[raghwendra][6 X 9 Tight][D:/informa_Publishing/REY_H100116_2400039/z_pro-
`duction/z_3B2_3D_files/978-1-4398-2575-4_CH0007_O.3d] [6/5/010/16:4:50]
`[167–197]
`
`7
`
`Freezing- and Drying-Induced Perturbations
`of Protein Structure and Mechanisms of
`Protein Protection by Stabilizing Additives
`
`John F. Carpenter and Ken-ichi Izutsu
`University of Colorado Health Sciences Center, Denver, Colorado, U.S.A.
`
`Theodore W. Randolph
`University of Colorado, Boulder, Colorado, U.S.A.
`
`INTRODUCTION
`There are numerous unique, critical applications for proteins in human
`healthcare (1–3). However, even the most promising and effective protein
`therapeutic will not be of benefit if its stability cannot be maintained during
`packaging, shipping, long-term storage, and administration. For ease of prepa-
`ration and cost containment by the manufacturer and ease of handling by the
`end user, an aqueous protein solution often is the preferred formulation.
`However, proteins are readily denatured (often irreversibly) by the numerous
`stresses arising in solution,
`for example, heating, agitation,
`freezing, pH
`changes, and exposure to interfaces or denaturants (4). The result is usually
`inactive protein molecules and aggregates, which compromise clinical efficacy
`and increase the risk of adverse side effects (5). Even if its physical stability is
`maintained, a protein can be degraded by chemical reactions (e.g., hydrolysis
`and deamidation), many of which are mediated by water. Thus, inherent protein
`instability and/or the logistics of product handling often precludes develop-
`ment of aqueous, liquid formulations (6,7). Also, simply preparing stable frozen
`products, which is relatively straightforward,
`is not a practical alternative
`because the requisite shipping and storage conditions are not technically and/or
`economically feasible in many markets.
`The practical solution to the protein stability dilemma is to remove the water.
`Lyophilization (freeze-drying) is most commonly used to prepare dehydrated pro-
`teins, which, theoretically, should have the desired long-term stability at ambient
`temperatures. However, as will be described in this review, recent infrared spec-
`troscopic studies have documented that the acute freezing and dehydration stresses
`of lyophilization can induce protein unfolding (8–11). Unfolding not only can lead to
`irreversible protein denaturation, even if the sample is rehydrated immediately, but
`can also reduce storage stability in the dried solid (12,13).
`Moreover, simply obtaining a native protein in samples rehydrated imme-
`diately after lyophilization is not necessarily indicative of adequate stabilization
`during freeze-drying or predictive of storage stability. Many proteins unfold
`during lyophilization but readily refold if rehydrated immediately (8,11,14).
`Without directly examining the structure in the dried solid, it is not possible to
`know whether an unfolded protein with poor storage stability is present or not.
`
`“This chapter is a direct repeat of the text that appeared in Freeze-Drying/Lyophilization of Pharma-
`ceutical and Biological Products, Second Edition, Revised and Expanded (Rey L and May J, eds.) 2004,
`Marcel Dekker, Inc., New York.”
`
`167
`
`Ex. 2017-0001
`
`

`
`[vijay][6 X 9 Tight][D:/informa_Publishing/REY_H100116_2400039/z_production/
`z_3B2_3D_files/978-1-4398-2575-4_CH0007_O.3d] [30/4/010/14:17:50] [167–
`197]
`
`168
`
`Carpenter et al.
`
`To develop a protein formulation that minimizes protein unfolding during
`freezing and drying, it is crucial that the specific conditions (e.g., pH and specific
`for optimum protein stability be established and the
`stabilizing ligands)
`appropriate nonspecific stabilizing additives (i.e., those excipients that generally
`stabilize any protein) be incorporated into the formulation. Other physical
`factors—the glass transition temperature and the residual moisture of the dried
`solid—must also be optimized to assure storage stability in the dried solid
`(reviewed in Ref. 15). These aspects of developing a lyophilized protein for-
`mulation will not be considered here because they are addressed in other
`chapters in this volume as are the interplay between formulation, lyophilization
`cycle design, cake structure, and long-term stability of proteins (15). Here we
`will describe how to design formulations that protect proteins during both
`freezing and drying and the mechanisms by which additives stabilize proteins
`and, also importantly, fail to do so. In addition, we will give an overview of the
`use of infrared spectroscopy to directly monitor protein conformation in frozen
`and dried samples. This structural
`information is crucial for the rational
`development of stable, lyophilized protein formulations.
`We wish to emphasize that the principles and mechanisms to be discussed
`should be generally applicable to any protein. However, each protein has unique
`physicochemical characteristics, which often manifest themselves as specific
`routes of chemical and physical degradation during storage. Although we will
`not address chemical degradation directly in this chapter (see earlier works in
`Refs. 15–17), it is important to realize that minimizing unfolding during freezing
`and drying can reduce such degradation during lyophilization and subsequent
`storage (13). Currently, it is not possible to predict if degradation of a given protein
`will be inhibited by simply designing a formulation to maintain native structure,
`nor is it clear as to why the efficacy of “general” protein stabilizers often varies
`depending on the protein being studied. Thus, there is a great need to increase the
`fundamental understanding of the mechanisms by which protein stabilizers act
`and to document, by case studies, the applicability of the general rules to indi-
`vidual proteins. With sufficient effort by academic and industrial researchers, this
`can be an iterative process in which progress can be made toward developing a
`general strategy for protein formulation that can be rationally modified for the
`successful lyophilization of each new protein product.
`
`PROTEIN STABILIZATION DURING
`LYOPHILIZATION/REHYDRATION
`Much of the early research on protein stabilization during lyophilization was
`with labile enzymes, which were found to be irreversibly inactivated, presum-
`ably due to aggregation of nonnative molecules, to varying degrees after rehy-
`dration. As such, attempts at improving the recovery of activity were focused on
`the entire process of lyophilization and rehydration. It was not known at what
`point(s) during the process the damage arose and the stabilizers were operative.
`Also, usually these studies tested the capacity of nonspecific stabilizers (i.e.,
`those that will generally protect any protein) to prevent irreversible protein
`denaturation (i.e., aggregation) and inactivation. However, for practical purposes,
`the first step in increasing the resistance of a given protein to lyophilization-
`induced damage is to choose the specific conditions that provide the greatest
`stability to that protein. In general, any factor that alters the free energy of
`
`Downloaded by [John Sekella] at 17:14 30 June 2015
`
`Ex. 2017-0002
`
`

`
`[vijay][6 X 9 Tight][D:/informa_Publishing/REY_H100116_2400039/z_production/
`z_3B2_3D_files/978-1-4398-2575-4_CH0007_O.3d] [30/4/010/14:17:50] [167–
`197]
`
`Perturbations of Protein Structure
`
`169
`
`unfolding in solution will tend to have the same qualitative effect during lyo-
`philization. For example, the stability of many enzymes during freeze-thawing is
`altered by the presence of substrates, cofactors, and/or allosteric modifiers (18).
`Even for nonenzyme proteins, specific ligands can be important components of
`the formulation. For example, the stability of fibroblast growth factors is greatly
`increased in the presence of heparin or other polyanionic ligands (reviewed in
`Ref. 19). The pH and specific ligands that confer optimum stability often are
`known from purification protocols, preformulation studies, and/or earlier efforts
`at designing a liquid formulation.
`However, most proteins are not adequately stabilized by specific solution
`conditions. Of the nonspecific stabilizers that have been tested, sugars have been
`shown to stabilize the most proteins during lyophilization and have been known
`to have this property for the longest time. To our knowledge, the first published
`report is the 1935 paper by Brosteaux and Eriksson-Quensel (20) in which they
`described the protection during dehydration/rehydration of several proteins by
`sucrose, glucose, and lactose. Subsequent detailed comparisons of sugars docu-
`mented that usually disaccharides provide the greatest stabilization (4,8,21,22). For
`protection during the lyophilization cycle itself, both reducing and nonreducing
`disaccharides are effective. However, reducing sugars (e.g., lactose and maltose)
`can degrade proteins during storage via the Maillard reaction (protein browning),
`a process that can be accelerated at intermediate residual moisture contents (22,23).
`Therefore, the choice of disaccharides is essentially limited to the nonreducing
`sugars, sucrose and trehalose. Since, as of early 1998, trehalose has not been used
`in any Food and Drug Administration (FDA) approved parenteral product,
`sucrose is usually the first choice for commercial protein drug formulations.
`Although the data are much more limited, polyvinylpyrrolidone (PVP)
`and bovine serum albumin (BSA) have also been shown to protect a few tetra-
`meric enzymes, that is, asparaginase, lactate dehydrogenase (LDH), and phos-
`phofructokinase (PFK), during lyophilization and rehydration (24,25). Another
`class of compounds that has been found to be useful in freeze-dried for-
`mulations are nonionic surfactants. For example, sucrose fatty acid monoester,
`3-[(3-cholamidopropyl)-dimethylammonio]-1-propa-nesulfonate (CHAPS), and
`Tweens have been found to increase recovery of b-galactosidase activity after
`freeze-drying and rehydration (26). Various surfactants have been shown to
`protect LDH during freeze-drying and rehydration (27). Hydroxypropyl-
`b-cyclodextrin, which is surface active (28,29), inhibited the inactivation of
`recombinant tumor necrosis factor (30), interleukin-2 (31,32), and LDH (27)
`during freeze-drying/rehydration.
`
`MECHANISMS OF STABILIZATION OF PROTEINS BY SUGARS
`DURING DEHYDRATION
`Most protein pharmaceuticals are multicomponent systems that contain protein,
`buffer salts, bulking agents, and stabilizers. Each component has its intended
`role in the formulation. For example, often a crystallizing excipient (e.g., man-
`nitol or glycine) is chosen as a bulking agent (15). In contrast, numerous studies
`have documented that stabilization of a protein during dehydration requires the
`presence of a compound that remains at least partially amorphous. When a
`protein formulation is frozen, the protein partitions into the non-ice phase with
`other amorphous components. The interaction between the protein and these
`amorphous components must be maintained during the entire freeze-drying
`
`Downloaded by [John Sekella] at 17:14 30 June 2015
`
`Ex. 2017-0003
`
`

`
`[vijay][6 X 9 Tight][D:/informa_Publishing/REY_H100116_2400039/z_production/
`z_3B2_3D_files/978-1-4398-2575-4_CH0007_O.3d] [30/4/010/14:17:50] [167–
`197]
`
`170
`
`Carpenter et al.
`
`process to assure recovery of a native protein in the dried solid and after
`rehydration (8,9,33–36).
`Although most carbohydrates used for protein formulations remain
`amorphous in frozen solutions and during drying (e.g., sucrose and trehalose),
`some exhibit eutectic phase separation from frozen solutions (34–39). For
`example, mannitol readily crystallizes during freeze-drying, but the degree of
`crystallization can be manipulated by altering processing conditions and for-
`mulation components (34–39). In the concentration range where it remains
`mostly amorphous, mannitol has been shown to protect enzymes during freeze-
`drying in a concentration-dependent manner (35,36). A relatively high mass
`ratio of protein to mannitol will serve to inhibit mannitol crystallization,
`whereas with excess mannitol, crystallization and loss of stabilization arise.
`Similarly, substantial stabilization has been achieved with solutes (including
`buffer salts) that alone crystallize but in combination interfere with each other’s
`crystallization. For example, Izutsu et al. (35) found that with a sufficiently high
`ratio of potassium phosphate to mannitol, mannitol remained amorphous and
`protected LDH during freeze-drying. However, when there was excess man-
`nitol, its crystallization obviated protein protection. Similarly, Pikal et al. (40)
`found that appropriate ratio of mannitol and glycine resulted in a sufficiently
`large amorphous fraction to protect freeze-dried human growth hormone.
`Although it is well established that an amorphous excipient is needed to
`protect proteins during dehydration, the nature of the protective interaction of
`amorphous solutes with the protein in the dried solid has been a subject of
`controversy in the literature. There are at least two nonexclusive mechanisms
`proposed. Before describing these mechanisms, we wish to emphasize that
`neither mechanism alone is sufficient to fully explain stabilization during lyo-
`philization. Both mechanisms focus only on the effect of stabilizers during the
`terminal stress of dehydration and essentially ignore the freezing step. As
`documented below, no matter what the nature of the interaction of the additive
`with the dried protein, the most important factor is that the additive(s) prevent
`unfolding during both freezing and dehydration.
`Proponents of one mechanism state that proteins are simply mechanically
`immobilized in the glassy, solid matrix during dehydration (41). The restriction of
`translational and relaxation processes is thought to prevent protein unfolding, and
`spatial separation between protein molecules (i.e., “dilution” of protein molecules
`within the glassy matrix) is proposed to prevent aggregation. Although it is clear
`that protective additives must partition with the protein into the amorphous phase
`of the dried sample, simply forming a glassy solid does not assure protein sta-
`bilization. First, if all that were needed to prevent damage to a protein is the
`formation of a glass, then the protein by itself should be stable. Clearly, this is not
`the case because proteins themselves should form an amorphous phase in the
`dried solid (42); however, most unprotected proteins are denatured during lyo-
`philization (8–14). In some cases adding another protein (e.g., BSA), which should
`simply add to the mass of the final protein glass, confers protection (25).
`One might further qualify the mechanism by proposing that the requisite
`mechanical restriction to unfolding and aggregation can only be achieved if
`another amorphous compound is present to provide immobilization and spatial
`separation of the protein drug molecules. However, then the question becomes
`what amount of additive is sufficient to provide the desired physical properties
`of the dried solid, which are not achieved with the protein alone? This question
`
`Downloaded by [John Sekella] at 17:14 30 June 2015
`
`Ex. 2017-0004
`
`

`
`[vijay][6 X 9 Tight][D:/informa_Publishing/REY_H100116_2400039/z_production/
`z_3B2_3D_files/978-1-4398-2575-4_CH0007_O.3d] [30/4/010/14:17:50] [167–
`197]
`
`Perturbations of Protein Structure
`
`171
`
`has not been answered or addressed in the literature. However, it is expected
`that, in general, the capacity of an additive to protect a protein specifically
`during dehydration should depend on the final additive protein mass ratio.
`Increasing this ratio will favor spatial separation and immobilization of the
`protein within the glassy matrix. Also, the mass ratios between all compounds
`in the dried solid affect the influence of the compounds on each other’s crys-
`tallization (e.g., with glycine and mannitol).
`Several studies have shown that formation of a glassy phase by an addi-
`tive, even when it is used in large excess relative to the protein, is not a sufficient
`condition for acute stabilization of proteins during lyophilization. For example,
`formulations of 100 mg/mL interleukin-l receptor antagonist, prepared with
`sucrose concentrations ranging from 0% to 10% (wt/vol), formed a glass during
`lyophilization and all had glass transition temperatures of 66  28C (13). Yet
`only in formulations containing 5% sucrose was lyophilization-induced
`unfolding prevented. Tanaka et al. (43) have found that the capacity of carbo-
`hydrates to protect freeze-dried catalase decreased with increased carbohydrate
`molecular weight. Dextrans were the largest and least effective of all of the
`carbohydrates tested, and the larger the dextran molecule the less it stabilized
`catalase. Although they did not determine whether their dried samples were
`amorphous, it is well known that as the molecular weight of the carbohydrate is
`increased, the glassy state is formed more readily (44–46). In addition, more
`recent studies have shown (T. Randolph, M. Zhang, S. Prestrelski, T. Arakawa,
`and J. Carpenter, unpublished data) that PFK was not protected, and LDH was
`inactivated further, by dextran during freeze-drying and rehydration. Differ-
`ential scanning calorimetry documented that the dried samples were amor-
`phous. The potential mechanistic bases for these observations will be described
`below. For now, it is important to stress the conclusion that it is necessary for
`stabilizing additives to remain amorphous to protect proteins during lyophili-
`zation, but glass formation alone appears not to be sufficient for stabilization of
`proteins against the severe stress of dehydration.
`There are several studies supporting the other mechanism, which is often
`referred to as the water replacement hypothesis. According to this hypothesis,
`sugars protect labile proteins during drying by hydrogen bonding to polar and
`charged groups as water is removed, thus preventing drying-induced denatura-
`tion of the protein. For example, in early studies, using solid-state Fourier trans-
`1 in the
`form infrared (FTIR) spectroscopy, it was found that the band at 1583 cm
`spectrum for lysozyme, which is due to hydrogen bonding of water to carboxylate
`groups, is not present in the spectrum for the dried protein (33). When lysozyme
`is dried in the presence of trehalose or lactose, the carboxylate band is retained in
`the dried sample, indicating that the sugar is hydrogen-bonding in the place of
`water. Similar results have been obtained with a-lactalbumin and sucrose (8).
`More recently, it has been documented that the carboxylate band can be titrated
`back by freeze-drying lysozyme in the presence of increasing concentrations of
`either trehalose or sucrose (S. Allison and J. Carpenter, unpublished observations).
`This effect correlates directly with an increased inhibition of protein unfolding in
`the presence of increasing amounts of sugar.
`Three other recent studies on enzyme preservation provide further sup-
`port for the water replacement mechanism. Tanaka et al. (43) have found that
`the capacity of a saccharide to protect catalase during freeze-drying is inversely
`related to the size of the saccharide molecule. They suggest that as the size of
`
`Downloaded by [John Sekella] at 17:14 30 June 2015
`
`Ex. 2017-0005
`
`

`
`[vijay][6 X 9 Tight][D:/informa_Publishing/REY_H100116_2400039/z_production/
`z_3B2_3D_files/978-1-4398-2575-4_CH0007_O.3d] [30/4/010/14:17:50] [167–
`197]
`
`172
`
`Carpenter et al.
`
`the saccharide increases, steric hindrance interferes with hydrogen bonding
`between the saccharide and the dried protein. In support of this contention,
`recent experiments have shown that the carboxylate band is only minimally
`detectable in the infrared spectrum of lysozyme freeze-dried in the presence of
`dextran (D. Barberi, T. Randolph, and J. Carpenter, unpublished observation). In
`addition, Tanaka et al. (43) found that the degree of stabilization was based on
`the saccharide sugar mass ratio, which is to be expected if protection is due to
`hydrogen bonding of the saccharide to the protein in the dried solid. More
`recently, by studying protein structure in the dried solid with FTIR spectros-
`copy, Prestrelski et al. (12) found that as the molecular weight of a carbohydrate
`additive was increased the capacity to inhibit unfolding of interleukin-2 during
`lyophilization decreased and the level of protein aggregation after rehydration
`increased. Also, it was clear that protection of the protein did not correlate
`directly with the formation of a glass (all samples were found to be amorphous)
`or with the glass transition temperature of the sample (the Tg increased as
`carbohydrate molecular weight increased). Rather, there was a negative corre-
`lation between stabilization and molecular weight, which is to be expected if
`protection during drying is due to the water replacement mechanism.
`Some of the most compelling evidence for the water replacement hypoth-
`esis comes from studies on the effects of freeze-drying on a model poly-peptide,
`poly-L-lysine (8). This peptide assumes different conformations in solution,
`which have been well characterized with FTIR spectroscopy, depending on the
`pH and temperature. At neutral pH, poly-L-lysine exists as an unordered pep-
`tide. At pH 11.2, the peptide adopts an a-helical conformation. Poly-L-lysine
`assumes an intermolecular b-sheet conformation (11) in the dried state, regard-
`less of its initial conformation in aqueous solution. The preference for b-sheet in
`the dried state appears to be a compensation for the loss of hydrogen bonding
`interactions with water. The b-sheet allows for the highest degree of hydrogen
`bonding in the dried sample. If poly-L-lysine is freeze-dried in the presence of
`sucrose, the original solution structure is retained in the dried state because
`sucrose hydrogen bonds in place of water, obviating the need to form b-sheet.
`
`INFRARED SPECTROSCOPIC STUDIES OF
`LYOPHILIZATION-INDUCED STRUCTURAL CHANGES
`Until recently, the only way to assess the capacity of an additive to stabilize a
`protein during lyophilization was to measure activity and/or structural
`parameters after rehydration. To confound matters further, it was proposed in
`the protein chemistry literature that dehydration did not alter a protein’s con-
`formation (47). Such a claim was clearly counter to the known contributions of
`water to the formation of the native, folded protein (48,49). Also, it was difficult
`to reconcile the finding that proteins could be irreversibly inactivated and
`aggregated after rehydration with the contention that protein structure was not
`perturbed by dehydration.
`Reconciliation of this apparent dilemma was provided by FTIR spectros-
`copy, which can be used to study protein secondary structure in any state (i.e.,
`aqueous, frozen, dried, or even as an insoluble aggregate). FTIR spectroscopy
`has long been used for quantitation of protein secondary structure and for
`studies of stress-induced alterations in protein conformation (50–52). Structural
`information is obtained by analysis of the conformationally sensitive amide I
`
`Downloaded by [John Sekella] at 17:14 30 June 2015
`
`Ex. 2017-0006
`
`

`
`[vijay][6 X 9 Tight][D:/informa_Publishing/REY_H100116_2400039/z_production/
`z_3B2_3D_files/978-1-4398-2575-4_CH0007_O.3d] [30/4/010/14:17:50] [167–
`197]
`
`Perturbations of Protein Structure
`
`173
`
`FIGURE 1 Comparison of infrared spectra of a-chymotrypsin in aqueous solution and dried
`solid state. The inset shows the second derivatives in the amide I region for the spectra in the
`main panel. Source: From Ref. 11.
`
`1. This band is due to the in-
`band, which is located between 1600 and 1700 cm
`plane C=O stretching vibration, weakly coupled with C–N stretching and in-
`plane N–H bending (50,51,53). Each type of secondary structure (i.e., a-helix,
`b-turn, and disordered) gives rise to a different C=O stretching frequency (50–54)
`and, hence, has a characteristic band position, which is designated by wave-
`1. Band positions are used to determine the secondary structural types
`number, cm
`present in a protein. The relative band areas (determined by curve fitting) can then
`be used to quantitate the relative amount of each structural component. Therefore,
`an analysis of the infrared bands in the amide I region can provide quantitative as
`well as qualitative information about protein secondary structure (50–54).
`To obtain this detailed structural information, it is necessary to enhance
`the resolution of the protein amide I band, which usually appears as a single
`broad absorbance contour (Fig. 1). The widths of the overlapping component
`bands are often greater than the separation between the absorbance maxima of
`neighboring bands. Because the band overlapping is beyond instrumental res-
`olution,
`several mathematical band-narrowing methods
`(i.e.,
`resolution
`enhancement methods) have been developed to overcome this problem (11,50–
`52,54). For studies of lyophilization-induced structural transitions, calculation of
`the second-derivative spectrum is recommended (11). This method is com-
`pletely objective and alterations in component bandwidths, which are due to
`protein unfolding, are preserved in the second-derivative spectrum.
`For most unprotected proteins (i.e., lyophilized in the presence of only
`buffer), the second-derivative spectra for the dried solid are greatly altered rela-
`tive to the respective spectra for the native proteins in aqueous solutions (8–14).
`For example, Figure 1 compares the original and second-derivative spectra for
`a-chymotrypsin in solution and in the dried solid. Second-derivative spectra for
`aqueous and dried lactalbumin and LDH, which are also greatly altered by lyo-
`philization, and granulocyte colony-stimulating factor (GCSF), which is mini-
`mally perturbed, are shown in Figure 2. For dozens of proteins studied to date,
`
`Downloaded by [John Sekella] at 17:14 30 June 2015
`
`Ex. 2017-0007
`
`

`
`[vijay][6 X 9 Tight][D:/informa_Publishing/REY_H100116_2400039/z_production/
`z_3B2_3D_files/978-1-4398-2575-4_CH0007_O.3d] [30/4/010/14:17:50] [167–
`197]
`
`174
`
`Carpenter et al.
`
`FIGURE 2 Second-derivative amide I spectra of granulocyte colony-stimulating factor (GCSF),
`a-lactalbumin, and lactate dehydrogenase (LDH) in aqueous solution (upper spectra) and dried
`solid (lower spectra) states. Source: From Ref. 11, employing data from Refs. 8 and 9.
`
`lyophilization induces varying degrees of shifts in band positions, loss of bands,
`and broadening of bands.
`The lyophilization-induced spectral alterations in the conformationally
`sensitive amide I region are due to protein unfolding and not simply to the loss
`of water from the protein. The intrinsic effects of water removal on the vibra-
`tional properties of the peptide bond, and hence protein infrared spectra, were
`found to be insignificant by Prestrelski et al. (8). If the direct vibrational effects of
`water removal were responsible for drying-induced spectral changes, then the
`infrared spectra of all proteins should be altered to the same degree in the dried
`solid, which is not the case.
`Two different behaviors of proteins unfolded in the dried solid are dis-
`played during rehydration: (i) The protein regains the native conformation upon
`rehydration (reversible unfolding), as observed for a-lactalbumin, lysozyme,
`chymotrypsinogen, ribonuclease, b-lactoglobulins A and B, a-chymotrypsin,
`and subtilisin (8,10,11,13,14,55,56). (ii) A significant fraction of the protein
`molecules aggregate upon rehydration (irreversible unfolding), as noted for
`LDH, PFK, interferon-g, basic fibroblast growth factor, and interleukin-2 (8–12).
`It has been documented with several proteins in the latter class that prevention
`of aggregation and recovery of activity after rehydration correlate directly with
`retention of the native structure in the dried solid (8–12). Thus, the mechanism
`by which stabilizing additives (e.g., sugars) minimize loss of activity and
`aggregation during lyophilization and rehydration is to prevent unfolding
`during freezing and drying (8–12).
`For example, the spectrum for interferon-g dried in the presence of 1 M
`sucrose is similar to that for the native aqueous protein, whereas that for the
`protein dried alone is greatly altered (Fig. 3). For analysis of these data, a
`
`Downloaded by [John Sekella] at 17:14 30 June 2015
`
`Ex. 2017-0008
`
`

`
`[vijay][6 X 9 Tight][D:/informa_Publishing/REY_H100116_2400039/z_production/
`z_3B2_3D_files/978-1-4398-2575-4_CH0007_O.3d] [30/4/010/14:17:50] [167–
`197]
`
`Perturbations of Protein Structure
`
`175
`
`FIGURE 3 Comparisons of second-derivative spectra of interferon-g in the dried solid and
`rehydrated states, with or without 1 M sucrose, with the spectrum of the native aqueous state.
`The spectrum of the native aqueous state is shown with the dashed line. The arrows indicate the
`band arising from nonnative intermolecular b-sheet. Source: From Ref. 11.
`
`baseline was fitted to the second-derivative spectra and have been normalized
`for total area (11,57). This data presentation is useful because it allows visual-
`ization of the relative shifts of area from one component band to another, and,
`hence, the redistribution from native to nonnative secondary structural ele-
`ments. For example, for the sample dried without sugar, there is a loss of a-helix
`1 band, which is
`as indicated by the decreased absorbance in the 1656 cm
`compensated by increased absorbance in bands for b-sheet and turns (*1640–
`1). These changes are attenuated when the protein is
`1645 and 1665–1695 cm
`lyophilized in the presence of sucrose, documenting an increased retention of
`native structure in the molecular population.
`After rehydration, the spectra of both samples are very native-like, indi-
`cating that the majority of nonnative molecules have refolded (Fig. 3). However,
`in the spectrum of the sample lyophilized without sucrose, the appearance of a
`1, which is assignable to intermolecular b-sheet structure,
`new band near 1625 cm
`1)
`and the decreased intensities in vibrational bands ascribed to a-helix (1656 cm
`1) structures, indicate the formation of protein aggregates
`and turn (1688–1665 cm
`upon rehydration (see Ref. 11 for a detailed review of the study of protein
`aggregation with infrared spectroscopy). In this sample, 18% of the protein formed
`insoluble aggregates. In contrast, in the sample lyophilized with sucrose, only 9%
`insoluble aggregate was noted after rehydration. This reduction in aggregation is
`1 band in the spectrum of the rehydrated
`reflected in a much weaker 1625 cm
`sample. In this case, 1 M sucrose does not provide complete protection during
`freeze-drying, presumably because it is inadequate at preserving the protein
`structure during the freezing step (see later).
`
`Downloaded by [John Sekella] at 17:14 30 June 2015
`
`Ex. 2017-0009
`
`

`
`[vijay][6 X 9 Tight][D:/informa_Publishing/REY_H100116_2400039/z_production/
`z_3B2_3D_files/978-1-4398-2575-4_CH0007_O.3d] [30/4/010/14:17:50] [167–
`197]
`
`176
`
`Carpenter et al.
`
`Also, unfolding of proteins that refold if immediately rehydrated can be
`inhibited by stabilizing additives (8,10,12–14). It appears crucial that even these
`proteins should be stabilized against
`lyophilization-induced unfolding to
`maintain stability during long-term storage in the dried solid (12,13,15). Thus, an
`important criterion for a successful freeze-dried formulation of any protein is
`retention of the native protein structure in the dried solid, which can be readily
`documented with FTIR spectroscopy.
`Although a qualitative visual comparison of second-derivative spectra can
`be useful to assess the influence of additives on protein structure during lyo-
`philization, a quantitative comparison is often also desirable. For research on
`lyophilization-induced structural transitions, two approaches can be employed.
`Occasionally, there is a need to know the secondary structural content. The
`relative band areas can then be determined with curve fitting (11,50–52,54). For
`example, the percentage of intermolecular b-sheet can be used to calculate the
`percentage of aggregated protein in dried samples (11,14).
`However, for the general assessment of protein stabilization needed to
`evaluate formulations, it is usually more meaningful to make an overall global
`comparison between two spectra. For this analysis, Prestrelski et al. (8,9) origi-
`nally developed a mathematical procedure to calculate the spectral correlation
`coefficient (similarity) between two second-derivative spectra. More recent
`analysis indicated that this method can provide misleading information (57). If
`the spectra have offset baselines, then the correlation coefficient is much lower
`than that expected based on a visual assessment of spectral similarity (Fig. 4,
`top). In contrast, if the