it has a lower free energy than the denatured state. The magnitude of the difference in free energy between the two states (i.e., the free energy of denaturation) dictates the relative stability of the native state. Any alteration in a system that decreases this difference will reduce stability, e.g., reduce the melting temperature of the protein. Conversely, increasing this free energy difference will increase the stability of the native state. The effects of additives on the relative stabilities of each state can be described by an application of the Wyman linkage function, which in this case can be defined as the link between ligand binding and stability of protein states binding the ligand. More detailed explanations can be found in publications by Timasheff, 11 Wyman, le and Wyman and Gill) 3 Binding of a ligand to a given state will reduce the free energy (chemical potential) of that state. The effect of ligand binding on protein stability depends on the difference in binding between the two states. If more ligand binds to the native state than to the denatured state, then the free energy of denaturation will be increased and the native state will be stabilized. This is the case for potent specific protein stabilizers (e.g., enzyme cofactors and other ligands that interact strongly with the native protein conformation). The opposite effect will be seen if more ligand binds to the denatured state. This general ligand-binding argument can also explain the mechanism for nonspecific protein stabilization and destabilization by additives. De- tailed reviews of this mechanism, which has been developed by Timasheff and colleagues, can be found elsewhere. H For the nonspecific additives, relatively high concentrations (ca. >0.3M) are needed to affect protein stability. This is because the interactions of the solute with the protein are relatively weak. Binding of these additives is actually a measure of the relative affinities of the protein for water and ligand. Therefore, the ligand interaction is referred to as preferential. Timasheff and colleagues have found that denaturants (e.g., urea and guanidine hydrochloride) are bound preferentially to proteins and that the degree of binding is greatest for the denatured state. 11 The free energy (chemical potential) of the denatured state is decreased more than that for the native state because more surface area for binding is exposed to solvent as the protein unfolds. Therefore, the free energy barrier between the two states is reduced, and the resistance to stress in the native state is reduced 11 S. N. Timasheff, Adv. Protein Chem. 51) 355 (1998). 12 j. Wyman, Adv. Protein Chem. 19, 223 (1964). 13 j. Wyman and S. J. Gill, in "Functional Chemistry of Biological Molecules." University Science Books, Mill Valley, CA, 1990.
`
`Ex. 2029-0005
`
`

`
`[ 161 INHIBITION OF PROTEIN AGGREGATION 241 (e.g., the melting point of the protein in lowered). If this effect is great enough, the protein will be denatured at room temperature. Conversely, Timasheff and colleagues have observed experimentally that there is a deficiency of stabilizing solutes (e.g., sugars and polyols) in the presence of the protein, relative to that seen in the bulk solution, u That is, the solutes are preferentially excluded from contact with the surface of the protein. In a thermodynamic sense the solute (ligand) has negative binding to the protein. Thus, there is an increase in the free energy (chemical potential) of the protein. The native state is stabilized because denaturation leads to a greater surface area of contact between the protein and the solvent and to greater preferential solute exclusion. Thus, even though there is an increase in the free energy of the native state, this effect is offset by the greater increase in the free energy of the denatured state. Long-Term Storage in Aqueous Solution During the time frame (i.e., several months) of storage that is needed for a therapeutic protein product, the formation of soluble aggregates and/ or protein precipitates is often a major problem. These degradation prod- ucts can arise under conditions greatly favoring the native state and in the absence of agitation (which is treated separately as a stress later). For example, the product may be prepared at the optimum pH and ionic strength, stored at temperatures well below the melting point of the protein, and still exhibit protein aggregation. Under these conditions, any spectroscopic measurement of the protein sample prior to storage indicates a "fully native" protein. In addition, in some cases, only minimal changes in secondary structure are detected in the aggregated species. Thus, a far more minor structural fluctuation than the unfolding of secondary structure is sufficient to cause problematic levels of aggregation during months of storage of aqueous therapeutic protein for- mulations. There are a number of reasons why the minor structural changes respon- sible for undesirable aggregations have been difficult to determine. The degree of structural change is probably too small for many analytical meth- ods to resolve. The change is also presumably a reversible process, unless the molecules involved are kinetically trapped by aggregation. The aggregate- competent species in the molecular population are most likely relatively scarce when compared to that of the most compact, native protein mole- cules. In addition, minor structural changes are constantly occurring so that what is defined as a "native" structure is actually an average of a continuous ensemble of molecules with fluctuating conformations.
`
`Ex. 2029-0006
`
`

`
`242 In Vitro PROTEIN DEPOSITION [ 16] An example of aggregation of a therapeutic protein during storage under minimally stressful conditions is provided by recombinant interleukin-1 receptor antagonist (IL-lra). TM Native IL-lra is monomeric and does not associate into higher order states. However, during storage at 30 ° , which is well below the melting point of 56 ° for the solution conditions used, the protein irreversibly forms a soluble dimer. The soluble dimeric IL-lra is biologically active and is almost structurally indistinguishable from native monomeric IL-lra, based on infrared and circular dichroism spectroscopies. These results are consistent with the hypothesis that the dimer forms from a species in the native state ensemble, which is only minimally different from the most compact species. Dimer formation is inhibited greatly by sucrose. TM This finding can be explained by the sucrose preferential exclusion mechanism presented ear- lier. In the presence of sucrose, protein species having an increase in surface area (or volume) are thermodynamically less favorable than the more com- pact species. Thus, in the presence of sucrose, the equilibrium is shifted away from the aggregate-competent species and, hence, with time there is less accumulation of aggregates. Furthermore, with rhlL-lra, the rates of both H-D exchange and reaction of side chains, which are chemically modified on exposure to the surface, are reduced in the presence of su- croseJ 5 Thus, with short-term measurements that are indicative of the conformational mobility of the native state, it can be seen that sucrose makes structural expansion less favorable. In equilibrium terms, the popula- tion is shifted toward the more compact species. In general, for proteins that form nonnative aggregates during long-term storage in aqueous solu- tion, the effect of thermodynamic stabilizers such as sucrose should be useful for increasing kinetic stability and minimizing aggregation. Pathway for Protein Aggregation and Calculation of Structural Expansion to Form Aggregation-Fostering Species Kinetic analyses of many protein aggregation pathways has resulted in the well-known Lumry-Eyring modelJ 6'17 The model, shown in Scheme i, involves a first-order reversible unfolding of the protein and subsequent aggregation of nonnative species in a higher order processJ 6,~7 In Scheme 14 B. S. Chang, R. M. Beauvais, T. Arakawa, L. N. Narhi, A. Dong, D. I. Aparisio, and J, F. Carpenter, Biophys. J. 71, 3399 (1996). 15 B. S. Kendrick, B. S. Chang, T. Arakawa, B. Peterson, T. W. Randolph, M. C. Manning, and J. F. Carpenter, Proc. Natl. Acad. Sci. U.S.A. 94, 11917 (1997). 16 Ko W. Minton, P. Karmin, G. M. Hahn, and A. P. Minton, Proc. Natl. Acad. Sci. U.S.A. 79, 7107 (1982). 17 R. Lurnry and H. Eyring, J. Phys. Chem. 58, 110 (1954).
`
`Ex. 2029-0007
`
`

`
`[16] INHIBITION OF PROTEIN AGGREGATION 243 k] (a) N ~ ~ A (b) A m + A km Scheme 1 SCHEME 1 Am+ I (Aggregate) 1, N refers to native protein and A to an intermediate conformational state preceding aggregation. Am refers to an aggregated form composed of m protein molecules. The rate constants for each reaction, i, are represented by ki. If the first step is in equilibrium, the model predicts that aggregation should follow second- or higher-order kinetics. However, with some proteins aggregation has been found to follow first-order kinetics. 18 This is an unexpected result for a bimolecular reaction and shows that the process is not rate limited by protein-protein collisions, but rather by the preceding unimolecular step shown in Scheme 2. In Scheme 2, N* is a transiently expanded conformational species within the native state ensemble, which is in equilibrium with N. Keq is the equilibrium constant for the reaction N to N*. N* is transformed irreversibly to an aggregation-competent state, A. State A undergoes further reaction to form insoluble aggregates, Am, composed of rn monomer units. The irreversible, unimolecular isomerization reaction of N* to A is the rate-limiting step in the formation of aggregates and has a rate constant denoted as kc. For unimolecular isomerizations, kc is not expected to depend on solvent vis- cosity. For proteins that aggregate via this pathway, a decrease in the aggrega- tion rate by sucrose, which is excluded preferentially from the surface of proteins, can be explained readily by the Timasheff mechanism outlined earlier.11.18 The degree of preferential exclusion and the increase in chemical potential are directly proportional to the surface area of protein exposed to solvent. By the LeChatelier principle, the system will minimize the thermodynamically unfavorable effect of preferential sucrose exclusion by favoring the state with the smallest surface area. This corresponding shift in the equilibrium (Keq in Scheme 2) toward compact species (N in Scheme 2), which can be explained by the Wyman linkage relationship, will lead to an overall decrease in the rate of protein aggregation. 18 B. S. Kendrick, J. L. Cleland, J. F. Carpenter, and T. W. Randolph, Proc. Natl. Acad. Sci. U.S.A. 95, 14142 (1998).
`
`Ex. 2029-0008
`
`

`
`244 In Vitro PROTEIN DEPOSITION [ 16] (a) N ~ Keq~ N* kc ~ i Iv (b) Am + A ~ Am+ ~ (Aggregate) km Scheme 2 SCHEME 2 Based on the aggregation pathway presented in Scheme 2 and insight into protein-sucrose thermodynamic interactions, a model has been devel- oped to calculate the expansion in protein surface area needed to form the intermediate state preceding aggregation. TM A detailed development of this model and an example using first-order aggregation of interferon-y can be found in Kendrick et al. 18 Application of the model requires determination of initial rates of loss of native protein as a function of sucrose concentration and use of thermodynamic parameters available from the literature (e.g., surface tension increment of sucrose in water). With interferon-y, it was found that only a 9% expansion in surface area of the most compact species of the native state was needed to foster protein aggregation. Such a minor change in conformation is consistent with the finding that proteins can aggregate under conditions that favor the native state greatly. Finally, these results indicate that for the many protein systems that are prone to aggregation both in vitro and in vivo, only modest or even undetectable applied stress may be sufficient to drive aggregation. Similarly, mutations that have apparently minimal effect on properties, such as free energy of unfolding, may cause aggregation because they promote a slight increase in the fraction of the native state population that is structurally expanded at any instant in time. Methods that probe for these minor changes (e.g., measurement of H-D exchange kinetics) may be needed to discern mechanistically why some systems aggregate, whereas others that are very similar do not. Freezing and Freeze-Thawing Stresses There are several points in the processing, shipping, and storage of proteins at which a solution may be frozen, either by design or accident. For example, large volumes of therapeutic protein solutions are stored frozen as an intermediate holding step. Freezing stress can also contribute to protein damage during the freeze-drying process. Liquid aqueous formu- lations that may be adequately stable during storage under controlled condi-
`
`Ex. 2029-0009
`
`

`
`[16] INHIBITION OF PROTEIN AGGREGATION 245 tions (e.g., at 4 °) can be damaged severely if frozen accidently during shipping. Often the consequence of freeze-thawing is the formation of protein aggregates. During freezing a protein is exposed to low temperatures and the forma- tion of ice. Cold denaturation has been documented for many proteins and by itself may be sufficient to account for at least some of the damage noted during freezing. 19 Also, the protein, which partitions into the non-ice liquid phase, is exposed to extremely high solute concentrations as the sample is frozen. If solutes that are destabilizing to the protein are present or if the protein is perturbed by high ionic strength, then this concentrating effect can contribute to protein denaturation. Also, there can be dramatic pH changes during freezing. For example, the dibasic form of sodium phosphate crystallizes in frozen solution, which results in a system that contains essen- tially solely the monobasic salt and has a very low pH, e.g., pH 4. 20 To minimize problems associated with pH changes, sodium phosphate should be avoided whenever possible. Other buffers such as Tris, which do not have large freezing-induced changes in pH, should be used. 2° Fortunately, to prevent freezing-induced damage to proteins, often it is not necessary to discern which stresses are responsible for the damage. The first step is to choose specific conditions (e.g., pH or ligand) that maximize the thermodynamic stability of the protein. 21 Next, nonspecific stabilizers can be added, if needed. A wide variety of compounds have been found to provide nonspecific protection to proteins by the preferential exclusion mechanism described earlier. 2a These include sugars, amino acids, polyols, certain salts, methylamines, alcohols, other proteins, and synthetic polymers. 21 Compared to sugars, polymers such as PEG, polyvinylpyrroli- done, and other proteins (e.g., albumins) are much more potent cryoprotec- tants and can be effective at concentrations <1.0% (w/v). 21 However, for most proteins, sufficient freezing protection can be obtained by using a disaccharide (e.g., sucrose), which has the added benefit of also protecting the protein during subsequent drying (see later). Relatively high sugar concentrations [e.g., >30% (w/v)] may be needed to confer adequate protec- tion during freezing. Often, increasing the initial protein concentration increases the percent- age recovery of native protein during freeze-thawing. 2°'2~ At least in some 19 p. L. Privalov, Crit. Rev. Biochem. Mol. Biol. 25, 281 (1990). 2o T. J. Anchordoquy and J. F. Carpenter, Arch. Biochem. Biophys. 332, 231 (1996). 21 j. F. Carpenter and B. S. Chang, in "Biotechnology and Biopharmaceutical Manufacturing, Processing and Preservation" (K. E. Avis and V. L. Wu, eds.), p. 199. Interpharm Press, Buffalo Grove, IL, 1996.
`
`Ex. 2029-0010
`
`

`
`246 In Vitro PROTEIN DEPOSITION [16] cases, in a given sample volume, there is the same mass of protein aggre- gated, independent of the initial protein concentration. 22 These results suggest that protein denaturation and aggregation may occur at the ice- water interface, which has a finite area under a given set of freezing condi- tions (e.g., sample volume and cooling rate). Other observations support this suggestion. Increasing the rate of cooling, which fosters the formation of smaller ice crystals and greater overall ice surface area, often leads to increased aggregation during freeze-thawing. 23 Strambini and Gabellieri 24 have directly documented with phosphorescence lifetime studies that per- turbation of protein tertiary structure in the frozen state was almost twofold greater for protein frozen by cooling at 100°/min than that for samples cooled at l°/min. Consistent with the suggestion that the ice-water interface can denature proteins is the finding that many different nonionic surfactants, at very low concentration [ca. <1% (w/v)], can inhibit protein aggregation greatly during freeze-thawing. 23 This protection has been attributed to the competi- tion of the surfactant with the protein for the ice-water interface. 22'23 This mechanism and other aspects of protection of proteins against interfacial stress will be considered further below. Freeze-Drying Stress To obtain the requisite storage stability for a pharmaceutical protein, it is often necessary to prepare a freeze-dried formulation, zl The problem is that the stresses encountered during this process can cause protein dena- turation and aggregation, the latter of which is usually noted after the protein sample has been rehydrated. 2a'z5 The effects of freezing, and how protein damage can be minimized during this stress, have already been addressed. This section focuses primarily on the dehydration stress. How- ever, it is important to realize that to recover a native, functional protein after freeze-drying and rehydration, it is essential that the protein be pro- tected during both freezing and drying streps. 21 The functional, three-dimensional structure of a protein is determined to a large degree by the interaction of the protein residues and backbone with water. 26 Thus, it is not surprising that, in the absence of stabilizing additives, protein unfolding is often induced by dehydration, which can 22 L. Kreilgard, J. L. Flink, S. Frojaer, T. W, Randolph, and J. F. Carpenter, I. Pharm. Sci. 88, 281 (1999). 23 B. S. Chang, B. S. Kendrick, and J. F. Carpenter, Z Pharm. Sci. 85, 1325 (1996). 24 G. B. Strambini and E. Gabellieri, Biophys. Z 70, 971 (1996). 25 S. J. Prestrelski, N. Tedeschi, T. Arakawa, and J. F. Carpenter, Biophys. J. 65, 661 (1993). z6 I. D. Kuntz and W. Kauzman, Adv. Protein Chem. 28, 239 (1974).
`
`Ex. 2029-0011
`
`

`
`[161 INHIBITION OF PROTEIN AGGREGATION 247 be brought about by freeze-drying or other drying methods (e.g., spray- drying) .21,25.27 Infrared spectroscopy has proven invaluable in studying these structural alterations because protein samples in both liquid and dried states can be examined. For an unprotected protein in the dried solid, the infrared spectrum in the conformationally sensitive amide I region is often altered greatly relative to that for the native protein in aqueous solution. 21'25'27 Broadening and shifting of the component bands indicate that unfolded protein molecules are present in the dried solid. 21'25'27 Unprotected lyophilized proteins often form nonnative aggregates on rehydration. 21,2s'27 The degree of aggregation correlates with the degree to which the secondary structure of the protein has been perturbed during freeze-drying. 21'25'27 In the dried solid there are unfolded protein molecules at essentially infinite concentration. When water is added, intermolecular contacts are favored over refolding and a substantial fraction of the molecu- lar population can form aggregates. Reducing protein unfolding during freeze-drying also serves to reduce the degree of aggregation noted after rehydration. Sometimes, simply alter- ing a specific solution condition, such as initial pH, can have dramatic effects. For example, with infrared spectroscopy, Prestrelski and colleagues 28 found that interleukin-2, lyophilized from a solution of pH 7.0, was unfolded. On rehydration, a large fraction of the sample was aggregated. Furthermore, the appearance of new infrared bands at 1617 and 1690 cm -1 suggested that the protein sample was at least partially aggregated in the dried solid. 27 When pH 5.0 was employed, the spectrum of the dried protein was almost identical to that for the native aqueous protein, and aggregation after rehydration was minimal. The resistance of many proteins to stresses im- posed in solution is known to show a strong pH dependence. For lyophiliza- tion, it is not known whether the influence of pH manifests itself during freezing, drying, or rehydration, where charge-charge repulsion may de- crease intermolecular interactions. Optimizing specific solution conditions for a given protein often is not sufficient to inhibit lyophilization-induced unfolding and the resulting ag- gregation. 21'25,27 A nonspecific stabilizer, such as sucrose or trehalose, is then needed. To date, these two disaccharides have been found to be the most effective stabilizers during lyophilization. 21'25'27 Importantly, they provide protection during both freezing and dehydration, such that the native secondary structure can be recovered in the dried solid. As a conse- quence, aggregation is reduced in the rehydrated sample. 27 j. F. Carpenter, S. J. Prestrelski, and A. Dong, Eur. J. Pharm. Biopharm. 45, 231 (1998). 28 S. J. Prestrelski, K. A. Pikal, and T. Arakawa, Pharm. Res. 12, 1250 (1995).
`
`Ex. 2029-0012
`
`

`
`248 In Vitro PROTEIN DEPOSITION [ 161 Sometimes, a nonionic surfactant, such as polyoxyethylene sorbitan monolaurate (Tween 20), can be used, in addition to the sugar, to inhibit aggregation further. The surfactant may contribute to the inhibition of protein unfolding during freezing and drying, z3 In addition, the surfactant may serve a "ch