The Stability Factor: Importance in Formulation Development
`
`Current Pharmaceutical Biotechnology, 2002, 3, 361-371
`
`361
`
`Rajesh Krishnamurthy1 and Mark C. Manning2,*
`
`1Human Genome Sciences, Inc., Rockville, MD and 2Center for Pharmaceutical Biotechnology, University of Colorado
`Health Sciences Center, Denver, CO, USA
`
`Abstract: Efficient development of stable formulations of protein pharmaceuticals requires an intimate knowledge
`of the protein and its chemical and physical properties. In particular, understanding the mechanisms by which a
`protein could degrade is critical for designing and testing formulations. This review describes the major pathways
`by which proteins can degrade, including denaturation, aggregation, oxidation, and interfacial damage. The
`methods to detect the degradation are covered, along with generalized strategies to retard or prevent each type of
`decomposition. Without an appreciation of the current best practices for devising stable formulations, the
`formulation process will be neither efficient nor optimal.
`
`INTRODUCTION
`
`Table 1.
`
`Summary of Some General Degradation Pathways
`
`formulation
`stabilization and
`Historically, protein
`development began as an empirical science owing the unique
`nature of each protein and the wide range of agents that can
`cause denaturation. More recently, as biotechnology-derived
`products began to appear on the market, there has been a
`concerted attempt to develop a “rational approach” to formu-
`lation development [1]. Such an approach necessarily relies
`on identifying the mechanism(s) of degradation in order to
`identify the appropriate countermeasures [2-5]. In fact, it is
`essential to recognize that protein instability actually refers
`to a number of degradation mechanisms, including both
`chemical and physical decomposition (Table 1). This review
`will examine four particular areas of protein instability:
`denaturation, aggregation, oxidation, and interfacial damage
`in detail. Once the specifics of any particular degradation
`pathway are understood, a more informed choice regarding
`excipients and formulation can be made, accelerating product
`development.
`
`THERMODYNAMIC STABILITY OF PROTEINS:
`DENATURATION
`
`In order to understand the thermodynamic stability of the
`overall structure of globular proteins, a summary of the
`interactions
`that affect protein
`folding
`is helpful.
`Manipulation of any of these elementary forces can result in
`either the stabilization or destabilization of the intrinsic
`stability of the protein. In this section, the dominant forces
`in protein, the stresses that can cause protein unfolding or
`denaturation, and analytical methods to evaluate the extent of
`denaturation are described. Finally, as is well established,
`addition of certain excipients can modulate the stability,
`through ligand binding mechanisms described by Wyman
`and Timasheff and co-workers [6-9]. An understanding of
`
`*Address correspondence to this author at the Center for Pharmaceutical
`Biotechnology, University of Colorado Health Sciences Center, Denver,
`CO; USA; Tel: 303-315-6162; E-mail: mark.manning@uchsc.edu
`
`Chemical Instabilities
`
`Physical Instabilities
`
`Oxidation
`
`Deamidation
`
`Proteolysis
`
`Denaturation
`
`Aggregation
`
`Precipitation
`
`Beta-Elimination
`
`Surface Adsorption
`
`Disulfide Scrambling
`
`these mechanisms is essential to formulating a protein
`properly in a liquid dosage form. It is important to note that
`additives,
`like sucrose, stabilize proteins by different
`mechanisms in the solid state than it does in solution (see
`below). In either case, knowledge of the mechanism of
`degradation and appreciation of the ability of various
`compounds to retard such processes is the basis for rational
`formulation design.
`
`Dominant Forces in Proteins
`
`Knowledge of the forces stabilizing proteins has come
`principally from X-ray crystallography and dynamic studies
`of the folding/unfolding process induced by heat or
`denaturants. The elementary interactions between amino acid
`residues that constitute the protein include electrostatic
`interactions, Van der Waals’ forces, hydrogen bonds, and
`hydrophobic interactions. Whether or not a particular amino
`acid segment of a protein (or the protein itself) is stable will
`depend on the net free energy summed over all of these
`interactions. Contrary to the perception that proteins are
`loose and floppy structures in solution, their residues are
`packed as tightly as they are in the crystalline state [10].
`
`Electrostatic Interactions
`The term, electrostatic interactions, describes interactions
`that occur between charged atoms or molecules. Depending
`
` 1389-2010/02 $35.00+.00
`
`© 2002 Bentham Science Publishers Ltd.
`
`Ex. 2032-0001
`
`

`
`362 Current Pharmaceutical Biotechnology, 2002, Vol. 3, No. 4
`
`Krishnamurthy and Manning
`
`on the ionization constant of the charged moiety, the
`magnitude of the interaction can be dependent on pH and/or
`ionic strength. In the early days of protein folding it was
`believed that the folding process was primarily driven by
`electrostatic interactions. The first quantitative model of
`electrostatic interactions was proposed by Linderstorm-Lang.
`This model assumes a continuous distribution of charge on
`the protein surface. While this model proved helpful in
`forming early ideas regarding protein folding, it was unable
`to account for the effects of specific charge groups. Thus, in
`recent
`literature, charge-based
`interactions have been
`delineated into two effects: classical electrostatic and specific
`charge interactions [11]. The classical electrostatic effects are
`the nonspecific interactions that arise when the protein is
`highly charged, such as at extreme pH values. The specific
`charge effects involve interactions, such as ion-pairing or salt
`bridges, between charged amino acids side chains that are in
`close spatial proximity. For example, the effect of salt
`bridges on the stability of isolated a
`-helices has been
`described [12, 13]. It is now believed that these specific
`interactions are important contributors to the difference in
`free energy between native and denatured states [14, 15].
`Certainly, there is ample evidence to suggest that salt
`bridges contribute to increased protein stability [16-18].
`However, it is not the dominant force in protein folding
`[11]. This conclusion stems from studies using model
`compounds on the expected change in the volume arising
`from electrostatic interactions [19] and the observation that
`thermodynamic stabilities of proteins show little dependence
`on pH or salt near the isoelectric point [20]. Structural
`studies indicate that ion pairs or salt bridges are neither
`highly conserved nor sufficiently numerous to be dominant
`in driving protein folding [16].
`
`Van der Waals’ Interactions
`The nonpolar interactions arise from interactions among
`fixed or induced dipoles. Although a nonpolar molecule has
`no net dipole when averaged over time, at a given instant
`there will be dipoles arising from fluctuations in the electron
`density around the molecule. These interactions, while
`numerous, are not believed to be dominant forces driving
`protein folding. This is because there are few methods that
`can distinguish the contributions to a protein’s stability from
`these interactions compared to hydrogen bonding. As a
`result, the magnitudes of these interactions are very difficult
`to assess experimentally. However, calculated values suggest
`that these forces can make significant contributions
`to
`protein stability [21, 22].
`
`Hydrogen Bonding
`A hydrogen bond occurs when a hydrogen atom is shared
`between two atoms, usually electronegative. The bond
`strength is dependent on the electronegativity and the
`orientation of the bond between the atoms (more linear
`hydrogen bonds are stronger). The importance of hydrogen
`bonding in protein folding and conformation has been
`reviewed [23, 24]. Of particular interest is the peptide
`hydrogen bond, which is responsible for the formation of
`secondary structures such as a
`-helix and b -sheets. In these
`ordered regions, buried carbonyl groups form hydrogen
`
`bonds with a peptide NH group. The energies of hydrogen
`bonds have been variously estimated between 2 - 10
`kcal/mole [23]. This is of importance because it suggests
`that while the bonds are stable enough to provide sufficient
`binding, they are also weak enough to permit rapid
`dissociation [10].
`
`Hydrophobic Interactions
`Kauzmann made the case for hydrophobic interactions
`being the dominant force in protein folding (i.e. it is able to
`explain why the folded state is preferred over the unfolded
`state) [25]. Two observations driving Kauzmann's conclu-
`sions were that nonpolar solvents denature proteins and that
`proteins denature at low temperatures. Since nonpolar solutes
`become more soluble in water at low temperatures [26],
`Kauzmann believed that the cold destabilization resembled
`nonpolar solvation. A sizable amount of research supports
`Kauzmann’s hypothesis. These include the observations that
`globular proteins possess a hydrophobic core [20, 27], where
`they largely avoid contact with water. Moreover,
`the
`conservation of these hydrophobic residues [16], and the
`resemblance between temperature-dependent unfolding of
`proteins and the temperature-dependent transfer of nonpolar
`solutes into water reinforces this view [26].
`
`At least three different descriptions of the nature of
`hydrophobic interactions exist [11]. Hydrophobic interac-
`tions have been visualized as the transfer of a nonpolar solute
`to an aqueous solution, as the rearrangement of
`the
`hydrogen-bonding network in water around a nonpolar
`solute, and as the transfer of nonpolar solutes into an
`aqueous solution only when a characteristic temperature
`dependence is observed. The most common visualization,
`from a formulation development perspective, is that of the
`ordering of water molecules around a nonpolar solute,
`thereby “driving” a hydrophobic molecule into the hydro-
`phobic region of a protein.
`
`Small molecules such as urea, diketopiperazines,
`benzene, and individual amino acids have been used to
`determine the stabilizing effects of hydrogen bonding[11],
`hydrophobic interactions [26] (by measuring the energy of
`transfer from aqueous solvent to non-polar solvents
`to
`simulate burial in the core of the folded protein), and
`electrostatic interactions [10]. While these small molecule
`experiments have helped us understand some of the factors
`influencing protein folding, a protein cannot be treated as a
`sum of small molecule models. Quantitative predictions that
`have relied on such assumptions have not been successful.
`One of the reasons is because in addition to the translational,
`rotational, and vibrational entropies normally associated with
`small molecules, there are other entropic contributions,
`which come into play when considering macromolecules
`such as proteins. One such entropic contribution
`is
`configurational entropy. Flory introduced the concept of
`configurational entropy of a polymer and attributed it to the
`(steric) constraints on the numerous ways of arranging
`polymer and solvent molecules [28]. To put it differently, it
`is a function of the properties of the protein and the solvent
`and cannot be completely described by small molecule
`studies [11].
`
`Ex. 2032-0002
`
`

`
`The Stability Factor: Importance in Formulation Development
`
`Current Pharmaceutical Biotechnology, 2002, Vol. 3, No. 4 363
`
`Stresses Causing Denaturation
`
`Tanford has defined denaturation as a major change from
`the native structure without an accompanying change in the
`primary structure [20]. Denaturation, or loss of some aspect
`of the globular structure of the protein, is peculiar to
`macromolecules, thus presenting a challenge unique to the
`formulation scientist. Proteins are dynamic ensembles and,
`at any given instant, there is equilibrium between native and
`non-native forms. The native and near-native forms, which
`can include partially folded versions of the protein, interact
`with each other as well as any other molecules in solution.
`Consequently, the presence of high concentrations of
`chaotropes or extreme pH conditions is not a prerequisite for
`shifting
`the equilibrium population of
`the ensemble,
`although such conditions will certainly have an effect.
`Appreciating the dynamic and transient nature of the protein
`ensemble helps understand the factors influencing long-term
`stability of a protein pharmaceutical.
`
`The vast amount of literature devoted to protein stability
`indicates that any approach that provides for increased
`stability towards one form of denaturation typically increases
`the stability towards all forms of denaturation. Therefore,
`only a few possible stresses will be examined in detail. In
`particular, the effects of chaotropes, elevated temperature, and
`pH extremes will be considered here.
`
`Reversible and Irreversible Denaturation
`As the native structure of a protein is disrupted because
`of a change in the environment (pH, heat, presence of a
`chaotrope), it is important to determine whether the perturba-
`tion is permanent or temporary. Reversible denaturation
`refers to loss of native structure that is recovered once the
`stress is removed. Sometimes reversible denaturation steps
`occur during a manufacturing process. For example, in order
`to obtain a crude isolate, the precipitation of human serum
`albumin occurs during the plasma fractionation process. In
`contrast, irreversible denaturation leads to a loss of the native
`structure that cannot be regained. The extent of irreversible
`denaturation is dependent on both the protein in question
`and stress applied [29]. Typically, the term irreversible
`denaturation indicates that some other degradation process
`has occurred, such as aggregation or chemical modification.
`One possible chemical modification leading to irreversibility
`is disulfide exchange. Cross-linking with disulfides leads to
`new covalent linkages that do not allow native structure to
`be recovered, no matter how
`long
`the sample sits
`unperturbed. It is also possible for chemical modifications to
`induce changes to the secondary and/or tertiary structure
`leading to aggregation [30].
`
`Temperature-Induced Unfolding
`A large amount of literature exists that demonstrates the
`adverse effects of increasing temperature on proteins [20, 23].
`However, it is important to note that once the temperature is
`elevated, and a significant fraction of the protein ensemble is
`unfolded, aggregation can proceed rapidly, resulting in
`irreversibility, making
`thermodynamic analysis of
`the
`unfolding
`data
`impossible.
`Therefore,
`obtaining
`thermodynamic information may require low concentrations
`of protein to be used.
`
`On the other hand, since freeze-drying has become a
`standard approach in pharmaceutical processing, denaturation
`at low temperatures (i.e., cold denaturation) has become of
`great interest [31]. While it is probable that more damage
`occurs from freeze concentration of solutes and ice-water
`interfacial damage [32], it is essential that the formulation
`scientist realize that proteins unfold at low as well as high
`temperature.
`
`Pressure-Induced Unfolding
`Protein denaturation can be caused by high static
`pressure, leading to the loss of ordered structure (secondary,
`tertiary, and quaternary). In the case of multimeric proteins,
`both increased association and dissociation of the multimers
`into monomers can be achieved, depending on the pressure
`applied. While the denaturation at high pressure is expected
`to be reversible [33, 34], irreversible denaturation has been
`observed in lactate dehydrogenase and several other enzymes
`[35, 36].
`
`While high pressure is usually considered to be
`destabilizing, the effects are quite sensitive to the amount of
`pressure applied. For example, high pressure has been
`observed to reduce the rate of aggregation from soluble
`aggregates [37]. Not only can aggregation rates be retarded,
`but high pressure can also be used to refold protein from
`insoluble aggregates [38]. The work using recombinant
`b -lactamase
`human growth hormone,
`lysozyme, and
`demonstrated that high pressure might be a useful tool to
`refold proteins from insoluble aggregates and inclusion
`bodies.
`
`Chaotrope-Induced Unfolding
`Urea and guanidine hydrochloride are two commonly
`used chaotropes (denaturants), since the denaturation has
`been found to be nearly complete for most proteins when
`exposed to high concentrations of the compounds in aqueous
`solution [39]. Under these conditions, the denatured state
`lacks large amounts of ordered structure and is believed to be
`as close to a truly unfolded state of the protein as possible.
`
`Analyzing Unfolding Curves
`
`The primary assumption in using denaturing curves for
`the determination of conformational stability is that protein
`denaturation is a cooperative process, i.e.,
`the native
`structure is lost simultaneously rather than incrementally
`[40]. This cooperativity is expected to lead to an all-or-none
`behavior that is responsible for abrupt transitions between
`the native and denatured states. The two-state assumption of
`protein unfolding is not necessary in order to perform the
`analysis though it greatly simplifies calculations [41,42].
`
`The unfolding of the protein can be monitored using any
`of a number of techniques, including UV absorption,
`fluorescence, infrared, and circular dichroism (CD) spectro-
`scopy. Whichever method is selected, it is important to pick
`a wavelength where the signal from the chromophore of
`interest differs significantly between the native and denatured
`states. The details of how to analyze the resultant curves can
`be found in reviews from Pace and co-workers [23, 24].
`
`Ex. 2032-0003
`
`

`
`364 Current Pharmaceutical Biotechnology, 2002, Vol. 3, No. 4
`
`Krishnamurthy and Manning
`
`Ultimately, what is obtained from the analysis of an
`unfolding curve is an estimate of the equilibrium constant
`between the native and unfolded states. This can be directly
`converted into a standard free energy of denaturation, that is,
`the relative difference in free energy between the native and
`unfolded states. However, for curves involving the use of
`chaotropes, such as urea or guanidine hydrochloride
`(GdmHCl), it ought to be remembered that the D G value is
`obtained under denaturing conditions. If the actual stability
`(i.e., the free energy of denaturation with no chaotrope
`present) is desired, then this value must be extrapolated to
`the D GH2O by one of three methods - the denaturant binding
`model, Tanford’s model, or the linear extrapolation model
`(Shirley, 1992) [40-42]. This value (D GH2O) can then be used
`to compare the relative stability of a protein in a particular
`formulation. It should be noted that chaotrope-induced
`denaturation can be used to measure the free energy of
`unfolding, even in the presence of stabilizers, such as sucrose
`[124].
`
`Experimental Approaches for Monitoring Denaturation
`
`Denaturation of a protein can be accomplished with a
`number of different stresses, as outlined above. The exact
`characteristic of the denatured state is dependent on the
`manner of denaturation, whether by heat, pH, chaotrope, and
`so on. Consequently,
`the experimental approaches
`to
`measure denaturation are also varied. This section briefly
`describes three such analytical methods that allow unfolding
`curves to be determined. In addition, the use of differential
`scanning calorimetry (DSC) to determine the heat changes
`associated with denaturation is also described.
`
`Differential Scanning Calorimetry (DSC)
`The differential scanning calorimeter (DSC) measures the
`change in the heat capacity of a protein as the temperature is
`increased. Privalov and Makhatadze [43] demonstrated that
`the positive change in heat capacity that accompanies
`unfolding is mainly due to the hydration of polar and
`nonpolar groups in a protein. The change in heat capacity is
`then used to determine the enthalpy and entropy of
`unfolding, which is then used to determine the free energy of
`unfolding. For example, Remmele et al. successfully used
`DSC to screen for different formulations for interleukin-1
`receptor and established a correlation between the tendency of
`the receptor to aggregate and the stability ranking based on
`the melting temperature [44].
`
`Gomez et al. [45] came up with a single function that can
`be used to represent the partial molar heat capacity of the
`native and unfolded states of the protein, which is a function
`of the molecular weight, the solvent accessible areas (of the
`polar and nonpolar residues) and the total area buried from
`the solvent. The universal function enables us to determine
`the stability of a protein provided we know the three-
`dimensional structure. While at first glance it might appear
`that such knowledge is not very useful, this approach
`improves our understanding of
`the bases for protein
`stabilization. A case in point is the calorimetric analysis of
`the heat shock protein, DnaK, by Montgomery et. al. [46].
`They were able to develop a quantitative model of the
`
`folding/unfolding behavior of DnaK and the cooperative
`domain structure of the protein under equilibrium based on
`knowledge of a homologous 44kDa N-terminal fragment of a
`related protein.
`
`However, the temperature marking the onset of unfolding
`(which is dependent on the scan rates employed) is often a
`better parameter to evaluate the stability of a protein in
`different formulations [47, 48]. Once a significant fraction of
`the protein ensemble is highly unfolded, the probability of
`aggregation increases dramatically, especially at the high
`protein concentrations used in DSC experiments [49].
`
`Circular Dichroism
`Circular dichroism (CD) spectroscopy allows one to
`characterize both the secondary structure composition of a
`protein in solution, as well as the localized environment
`around the aromatic side chains in the protein (often thought
`to indicate the state of the tertiary structure). The role of CD
`spectroscopy in assessing changes in protein structure has
`been widely reviewed. Furthermore, its place in assessing
`protein stability has been described as well [50].
`
`The use of CD to monitor unfolding is widespread, as
`one can follow both secondary and tertiary structure changes,
`as has been done with insulin [51]. This is especially true for
`-helical proteins, which display a characteristic negative
`band at 222 nm. Intensity at this wavelength has been long
`thought to correlate directly to the a
`-helical content of the
`protein [52]. Therefore, a great many unfolding curves plot
`intensity at 222 nm vs. the particular stress (e.g., increased
`temperature, concentration of chaotrope), resulting in a
`sigmoidal curve to be analyzed as described earlier.
`
`Fluorescence Spectroscopy
`In general, fluorescence based methods are more popular
`since it requires far less protein than the other spectroscopic
`techniques [53]. The use of fluorescence spectroscopy in
`urea/GdmHCl curves is outlined above.
`In addition,
`fluorescence spectroscopy can shed some light on the tertiary
`structure of proteins. The emission spectrum of tryptophan
`provides information on the location of tryptophan. One
`advantage of fluorescence spectroscopy is its ability
`to
`provide information on
`insoluble as well as soluble
`aggregates. This advantage was utilized by Shahrokh et al.
`to determine the mechanism of particulate formation in basic
`fibroblast growth factor [54].
`
`Infrared Spectroscopy
`Infrared (IR) spectroscopy is widely used to measure the
`secondary structure content of a protein, as it can be used for
`both solids and solutions [55]. For example, infrared
`spectroscopy is used widely to assess the effectiveness of
`freeze-drying formulations [56]. The techniques can be used
`qualitatively, as in the study by Lin et. al. [57], where they
`demonstrated a correlation between the extent of unfolding in
`dried human serum albumin, as measured by infrared
`spectroscopy after processing, and the amount of soluble
`aggregates upon dissolution. In addition, it can provide
`unfolding curves similar to those generated using other
`methods [58].
`
`Ex. 2032-0004
`
`a
`

`
`The Stability Factor: Importance in Formulation Development
`
`Current Pharmaceutical Biotechnology, 2002, Vol. 3, No. 4 365
`
`Mass Spectrometry
`Usually, mass spectrometry is not considered when
`evaluating what technique to use
`to monitor protein
`unfolding. However, a novel use of mass spectrometry was
`recently reported by Ghaemmaghami et al. that could alter
`the way mass spectrometry is employed in protein stability
`studies [59]. The particular application of the technique relies
`on the fact that certain amide protons on a protein exchange
`with the solvent only when there is global unfolding. By
`monitoring the degree of exchange in a protein dissolved
`D2O/H2O in the presence of denaturants (in this case,
`GdmHCl), they were able to measure the average increase in
`molecular weight at each GdmHCl concentration. As the
`protein unfolds, the mass increase is much larger, and one
`can obtain an unfolding curve. While their paper studied the
`stability of different variants of small protein, the concept
`could be used to screen the effects of different formulations.
`For example, it is known that addition of sucrose shift urea
`unfolding curves, demonstrating its ability to stabilize a
`protein even in the presence of a chaotrope [60]. Although
`the authors were unable to make absolute determinations of
`the free energy of unfolding, the relative rank ordering of the
`mutants was correct compared to
`traditional unfolding
`studies.
`
`Stabilization in Aqueous Solution
`
`Probably, one of the most important lessons to be
`learned for a formulation scientists working on proteins is
`the mechanism by which low molecular weight solutes affect
`protein conformational stability. Timasheff and co-workers
`have described the general process for many years [6, 8, 9,
`61, 62], building on concepts of Wyman [7]. In general,
`stabilizers act either as ligands that preferentially bind or are
`excluded from the surface of the protein. Those that bind
`directly to the native state, thereby lowering the free energy
`of that state relative to the unfolded state, lead to net
`stabilization of
`the protein. Substrate stabilization of
`proteins is a well-known consequence of this mechanism
`[63]. Binding of metals can serve the same purpose as well.
`For example, the stability of human growth hormone is
`increased by the binding of zinc [64]. Other metals, such as
`calcium and magnesium, have been shown enhance stability
`as well [65].
`
`In addition to stabilization from direct ligand binding,
`there is another mechanism for increasing the thermodynamic
`stability of a protein. In this case, additives that exhibit
`negative (or excluded) binding to the native state can also
`stabilize proteins, as is seen with sugars, amino acids, and
`certain salts [6]. In this case, the additive displays negative
`binding to the native state, resulting in an increase in its
`chemical potential. In other words, such an interaction
`destabilizes the native state. However, it is essential to
`examine the interaction with the unfolded state as well. If the
`same ligand is also excluded from the denatured state, given
`the larger surface area of most unfolded states, the increase in
`chemical potential will be even greater for this state. The
`result is a net increase in the free energy difference between
`the native and unfolded state, exhibiting itself as a net
`stabilization relative to denaturation.
`
`Stabilization in the Solid State
`
`to
`is a key process
`While preferential exclusion
`understand when considering the design of a stable liquid
`formulation, a comprehensive strategy to achieve stable
`liquid formulations has not yet emerged. In contrast, the
`ability to design a stable lyophilized protein formulation
`rationally is more highly developed [1]. In this case, one
`must guard against two significant stresses, freezing and
`drying. The former requires all of the skills of stabilizing a
`protein against damage in solution, as well as preventing
`interfacial damage that can occur (see below). Furthermore,
`one must understand that if freezing is slow, there can be
`significant increases in solute concentration in the maximally
`freeze concentrated state, possibly as much as 15-fold. In
`other words, a 20 mM buffer could become 300 mM before
`the solution is completely frozen. Once freezing commences,
`there can be damage at the ice-water interface; in the same
`way, proteins can experience degradation at air-water and
`solid-water interfaces [32]. These concerns should lead the
`formulation scientist to evaluate the freeze-thaw stability of a
`new drug candidate. Studies describing how one approaches
`this problem have been reported [66, 67]
`
`During drying, another challenge presents itself. Now
`water of hydration is being stripped away. Therefore, the best
`stabilizers are those that adequately replace the water being
`removed [68]. In general, this means sugars, such as
`trehalose and sucrose. While glucose and maltose also work,
`they are reducing sugars and Maillard-type reactions could
`ensue [69]. It appears that there is a certain minimal amount
`of sugar required to stabilize a lyophilized protein, roughly
`about 1:1 on a weight basis. A detailed example of this
`behavior can be seen in a recent study on the stabilization of
`HER2 antibody in the freeze-dried state [70]. In brief, there
`needs to be a sufficient amount of an amorphous phase where
`the protein can have enough water molecules replaced by
`sugar molecules in order to maintain structural stability and
`limit motion. However, one must be aware of the possibility
`of phase separation behavior [71], where the protein and the
`amorphous phase do not mix intimately. If the sugar does
`not reside in the immediate vicinity of the protein, no
`stabilization occurs. This behavior has been seen with
`polymeric excipients, such as dextran [72]. Similarly, if the
`stabilizing sugar begins to crystallize, phase separation can
`occur as well [73, 74].
`
`AGGREGATION OF PROTEINS
`
`Aggregation and precipitation of proteins is of concern
`because of potential alterations in immunogenicity, toxicity,
`and efficacy [75, 76]. The ability of proteins to form
`aggregates and precipitates has been widely reviewed [77-79].
`Aggregation can occur during any number of processing
`steps,
`such
`as
`purification,
`concentration,
`filling,
`freeze/thaw, lyophilization, and during delivery.
`
`For purposes of this discussion, aggregation of proteins
`refers to the association of nonnative forms of the protein,
`often occurring in conjunction with some other degradation
`process, such as denaturation or interfacial damage. The
`classic model of protein aggregation envisages the protein
`
`Ex. 2032-0005
`
`

`
`366 Current Pharmaceutical Biotechnology, 2002, Vol. 3, No. 4
`
`Krishnamurthy and Manning
`
`existing in either a native (N), intermediate (I), and unfolded
`(U) state. Kinetic models of protein aggregation have been
`constructed, resulting in the well-established Lumry-Eyring
`model [80, 81]. These mechanisms assume aggregation
`proceeds from the unfolded state, where the native state
`converts to the aggregation-competent state (A), followed by
`association of A to form multimers, Am+1 (see Scheme 1). If
`reaction (b) in Scheme 1 is rate-limiting, then the kinetics
`are second-order in protein concentration. A modification of
`the well-known mechanism has been devised to account for
`aggregation processes that follow first-order kinetics (see the
`work by Kendrick et al. described below).
`
`Once nonnative multimers form, they can either stay in
`solution (soluble aggregates) or precipitate (insoluble
`aggregates). It is important to note that precipitation is
`distinct from processes such as salting out. Whereas
`precipitation occurs from association of protein molecules
`that are structurally perturbed, salting out leads to insoluble
`
`5. This growth continues until the solubility limit of the
`aggregate is exceeded and the protein falls out of solution
`as a precipitate. Such a precipitate does not usually
`dissolve to give the native protein again.
`
`Controlling Aggregation
`
`In theory, aggregation is a second (or higher) order
`kinetic process and is therefore highly dependent on protein
`concentration. However, this assumes knowledge of the
`kinetics of protein aggregation [Note: The order of a reaction
`with respect to a specific species involved in the reaction
`refers to the power to which the concentration of that species
`is
`raised
`in a
`rate expression of
`the
`form
`rA =
`
`
`kCAaCBb….where rA is the reaction rate, k is the reaction rate
`constant, CA is the concentration of species A, CB is the
`concentration of species B and so on. By default, in
`reactions involving proteins, the order is with reference to
`
`Scheme 1.
`
`protein that is still quite native-like and active. For example,
`salting