Physical stability of proteins in aqueous solution: Mechanism and ...
`Pharmaceutical Research; Sep 2003; 20, 9; ProQuest Centralpg. 1325
`
`Pharmaceutical Research, Vol. 20, No. 9, September 2003 (© 2003)
`
`Review
`
`Physical Stability of Proteins in Aqueous Solution:
`Mechanism and Driving Forces in Nonnative
`Protein Aggregation
`
`Eva Y. Chi/ Sampathkumar Krishnan/ Theodore W. Randolph/ and John F. Carpenter3
`
`4
`•
`
`Received February 26, 2003; accepted May 1, 2003
`
`Irreversible protein aggregation is problematic in the biotechnology industry, where aggregation is
`encountered throughout the lifetime of a therapeutic protein, including during refolding, purification,
`sterilization, shipping, and storage processes. The purpose of the current review is to provide a funda(cid:173)
`mental understanding of the mechanisms by which proteins aggregate and by which varying solution
`conditions, such as temperature, pH, salt type, salt concentration, cosolutes, preservatives, and surfac(cid:173)
`tants, affect this process.
`
`KEY WORDS: formulation; pharmaceuticals; denaturation; second virial coefficient; conformational
`stability.
`
`INTRODUCTION
`
`The issue of protein stability was first explained on a
`fundamental level by Hsien Wu in 1931 (1), when he pro(cid:173)
`posed a theory on protein denaturation after publishing 12
`papers on his experimental observations on this topic (2). In
`1954, Lumry and Eyring published a seminal paper (3) -
`"Conformation Changes of Proteins" -that laid the ground(cid:173)
`work for what we know today about protein structure, fold(cid:173)
`ing, stability, and aggregation.
`Protein stability is a particularly relevant issue today in
`the pharmaceutical field and will continue to gain more im(cid:173)
`portance as the number of therapeutic protein products in
`development increases. Proteins provide numerous unique
`and critical treatments for human diseases and conditions
`(e.g., diabetes, cancer, hemophilia, myocardial infarction).
`There are already dozens of protein products on the market
`and hundreds more in preclinical and clinical development
`( 4). However, if a therapeutic protein cannot be stabilized
`adequately, its benefits to human health will not be realized.
`The shelf life required for economic viability of a typical pro(cid:173)
`tein pharmaceutical product is 18-24 months (5). Achieving
`
`1 Department of Chemical Engineering, Center for Pharmaceutical
`Biotechnology, ECCH 111, Campus Box 424, University of Colo(cid:173)
`rado, Boulder, Colorado.
`2 Department of Pharmaceutics and Drug Delivery, Amgen Inc.,
`Thousand Oaks, California.
`3 Department of Pharmaceutical Sciences, School of Pharmacy, Uni(cid:173)
`versity of Colorado Health Sciences Center, Denver, Colorado.
`4 To whom correspondence should be addressed. (e-mail:
`john.carpenter@uchsc.edu)
`ABBREVIATIONS: rhiFN--y, recombinant human interferon--y;
`rhGCSF, recombinant human granulocyte colony-stimulating factor;
`llGunf• free energy of unfolding; B 22, osmotic second virial coefficient;
`GdnHCl, guanidine hydrochloride.
`
`this goal is particularly difficult because proteins are only
`marginally stable and are highly susceptible to degradation,
`both chemical and physical (6-9). Chemical degradation re(cid:173)
`fers to modifications involving covalent bonds, such as deami(cid:173)
`dation, oxidation, and disulfide bond shuffling. Physical deg(cid:173)
`radation includes protein unfolding, undesirable adsorption
`to surfaces, and aggregation (6,8-10). Nonnative aggregation
`is particularly problematic because it is encountered routinely
`during refolding, purification, sterilization, shipping, and stor(cid:173)
`age processes. Aggregation can occur even under solution
`conditions where the protein native state is highly thermody(cid:173)
`namically favored (e.g., neutral pH and 37°C) and in the ab(cid:173)
`sence of stresses. This review examines the mechanisms and
`driving forces in nonnative protein aggregation.
`Nonnative protein aggregation (hereafter referred to
`simply as "aggregation") describes the assembly from initially
`native, folded proteins of aggregates containing nonnative
`protein structures. Aggregation is often irreversible, and ag(cid:173)
`gregates often contain high levels of nonnative, intermolecu(cid:173)
`lar [3-sheet structures (11). Protein aggregation behaviors,
`such as onset, aggregation rate, and the final morphology of
`the aggregated state (i.e., amorphous precipitates or fibrils)
`have been found to depend strongly on the properties of a
`protein's solution environment, such as temperature, pH, salt
`type, salt concentration, cosolutes, preservatives, and surfac(cid:173)
`tants (10,12-16) as well as the relative intrinsic thermody(cid:173)
`namic stability of the native state (17-20).
`This review first examines how different solution condi(cid:173)
`tions affect protein stability. Case studies and fundamental
`insights into how each solution condition affects protein sta(cid:173)
`bility are discussed. The second part of this review discusses
`characteristics, mechanisms, energetics, and driving forces of
`nonnative protein aggregation. Recent studies on two thera(cid:173)
`peutic proteins -recombinant human interferon-)' (rhiFN(cid:173)
`'Y) and recombinant human granulocyte colony-stimulating
`
`1325
`
`0724·8741/03/0900·1325/0 © 2003 Plenum Publishing Corporation
`
`Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
`
`Ex. 2018-0001
`
`Chi, Eva Y;Sampathkumar Krishnan;Randolph, Theodore W;Carpenter, John F
`

`
`1326
`
`Chi et al.
`
`factor (rhGCSF) - will be used to provide insight into the
`mechanistic issues. The present review is not aimed at pro(cid:173)
`viding a comprehensive literature review in the area of pro(cid:173)
`tein stability and aggregation. Each protein is unique both
`chemically and physically and therefore will exhibit unique
`stability behavior. The purpose of the current review is to
`provide a fundamental understanding of the mechanisms by
`which proteins aggregate and by which varying solution con(cid:173)
`ditions may affect this process. This insight in turn can be used
`in the rational design of stable aqueous formulations of thera(cid:173)
`peutic proteins.
`
`FACTORS INFLUENCING PROTEIN STABILITY
`
`Temperature
`
`Most proteins fold to a specific globular conformation
`that is essential for their biologic functions. The thermody(cid:173)
`namic stability of the native protein conformation is only mar(cid:173)
`ginal, about 5-20 kcal/mole in free energy more stable than
`unfolded, biologically inactive conformations under physi(cid:173)
`ologic conditions (21-25). This thermodynamic stability is
`much weaker than covalent or ionic bonds ( -150 kcal/mole)
`(26) or the thermal energy of a protein (5-20 kcal/mole is less
`than one tenth of k 8 T per residue, where k 8 is Boltzmann's
`constant and Tis the absolute temperature) (21). The small
`net conformational stability of protein results from a unique
`balance between large stabilizing and large destabilizing
`forces. Contributions to the free energy of folding arise from
`hydrophobic interactions, hydrogen bonding, van der Waals'
`forces, electrostatic forces (classic charge repulsion or ion
`pairing), and intrinsic propensities (local peptide interactions)
`(21). The main force opposing protein folding is the protein's
`conformational entropy. Both local entropy (e.g., transla(cid:173)
`tional, rotational, and vibrational degrees of freedom on the
`molecular scale) and nonlocal entropy (e.g., excluded volume
`and chain configurational freedom) are increased on unfold(cid:173)
`ing (21). Because of the small conformational stability of the
`protein native state, relatively small changes of external vari(cid:173)
`ables (e.g., temperature, pH, salt, etc.) in the protein-solvent
`system can destabilize the structure of the protein, i.e., induce
`its unfolding.
`The thermodynamic stability of the native protein con(cid:173)
`formation, characterized by the free energy of unfolding
`(~Gunf), typically shows a parabolic profile as a function of
`temperature (27-30). ~Gunf therefore becomes negative at
`two temperatures, accounting for the unfolding of proteins at
`both high (e.g., 5G-100°C) and low temperatures (e.g., less
`than 10°C) (21,27,29,30). The molecular origin of the effect of
`temperature on ~Gunf is complex and is the subject of much
`ongoing research (21,22,29-31) and is not considered further
`here.
`It has long been known that incubating protein solutions
`at high temperatures results in physical degradation. Al(cid:173)
`though thermally induced denaturation may be reversible for
`some proteins, high temperatures usually lead to irreversible
`denaturation because of aggregation. Examples include the
`concomitant unfolding and aggregation of recombinant hu(cid:173)
`man Flt2 ligand (32), streptokinase (33), recombinant human
`keratinocyte growth factor (34,35), recombinant consensus in(cid:173)
`terferon (36), rhiFN-)' (37), and ribonuclease A (38,39).
`
`Typically, high temperatures perturb the native protein
`conformation to a sufficient degree to promote aggregation
`(13). Importantly, it is usually observed during heating that
`aggregation starts at temperatures well below the equilibrium
`melting temperature of the protein (11). This observation
`suggests that aggregates are not formed from fully unfolded
`molecules. Rather, as discussed in more detail below, it ap(cid:173)
`pears that partially unfolded protein molecules are the reac(cid:173)
`tive species that form aggregates.
`Temperature also strongly affects reaction kinetics be(cid:173)
`cause rate constants increase exponentially with temperature
`for activated reactions. Increasing temperature increases the
`thermal kinetic energy of reactants. As a result, reactant col(cid:173)
`lision frequency, as well as the probability of collisions with
`enough energy to overcome activation energies, increases
`with increasing temperature ( 40). For diffusion-controlled re(cid:173)
`actions, increasing temperature increases the rate of diffusion
`of reactant species, thus increasing the rate of reaction. Pro(cid:173)
`tein aggregation rates are similarly increased at high tempera(cid:173)
`tures.
`
`Solution pH
`
`pH has a strong influence on aggregation rate. Proteins
`are often stable against aggregation over narrow pH ranges
`and may aggregate rapidly in solutions with pH outside these
`ranges. Examples include recombinant factor VIII SQ (41),
`low-molecular-weight urokinase (42), relaxin (43), rhGCSF
`(17,20), deoxyhemoglobin (44), interlukin-1[3 (45), ribonucle(cid:173)
`ase A (46), and insulin (47).
`Solution pH determines the type (positive or negative)
`and total charge on the protein, thereby affecting electrostatic
`interactions. There are two different ways in which electro(cid:173)
`static interactions can affect protein stability. First, classic
`electrostatic effects are the nonspecific repulsions that arise
`from charged groups on a protein when it is highly charged,
`for example, at pH far removed from the isoelectric point (pi)
`of the protein (21). As the number of charged groups on a
`protein is increased by increasing the acidity or basicity of the
`solution, increased charge repulsion within the protein desta(cid:173)
`bilizes the folded protein conformation because the charge
`density on the folded protein is greater than on the unfolded
`protein. Thus, pH-induced unfolding leads to a state of lower
`electrostatic free energy (21). Second, specific charge inter(cid:173)
`actions, such as salt bridges (or ion pairing), can also affect
`protein conformational stability. In contrast to the nonspe(cid:173)
`cific electrostatic effect, where increasing charges destabilize
`the folded state, salt bridges stabilize it (21,48,49).
`In addition to their effects on protein conformation,
`charges on protein molecules also give rise to electrostatic
`interactions between protein molecules. When proteins are
`highly charged, repulsive interactions between proteins stabi(cid:173)
`lize protein solution colloidally, making assembly processes
`such as aggregation energetically unfavorable (20,26,37,38).
`When proteins possess both positively and negatively charged
`groups (e.g., at pH values close to the pi), anisotropic charge
`distribution on the protein surface could give rise to dipoles.
`In such cases, protein-protein interactions could be highly
`attractive, making assembly processes such as aggregation en(cid:173)
`ergetically favorable (20,50).
`
`Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
`
`Ex. 2018-0002
`
`

`
`Stability of Proteins in Aqueous Solution
`
`1327
`
`Ligands and Cosolutes
`
`. The :Vyman linkage function and related theories ap(cid:173)
`phed .by T1masheff et al. [see, for example, Timasheff (51)] to
`protem conformational stability can be used to explain the
`effects of cosolutes, such as strong binding ligands, excipients,
`and salts, on protein physical stability. By the Wyman linkage
`function, differential binding of ligand in a two-state equilib(cid:173)
`rium will shift the equilibrium toward the state with the
`greater binding. Thus, for example, binding of polyanions to
`the native state of acidic fibroblast growth factor (52) or na(cid:173)
`ti~e recombi~~n~ keratinocyte growth factor (34) greatly
`sh1fts the eqmhbnum between the native and unfolded states
`to favor the native state. Likewise, binding of Zn2 + to human
`growth hormone increases the free energy of unfolding (53).
`The Wyman linkage function can also be used to explain
`the effect of weakly interacting ligands (i.e., cosolutes) that
`affect protein conformational stability and equilibrium solu(cid:173)
`bility at relatively high concentrations. It has been recognized
`for over a century that high concentrations (2:1 M) of certain
`solutes (e.g., sugars, polyols, and certain salts, such as ammo(cid:173)
`nium sulfate) stabilize the native state of proteins, whereas
`other solutes act as protein denaturants [e.g., urea and gua(cid:173)
`nidine hydrochloride (GdnHCl)] (51,54,55). These observa(cid:173)
`tions can be explained by differences in binding of these
`weakly interacting solutes to native and unfolded states. De(cid:173)
`naturants bind to the unfolded state to a higher degree than to
`the native state, thus favoring unfolding.
`~rotein stabilizers such as sucrose and glycerol are pref(cid:173)
`erentially excluded from the surface of a protein molecule,
`and the degree of exclusion is proportional to its solvent(cid:173)
`~xposed sur~ace area (51,56,57). These cosolutes are depleted
`m the domam of the protein, and as a result, water is enriched
`in that domain. Preferential exclusion can thus be interpreted
`as negative binding. During unfolding, protein surface area
`increases, leading to a greater degree of preferential exclusion
`(~.g., ~ar~er negative binding). The net effect of greater nega(cid:173)
`tive bmdmg to the unfolded state is to favor the native state.
`. Another way to state the mechanism by which preferen(cid:173)
`tial exclusion stabilizes the native state is to consider that this
`inte~action coincides with an increase in protein chemical po(cid:173)
`tential. By LeChatelier's principle, the system will tend to
`minimize this unfavorable effect. Thus, protein states with
`reduced surface area that exhibit lower preferential exclusion
`are favored over more solvent-exposed states. As a result, the
`free ~nergy of unfolding is increased in the presence of pref(cid:173)
`erentially excluded solutes. In addition, assembly of mono(cid:173)
`mers into native oligomers, which reduces the specific protein
`surface area exposed to solvent, is also favored. The same
`mechanism also explains the decrease in equilibrium solubil(cid:173)
`ity of proteins in the presence of preferentially excluded sol(cid:173)
`utes, such as ammonium sulfate (58).
`Ligands and cosolutes that alter protein conformational
`stability also influence the rate of formation of nonnative
`aggregates. For example, in the presence of polyanions, ag(cid:173)
`gregation of acidic fibroblast growth factor (52) and native
`n~c?mbinant keratinocyte growth factor (34) is greatly in(cid:173)
`hibited. It has also been shown that the addition of weakly
`interacting preferentially excluded solutes can reduce the
`rate of protein aggregation. For example, sucrose has been
`shown to inhibit aggregation of hemoglobin (59), rhiFN-)'
`(60,61), keratinocyte growth factor (62), immunoglobulin
`
`light chains (15,19), and rhGCSF (17). In contrast, cosolutes
`(e.g., GdnHCl) that exhibit greater binding to the denatured
`state can accelerate aggregation (15,18,19,63). The mecha(cid:173)
`nism for the effects of cosolutes on protein aggregation is
`discussed below.
`
`Salt Type and Concentration
`
`Electrolytes have complex effects on protein physical sta(cid:173)
`bility by modifying conformational stability, equilibrium solu(cid:173)
`bility (e.g., salting-in and salting-out), and rate of formation of
`nonnative aggregates (38,64-67). For example, Yamasaki et
`al. found that bovine serum albumin could be stabilized
`against thermal unfolding by kosmotropic salts such as
`NaSCN and NaC10 4 and destabilized by chaotropic salts at
`high ionic strength (68). However, low concentrations of
`chaotropes (10-100 mM) stabilized bovine serum albumin
`(68). The equilibrium solubility of recombinant human tissue
`factor pathway inhibitor was decreased in the presence of
`NaCl (69). The rates of aggregation of recombinant factor
`VIII SQ (41) and recombinant keratinocyte growth factor
`(70) were decreased in the presence of NaCl. In contrast,
`NaCl increased the aggregation rate for rhGCSF (20).
`Salts bind to proteins. Ions can interact with unpaired
`charged side chains on the protein surface. Binding of multi(cid:173)
`valent ions to these side chains can cross-link charged resi(cid:173)
`dues on the protein surface, leading to the stabilization of the
`protein native state (65). Because the peptide bond has a
`large dipol.e moment resulting from a partial positive charge
`on the ammo group and partial negative charge on the car(cid:173)
`bonyl oxygen, ions can bind to peptide bonds (67), potentially
`destabilizing the native state. Consistent with the Wyman
`linkage theory described above, destabilization occurs if ions
`bind more strongly to nonnative than to native protein states
`(65).
`Electrolytes modulate the strength of electrostatic inter(cid:173)
`actions between the charged groups, both within the protein
`and between protein molecules. Thus, whereas intramolecu(cid:173)
`~ar charge-charge interactions affect conformational stability,
`mtermolecular electrostatic interactions affect equilibrium
`and rate of aggregate formation, as is described in more detail
`below.
`At low concentrations, the predominant effect of ions in
`solution results from charge shielding, which reduces electro(cid:173)
`static interactions. However, at high concentrations of certain
`salts, in addition to charge-shielding effects, preferential bind(cid:173)
`ing of ions to the protein surface can result in a decrease in
`thermodynamic stability of the native conformation and an
`increase in equilibrium solubility (64). Other salts that are
`preferentially excluded from protein surface show stabilizing
`or salting-out effects (71).
`The net effect of salt on protein stability is thus a balance
`of. the multiple mechanisms by which salt interacts with pro(cid:173)
`tem molecules and by which salt affects protein-protein in(cid:173)
`teractions. Because pH determines the type, total, and distri(cid:173)
`bution of charges in a protein, salt-binding effects may be
`strongly pH dependent.
`
`Preservatives
`
`Antimicrobial preservatives, such as benzyl alcohol and
`phenol, are often needed in protein liquid formulations to
`ensure sterility during its shelf life. In particular, multidose
`
`Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
`
`Ex. 2018-0003
`
`

`
`1328
`
`Chi et al.
`
`formulations of proteins require effective preservatives to
`prevent microbial growth after the first dose has been re(cid:173)
`moved from a product vial. Preservatives are also required for
`certain drug delivery systems, e.g., injection pens that are
`used for multiple doses, minipumps that are used for continu(cid:173)
`ous injection, and topical applications for wound healing.
`However, preservatives often induce aggregation of protein
`in aqueous solution. For example, preservatives (e.g., phenol,
`m-cresol, and benzyl alcohol) have been shown to induce
`aggregation of human growth hormone (16), recombinant in(cid:173)
`terleukin-1 receptor (72), human insulin-like growth factor I
`(73), and rhiFN-)' (74).
`The mechanism for preservative-induced protein aggre(cid:173)
`gation is not well understood. However, it has been observed
`that addition of benzyl alcohol perturbed the tertiary struc(cid:173)
`ture of rhiNF-)' without affecting its secondary structure, and
`the rate of rhiNF-)' aggregation increased as the molar ratio
`of benzyl alcohol to protein increased (74). Also, preserva(cid:173)
`tives reduced the apparent melting temperature of recombi(cid:173)
`nant interleukin-1 receptor (72). These results suggest that
`preservatives bind to and populate unfolded protein states
`that are prone to aggregation. However, further research is
`needed to test this hypothesis and to determine rational strat(cid:173)
`egies to inhibit preservative-induced protein aggregation.
`
`Surfactants
`
`Nonionic surfactants are often added to protein solutions
`to prevent aggregation and unwanted adsorption (e.g., to fil(cid:173)
`ter and container surfaces) during purification, filtration,
`transportation, freeze-drying, spray-drying, and storage. Sur(cid:173)
`factants are amphiphilic molecules that tend to orient so that
`the exposure of the hydrophobic portion to the aqueous so(cid:173)
`lution is minimized. For example, surfactants adsorb at air/
`water interfaces, forming a surface layer of surfactant mol(cid:173)
`ecules oriented so that only their hydrophilic ends are ex(cid:173)
`posed to water. Such orientation and surface adsorption can
`also occur at solid/water interfaces such as those found in
`vials, syringes, tubing, and other containers [for a review see
`Randolph et al. (10) and references therein]. Protein mol(cid:173)
`ecules are also surface active and adsorb at interfaces. Surface
`tension forces at interfaces perturb protein structure, often
`resulting in aggregation. Surfactants inhibit interface-induced
`aggregation by limiting the extent of protein adsorption
`(10,75).
`As with other cosolutes, differential binding of surfac(cid:173)
`tants to native and unfolded states of protein influences the
`protein's conformational stability. For some proteins, surfac(cid:173)
`tants bind more strongly to the native state and increase the
`free energy of denaturation [e.g., human growth hormone
`(76)]. A more common effect is preferential binding of sur(cid:173)
`factants to the unfolded state, resulting in a decrease in the
`native protein state stability (10). Despite that surfactants
`often cause a reduction in thermodynamic stability of protein
`conformation, surfactants still can kinetically inhibit proteins
`aggregation at interfaces. In addition, surfactants have been
`shown to act as chemical chaperones, increasing rates of pro(cid:173)
`tein refolding and thus reducing aggregation (77). The read(cid:173)
`ers are directed to the following reviews, and references
`therein, for further information: Randolph et al. (10), Jones et
`al. (75), and Jones (78).
`
`MECHANISM OF PROTEIN AGGREGATION
`
`Structural Transitions Accompanying Aggregation
`
`Protein aggregation is accompanied by the loss of native
`protein structure. Such structural transitions have been well
`documented by Fourier transform infrared spectroscopy
`(FTIR) studies [for a review, see Dong et al. (11) and refer(cid:173)
`ences therein] A common feature of protein aggregates -
`formed in response to thermal, chemical, or physical stresses,
`or even in the absence of any applied stress -
`is an increased
`level of nonnative intermolecular [3-sheet structures (11 ). This
`structural transition occurs regardless of the initial secondary
`structural composition of the native protein (11) or the final
`morphology (amorphous or fibrillar) of the aggregates
`(14,15,17-19).
`
`Characterization of the Aggregation-Competent Species
`
`Based mostly on studies of thermally induced precipita(cid:173)
`tion, research on protein aggregation first led to the proposal
`that protein aggregates form from the fully unfolded state
`[reviewed by Dong et al. (11)]. Subsequent research has led to
`the hypothesis that partially unfolded states aggregate (14,79-
`86). These partially unfolded states (also called molten glob(cid:173)
`ules or acid-denatured "A" states) generally adopt a col(cid:173)
`lapsed conformation that is more compact than the unfolded
`state and has substantial secondary structure and little tertiary
`structure (79). They have large patches of contiguous surface
`hydrophobicity and are much more prone to aggregation than
`both native and completely unfolded conformations (14).
`Recently, several studies have found that even under
`physiologic solution conditions that are not perturbing of pro(cid:173)
`tein tertiary structure and that thermodynamically greatly fa(cid:173)
`vor the native state, proteins can form aggregates and pre(cid:173)
`cipitate (17-19,87). The protein native conformation is flex(cid:173)
`ible and does not exist as a discrete, single structure
`(12,63,88,89). Rather, at any instant in time, there exists an
`ensemble of native substates with a distribution of structural
`expansion and compaction. Kendrick et al. showed that the
`aggregation of rhiFN-)' proceeds through a transiently ex(cid:173)
`panded conformational species within the native state en(cid:173)
`semble (61). Compared to the most compact native species,
`the expanded species has a 9% increase in surface area
`(18,87). This conformational expansion is only about 30% of
`that required for the complete unfolding of rhiFN-)' (87).
`Furthermore, Webb et al. showed that the surface area in(cid:173)
`crease to form the structurally expanded species that precedes
`rhiFN-)' aggregation is independent of GdnHCl concentra(cid:173)
`tion, pressure, or temperature, suggesting a common interme(cid:173)
`diate for aggregation under these various stresses (87).
`Krishnan et al. recently showed that under physiologic
`conditions (neutral pH, 37°C, with no added denaturants),
`where the native state is greatly favored thermodynamically,
`rhGCSF aggregated readily (17). The surface area increase
`needed to form the expanded conformation leading to aggre(cid:173)
`gation was only approximately 15% of that for unfolding (17).
`
`Aggregation Models, Energetics, and Rates
`
`In order to transform protein molecules from natively
`folded monomers (or higher-order native assemblies, e.g., na(cid:173)
`tive dimers) to structurally perturbed, higher-order aggre-
`
`Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
`
`Ex. 2018-0004
`
`

`
`Stability of Proteins in Aqueous Solution
`
`1329
`
`N H TS* --7 A 1
`
`(1a)
`
`Ar +Am --7 Am+I
`(1b)
`Scheme L Lumry-Eyring framework of protein aggregation.
`
`(2a)
`2N H 2N* --7 A2
`(2b)
`N* +An --7 An+l
`(2c)
`A 2 +Am --7 Am+2
`Scheme 2. rhGCSF aggregation mechanism with N* as the transition
`state species (17).
`
`gates, protein molecules in the native state need to undergo
`both structural changes and assembly processes. The aggre(cid:173)
`gation pathways of many proteins have been analyzed in the
`well-known Lumry-Eyring framework (3,18,90,91). A repre(cid:173)
`sentation of this framework, shown in Scheme 1, involves a
`reversible conformational change of a protein (Scheme 1a)
`followed by irreversible assembly of the nonnative species to
`form aggregates (Scheme 1b) (3,90,91).
`In Scheme 1, N is the native protein, TS* represents the
`transition state preceding the formation of an aggregation
`intermediate A 1, and Am and Am+I are aggregates containing
`m and m + I protein molecules, respectively.
`It is generally known that the rate of a reaction is con(cid:173)
`trolled by both thermodynamics and kinetics. The transition
`state theory used in the model depicted in Scheme 1 can be
`graphically represented on a reaction coordinate diagram as
`shown in Fig. 1. The free energies of reactant (N), transition
`state (TS*), and products (An and Am) are shown on an ar(cid:173)
`bitrary free energy y-axis. The x-axis represents the course of
`individual reaction events. Am is expected to be favored ther(cid:173)
`modynamically and therefore has the lowest free energy.
`Each reaction proceeds through energy barriers (curved lines
`in Fig. 1), which represent energies of the different molecular
`configurations between reactants and products. The maxi(cid:173)
`mum energy configuration is the transition state, and the free
`energy difference between the transition state and reactant is
`called activation free energy (JiG*). For a multiple-step reac(cid:173)
`tion, such as protein aggregation, the step that has the highest
`JiG* is the rate-limiting step.
`Scheme 1 describes a reversible reaction to form a tran(cid:173)
`sition state, followed by irreversible reactions. The reaction
`order for the rate-limiting step determines the apparent order
`of the aggregation reaction. A number of proteins have been
`found to follow first order aggregation kinetics (3,18), sug(cid:173)
`gesting that the rate-limiting step is unimolecular (e.g., a con(cid:173)
`formational change) rather than a bimolecular reaction lim(cid:173)
`ited by collision frequency.
`In contrast, the aggregation of rhGCSF in pH 7 phos-
`
`Transition
`state
`(TS')
`
`Activation
`free
`energy
`
`Free
`Energy
`(G)
`
`Native
`protein
`(N)
`
`Aggregation
`competent
`intermediate
`(A.)
`
`Aggregate
`(Am)
`
`Aggregation Reaction Coordinate
`
`Fig. L Schematic reaction coordinate diagram of a protein aggrega(cid:173)
`tion (Scheme 1) on an arbitrary free energy y-axis. Curved lines
`illustrate kinetic energy barriers.
`
`phate buffer saline (PBS) follows a second-order reaction,
`suggesting that the rate-limiting step is biomolecular (17).
`Krishnan et al. proposed the mechanism in Scheme 2 for rh(cid:173)
`GCSF aggregation. Native rhGCSF (N) undergoes a bimo(cid:173)
`lecular, second-order irreversible reaction (2N --7 A2 ) to form
`a dimeric aggregation-competent intermediate A2 • This step
`proceeds through the formation of a transition state N*,
`which is a transiently expanded conformational species within
`the native state ensemble (17). N* then irreversibly dimerizes
`to form A2 , and this step is rate limiting (17). A2 then under(cid:173)
`goes assembly reactions to form aggregates. N* could also
`react irreversibly with an existing aggregate An to form a
`larger aggregate An+l· The reaction coordinate diagram for
`Scheme 2 is shown in Fig. 2. Also shown in Fig. 2 is the
`unfolded state (U), which is thermodynamically unstable with
`respect to the native state by 9.5 kcaVmole (17,20).
`
`Role of Conformational Stability
`
`It is apparent that the intrinsic conformational stability of
`the protein native state plays an important role in aggrega(cid:173)
`tion. First, aggregation is accompanied by the loss of native
`protein structures. Second, partially unfolded protein mol(cid:173)
`ecules are especially prone to aggregation. Third, the aggre(cid:173)
`gation transition state of some proteins has been identified as
`a structurally expanded species within the protein native state
`ensemble (17,18). Hence, aggregation is governed by the con(cid:173)
`formational stability of the protein native state relative to that
`of the aggregation transition state.
`Kendrick et al. showed that the addition of a thermody(cid:173)
`namic stabilizer (e.g., sucrose) that increased J1G,,1 of
`
`Free
`Energy
`(G)
`
`N
`
`Aggregation Reaction Coordinate
`
`Fig. 2. Schematic reaction coordinate diagram of rhGCSF aggrega(cid:173)
`tion in pH 7 PBS. N* is the transitions state species, and llGt NN* is the
`activation free energy of aggregation. A2 is the dimeric aggregation
`intermediate. Dotted arrows illustrate, relative to protein native state
`(N), shifts in the free energies of unfolded state (U) and N* when
`sucrose is added (20).
`
`Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
`
`Ex. 2018-0005
`
`

`
`1330
`
`Chi et al.
`
`rhiFN-)' decreased its aggregation rate (Fig.