J. F. Carpenter et al. (eds.), Rational Design of Stable Protein Formulations
`© Kluwer Academic/Plenum Publishers, New York 2002
`
`Apotex Exhibit 1027
`Page 1 of 17
`
`

`

`160
`
`Theodore W. Randolph and LaToya S. Jones
`
`The mechanisms of degradation of protein structure and activity are often
`categorized in two broad classes, chemical and physical. Chemical degradation
`refers to those modifications involving covalent bonds, such as deamidation,
`oxidation and disulfide bond shuffling. Physical degradation includes unfolding
`of the protein, undesired adsorption of the protein to surfaces, and aggregation.
`The two categories are not completely independent of one another. For example,
`protein oxidation may result in a greater proclivity to aggregate, and the rate of
`non-native disulfide bond formation may be higher in aggregated proteins.
`Surface-active agents, or surfactants, are often added to protein solutions to
`prevent physical damage during purification, filtration, transportation, freeze-
`drying, spray-drying and storage. Surfactants are amphiphilic, containing a polar
`head group and a non-polartail. This dual nature causes surfactants to adapt spe-
`cific orientations at interfaces and in aqueoussolutions. It is this characteristic
`that lies at the root of the mechanisms by which surfactants affect the physical
`stability of proteins.
`A well-known exampleis the anionic surfactant sodium dodecyl sulfate, or
`SDS. The sulfate anion is the hydrophilic head group of SDS, while the long
`aliphatic dodecyl chain formsthe tail group. Ionic surfactants such as SDS have
`been knownsincethe late 1930’s as effective protein denaturants (Anson, 1939),
`and are commonly used for this purpose, e.g., as a pre-treatment for proteins in
`polyacrylamide gel electrophoresis (SDS-PAGE). In contrast, surfactants used as
`stabilizing agents in protein formulations are typically non-ionic (Loughheed et
`al., 1983; Twardowskiet al., 1983; Chawla et al., 1985). This chapter will focus
`on non-ionic surfactants; protein interactions with ionic surfactants have been
`reviewed elsewhere (Jones, 1996). An example non-ionic surfactant is poly-
`oxyethylene sorbitan monolaurate (Tween 20°), shown in Figure 1. In this
`molecule, the hydrophilic polyoxyethylene units form the head group, while the
`hydrophobic monolaurate group is the tail. Tween 20 is often added to formula-
`tions due to its ability to protect proteins from surface-induced denaturation
`(Changetal., 1996; Jones et al., 1997; Bam et al., 1998; Kreilgaard et al., 1998;
`Maaetal., 1998).
`There are a number of mechanisms by which surfactants can prevent or
`promote damage to proteins. Some of these mechanisms are generic to all
`excipients, and can be explained in the solution thermodynamic framework ofthe
`Wyman linkage theory (Wyman and Gill, 1990) and the preferential exclusion
`mechanisms developed by Timasheff and colleagues (Arakawa and Timasheff,
`1982, 1983, 1984a,b, 1985a,b,c; Arakawaet al., 1990; Timasheff, 1998). Others
`derive from the amphiphilicity of surfactants and the resulting effect of micro-
`scopic ordering of surfactant molecules at interfaces, which in turn affects the
`kinetics and thermodynamics of protein interfaces. In this chapter, we discuss a
`numberof these mechanismsandtheir implications for proteinstability.
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 2 of 17
`Page 2 of 17
`
`

`

`Surfactant-Protein Interactions
`
`161
`
`O(CH2CH,)x-H
`
`Polyethylene glycol ether
`Triton X-100, x=9-10 (average)
`Triton X-114, x=7-8 (average)
`
`2
`
`_-— (CH2CH20),H
`oo(CH2CH20)yH
`
`H(OCH2CH2)w
`
`(CH2CH,0),H
`
`wtxty+z=20
`Polysorbate
`Tween 20, R=C,,H53CO,
`Tween 80, R=C7H33CO,
`
`Figure 1. Example non-ionic surfactants.
`
`PROTEINS AND SURFACTANTSAT SURFACES
`
`Because of their dual hydrophobic/hydrophilic nature, surfactants in solu-
`tion tend to orient themselves so that the exposure of the hydrophobic portion of
`the surfactant to the aqueous solution is minimized. Thus, in systems containing
`air/water interfaces, surfactants will
`tend to accumulate at
`these interfaces,
`forming a surface layer of surfactant oriented in such a fashion that only their
`hydrophilic ends are exposed to water. Such orientation and surface adsorption
`can also occurat solid/water interfaces such as those foundin vials, syringes, and
`other containers. Protein molecules also exhibit surface activity (for a review see
`(Magdassi, 1996), and references therein) and as such will also tend to adsorb to
`and orient at these interfaces.
`From classical thermodynamics, the excess surface internal energy dUj of
`a surface with area A at a temperature T is related to the excess surface entropy
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 3 of 17
`Page 3 of 17
`
`

`

`162
`
`Theodore W. Randolph and LaToyaS. Jones
`
`S? and the chemical potential and number of surface excess moles of each
`adsorbed species:
`
`dU? = TdS? + 0dA+ DY dny,
`i=2
`
`(1)
`
`Here the subscript 1 refers to a dividing surface chosen so that there is no excess
`adsorption of species 1, the solvent (water) (Gibbs, 1961), and o is the surface
`tension. The equilibrium criterion, 5S = 0, requires that 6 be constant across the
`surface. Thus, if a protein adsorbs to an interface, at equilibrium the surface
`tension forces must be continuousand constant across the wholeinterface, includ-
`ing acrossthe protein. Thestability criterion at equilibrium requires that:
`
`00(3). >0
`
`(2)
`
`If the surface tension of the interface is greater than the internal tension in the
`protein, then in order to meet these two conditions, the surface area ofthe protein
`must increase until the two tensions are equal, i.e., the protein must unfold. In
`some cases, nearly complete loss of native activity is lost upon adsorbing to
`the interface (Rothen, 1947; Verger et al., 1973). The Gibbs adsorption equation
`relates the surface tension to the concentration of adsorbed species at an
`interface:
`
`-do = sfdT + YTisdy,
`i=2
`
`(3)
`
`where [;, and s?dT are, respectively, the excess surface adsorption and excess
`surface entropy of component/, bothrelative to a dividing surface with no surface
`excess of solvent (1), and 1; is the chemical potential of species i. Adsorption of
`protein to the interface thus lowers the interfacial tension, making unfolding less
`likely as adsorption progresses.
`If the process of surface adsorption and unfolding of protein were to stop
`after the formation of an equilibrium monolayer, the amount of adsorbed protein
`would be so small as to be generally of no consequence. Indeed, a significant
`amountof work has been dedicated to development of methods with sensitivities
`high enoughto characterize the minute amount of protein adsorbedto the inter-
`face (Tupyet al., 1998; Vermeer and Norde, 2000). However, depending on the
`degree of surface hydrophobicity and characteristics of the protein in question,
`additional processes can occurin the adsorbedfilms, leading to behavior that is
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 4 of 17
`Page 4 of 17
`
`

`

`Surfactant-Protein Interactions
`
`163
`
`no longer described by the Gibbs’ adsorption equation. Among the possible
`processes are gas-to-liquid surface phase transitions, surface precipitation, and
`the formation of surface sublayers. In many cases, the adsorbed molecules can
`be rapidly released from the surface and surface sublayers, and may exchange
`with bulk protein molecules. Alternatively, because of slow refolding kinetics,
`proteins can become irreversibly adsorbed at
`the interface, or have a rate
`of exchange somewhere between these extremes (Dickinson, 1999; Norde and
`Giacomelli, 1999). These processes of adsorption and release of structurally-
`perturbed protein molecules into the bulk solution have been implicated as one
`of the causes of protein aggregation and denaturation.
`When discussing protein adsorption at interfaces, globular proteins are
`typically characterized as being either “hard” or “soft,” having a low or high
`degree of flexibility, respectively, and by their degrees of hydrophobicity. Soft,
`hydrophobic proteins, attaining monolayer coverage of the air/water interface in
`the matter of minutes, are generally more surface reactive at hydrophobic sur-
`faces than hard, hydrophilic proteins, attaining coverage of the same surfaces in
`a matter of hours (Tripp et al., 1995). The driving force for protein adsorption is
`the decrease in the entropy of the water molecules that are ordered around the
`hydrophobic protein domains whenthe protein is in the bulk solution. Thus, the
`role of the relative degree of hydrophobicity on protein surface adsorption is
`rather straightforward: given two proteins only differing in their hydrophobici-
`ties, the more hydrophobic protein will have a greater number of productive
`interactions with the surface and will form the monolayer more quickly.
`Middelberg et al. (2000) proposed that the difference in adsorption kinetics of
`Lac21 and Lac28 peptides is because the monomeric Lac21 has more hydropho-
`bic residues exposed than its tetratmeric counterpart, Lac28; thus, Lac21 more
`readily forms a monolayer at an octane-water interface. Protein flexibility
`is important in protein spreading that occurs at the interface. A flexible protein
`can expose additional non-polar residues, leading to an increased strength in
`binding to the surface. Finally, the protein flexibility dictates the number of
`proteins that can adsorb at the interface, and their spreading rate (Norde and
`Giacomelli, 1999).
`Protein adsorption to a hydrophobic surface does not necessarily lead to a
`complete loss of “native” structure: some proteins actually gain structure. For
`example, melittin, a honeybee venom peptide, increases its o-helical content
`slightly when adsorbed to hydrophobic quartz. In contrast, adsorption of a
`tetramer of the same peptide is thought to require a loosening of the o-helical
`content. The orientation of the adsorbed helices in the peptide is parallel to the
`quartz plane, with the hydrophobic moieties facing the plane (Smith and Clark,
`1992). Caessens et al. (1999) also report an increase in helical content on the
`adsorption of the predominately random coil B-casein and B-casein peptides at
`the teflon/waterinterface.
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 5 of 17
`Page 5 of 17
`
`

`

`164
`
`Theodore W. Randolph and LaToyaS. Jones
`
`Surfactants will also adsorb to the interface. The Gibbs adsorption isotherm
`again predicts that this interaction will lower the surface tension. Thus, if sur-
`factants co-adsorb to an interface together with proteins, there will be a smaller
`driving force for surface adsorption. Furthermore, competition between the
`protein and the surfactant for the interface may reduce equilibrium protein
`adsorption. For example, Tween 20 addition displaces beta-lactoglobulin films
`from air-water interfaces (Roth et al., 2000). This effect is not universal: the ionic
`surfactant sodium dodecyl sulfate forms a complex with high surface activity that
`exhibits enhanced surface adsorption (Green et al., 2000). Finally, because the
`surface tension is lowered after surfactant is adsorbed, less damageto the protein
`that does adsorbed may occur. For example, a loss of o-helix and an increase in
`B-turn structures occur when bovine serum albumin adsorbs to polystyrene par-
`ticles. This effect is decreased when the surface is more crowded (Norde and
`Giacomelli, 2000).
`Protein stabilization by nonionic surfactants can often be observed by for-
`mulating with micromolar concentrations of surfactant. This is due to the high
`surface-activity of this class of excipients, which renders a higher effective con-
`centration of surfactant molecules at interfaces than in the bulk solution. When
`the concentration of the surfactant is much lower than its critical micelle con-
`centration (CMC), surfactant moleculeslie flat at the air/water and hydrophobic
`solid/water interfaces (Figure 2A)
`(Porter, 1994). As the concentration is
`increased, more molecules adsorb to these interfaces, such that the surface con-
`centration remains linearly proportional
`to the bulk concentration (Tanford,
`1973). This crowding forces the surfactant molecules to order themselves such
`that the hydrophilic groups are oriented towards the bulk water and the hydro-
`carbon chains are pointed towards the air or hydrophobic solid (Porter, 1994;
`Fainerman et al., 2000) (Figure 2B). At sufficiently high surfactant concentra-
`tions (i.e., at or above the CMC), there is an oriented monolayer of surfactant
`molecules and maximum surfactant absorption, at the interface (Figures 2C &
`D). The surface saturation is responsible for the sharp slope change (to essen-
`tially zero) observed in experimental plots of surface properties (surface tension,
`osmotic pressure) versus surfactant concentration (Porter, 1994) (Figure 3). It
`should be notedthat the linear variations in the surface properties shown in Figure
`3 are indicative that the surface activity of a surfactant is due to the hydropho-
`bic effect: the variation would be cooperative if hydrocarbon self-affinity was the
`appropriate explanation (Tanford, 1973). Surfactant micelles are formed in the
`bulk phase whenthe concentration is of the surfactant is above the CMC (Figure
`2D). Thus, a range of surfactant concentrations could inhibit protein denaturation
`at an interface; however, the necessity of CMClevels of surfactant to completely
`inhibit protein damage would be a strong indication that the damage is caused by
`adsorption of protein at the interface and that inhibiting protein adsorption at the
`interface plays a role in preventing protein aggregation.
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 6 of 17
`Page 6 of 17
`
`

`

`Surfactant-Protein Interactions
`
`165
`
`A
`
`B
`
`C
`\
`YN 2Xad
`D
`— deaneeia| SSS,|HART|TY
`AION
`
`a
`
`&
`
`SP ay ah’
`
`Figure 2. Simplified models of the interfacial behavior of a nonionic surfactant at several concen-
`trations in water. Circles—polar head groups(e.g., polyoxyethelenes). Rectangles—hydrophobictails
`(e.g., hydrocarbon chains). Models abovethe dotted line is for the hydrophobicsolid (ice)/waterinter-
`face and those aboveare for the air/water interface. (A) Surfactant concentration is well below the
`CMC(B) Surfactant concentration is greater than in A, but still below the CMC.(C) Surfactant con-
`centration is at the CMC.(D) The surfactant concentration is above the CMC. (This model is adapted
`from Figures 4.4 and 4.8 of Porter (Porter, 1994)).
`
`TOnnantpahian
`S
`concentration
`oO
`
`Surface tension
`Osmotic pressure
`
`Figure 3. Representation of changes two properties in determining the CMCofa surfactant.
`
`Decreased protein adsorption at interfaces (e.g., air/liquid, ice/liquid) in the
`presence of nonionic surfactants such as Tween 20 can beattributed to the surface
`activity of nonionic surfactants (Changet al., 1996; Kreilgaard etal., 1998; Miller
`et al., 2000a,b) and, in somecases,direct interactions between the surfactant and
`protein molecules (Dickinson, 1998; Bam et al., 1998; Miller et al., 2000a,b). In
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 7 of 17
`Page 7 of 17
`
`

`

`166
`
`Theodore W. Randolph and LaToyaS. Jones
`
`mixed protein/surfactant systems in which the surfactant binds to hydrophobic
`regionsofthe protein, the protein is less surface-active than it would be in a solu-
`tion devoid of the nonionic surfactant. This explains the increase in surface
`tension relative to that of the pure protein solution that can be observed at
`extremely low surfactant concentrations. Adsorption of surfactant and protein
`molecules in the mixed system is competitive. Nonionic surfactants usually bind
`tighter than proteins or protein—-surfactant complexes at interfaces (Dickinson,
`1998). Thus, above a critical concentration of the surfactant, protein adsorption
`becomes negligible and the adsorption isotherms for mixed surfactant/protein
`systems can be roughly identical that of a pure surfactant solution, as observed
`by Miller et al. (2000a) for the HSA/C,)DMPO when the C;,DMPOconcentra-
`tion exceeds 107 mol/cm’.
`
`PROTEIN-SURFACTANT INTERACTIONS IN SOLUTION
`
`In addition to altering the interaction of proteins with surfaces, non-ionic
`surfactants can also interact directly with proteins in solution. For example,
`Tween 20 acts as a chemical chaperone,aiding in the refolding of proteins (Bam
`et al., 1998; Kreilgaard et al., 1999). In vivo, proteins fold while at average con-
`centrations of approximately 35 mg/ml (Hartl, 1996): in vitro, non-native protein
`moleculesat this concentration (e.g., due to freeze concentration) usually aggre-
`gate. Inside cells, protein folding is aided by naturally occurring molecular chap-
`erones. Unlike folding catalysts that have steric information to guide the protein
`folding, molecular chaperones act by non-covalently binding to partially-folded
`proteins, usually via hydrophobic interactions, to prevent misfolding or aggrega-
`tion while the protein is attempting to adopt its native conformation (Gatenby
`and Ellis, 1990; Hartl, 1996). Chaperone-assisted protein refolding can prevent
`the protein from falling into kinetic traps or simply allow more time for the
`protein to refold (Gatenby and Ellis, 1990). The hydrophobic effect, the driving
`force of protein folding and surface activity of Tween 20, is also implicated in
`the interactions between the exposed hydrophobic regions of the partially folded
`protein and the hydrocarbontail of the surfactant. These mixed surfactant/protein
`complexes and protein folding are dynamic processes, eventually the protein
`attains a conformation in which the hydrophobic groups are not as surface
`exposed. Unless there are hydrophobic patches in the native protein conforma-
`tion, the surfactant molecules will not necessarily bind to the protein, as in the
`case of molecular chaperones. In vitro, surfactants such as Tweens (Bam etal.,
`1996), polyethylene glycol (Cleland and Wang, 1990; Cleland et al., 1992;
`Cleland and Randolph, 1992; Cleland 1993; Cleland and Wang, 1993), Triton
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 8 of 17
`Page 8 of 17
`
`

`

`Surfactant-Protein Interactions
`
`167
`
`X-100 (Donate et al., 1998), and lubrol (Donate et al., 1998), have been impli-
`cated in aiding protein refolding by acting as chemical chaperones.
`Surfactants, such as Tween 20, can also affect the thermodynamic confor-
`mational stability of a protein. As discussed in other chapters in this book, and
`in detail in references from Timasheff and co-workers (Lee and Timasheff, 1981;
`Arakawa and Timasheff, 1982; Arakawa and Timasheff, 1985; Timasheff, 1998),
`thermodynamicstability is increased if a ligand exhibits greater binding to the
`native state of a protein than to a non-native state. However, with many excipi-
`ents (e.g., sucrose) excluded volume effects produce a non-specific negative
`binding to the native state, and a concomitantly larger negative binding to
`expanded, non-native conformations. This differential negative binding also
`results in a stabilization of the native state. At the low concentrations of surfac-
`tant (ca. 100 micromolar) typically used in formulations of therapeutic proteins,
`thermodynamic effects due to excluded volume can usually be neglected. More
`important is specific binding of surfactants to either the native or unfolded states
`of a protein. Randolph and colleagues report that some proteins and nonionic sur-
`factants, including Tween 20, form mixedprotein: detergent complexes (Bam et
`al., 1995, 1996, 1998; Jones, et al., 1999). In the presence of Tween 40 at a 4:1
`surfactant: protein molar ratio, native recombinant human growth hormoneis sig-
`nificantly stabilized: the denaturation midpoint in guanidine hydrochloride solu-
`tions increases from 4.6M guanidine hydrochloride in the absence of Tween 40
`to 5.9M in the presence of Tween 40 (Bam etal., 1996). Likewise, 4:1 Tween
`40 increases the AG of unfolding of recombinant human growth hormone by
`4.1 kcal/mol (Bam et al., 1996), and 10:1 Tween 40 increases its melting point
`slightly from 88.8 to 89.4° C (Bam etal., 1998). Stabilization of recombinant
`human growth hormoneby surfactants results in reduced aggregation in agitated
`solutions (Bam et al., 1998). With bovine serum albumin in the presence of sur-
`factant there. are decreased amounts of thermally-induced protein aggregates,
`relative to surfactant-free controls (Arakawaand Kita, 2000). Conformationalsta-
`bilization of proteins by non-ionic surfactants is not universal; stability of IgG is
`unaffected by low concentrations of Tween 20 (Vermeer and Norde, 2000), and
`recombinant human interferon-y shows lower free energies of unfolding in the
`presence of Tween 20 (Webbetal., 2000).
`
`SURFACTANT EFFECTS ON PROTEIN ASSEMBLY STATE
`
`The hydrophobic portion of non-ionic surfactants can bind to hydrophobic
`patches on proteins. This naturally causes the surfactant to order itself so that
`more hydrophilic groups are solvent exposed, resulting in a “hydrophobicity
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 9 of 17
`Page 9 of 17
`
`

`

`168
`
`Theodore W. Randolph and LaToya S. Jones
`
`the protein-surfactant
`reversal”. This “hydrophobicity reversal” means that
`complex is more hydrophilic that either the surfactant or protein alone, and effec-
`tively increases the solubility of the complex. This, in turn can reduce the pro-
`pensity of the protein to form higher-order aggregates. For example, bovine
`mitochondrial cytochrome bc1 is dimeric in solutions at low ionic strength and
`low surfactant levels. At Tween 20 concentrations above 5mg/mgprotein, a
`homogeneous, monomeric,reversible and enzymatically active protein-surfactant
`complex is formed (Musatov and Robinson, 1994). Likewise, in freeze-thaw
`studies of recombinant factor XIII, addition of Tween 20 at concentrations near
`the CMCblockedthe progression of aggregates from a relatively low molecular
`weight, soluble fraction to insoluble aggregates. Figure 4 shows levels of
`
`=nOaoOo
`
`
`
`aggregates
`
`
`nho
`
`%nativeprotein aOo
`
`
`%solubleaggregates
`%insoluble
`
`= oOo
`
`o
`
`0
`
`50
`
`100
`
`150
`
`200
`
`250
`
`Tween 20 concentration [iM]
`
`Figure 4. Recovery of native rFXIII (A) and formation of soluble (B) and insoluble aggregates (C)
`following 10 freeze-thaw cycles of 1 mg/ml (—e®—, 5 mg/ml (.---O----) and 10 mg/ml rFXIII (—¥
`as a function of Tween 20. Results are plotted as mean values +/— standard deviation for duplicate
`samples (reproduced from Kreilgaard et al., 1998).
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 10 of 17
`Page 10 of 17
`
`

`

`Surfactant-Protein Interactions
`
`169
`
`aggregation for factor XIII after freeze thaw cycling in the presence and absence
`of Tween 20. Note that Tween addition did not completely block aggregation,
`but was very effective at preventing the formation of insoluble aggregates
`(Kreilgaard et al., 1998).
`Non-ionic surfactants can also have the opposite effect on protein assembly
`state. In cases where a non-ionic surfactant destabilizes the conformation of
`
`a protein, this effect may compete against the solubilizing effect of surfactant
`binding and hydrophobicity reversal. For example, Bax, a monomeric protein that
`regulates apoptosis, readily forms dimers in the presence of Tween 20. However,
`these dimers are apparently non-native, as they do not expose the characteristic
`N-terminal Bax epitope (Hsu and Youle, 1998). In the case of the hydrophobic
`lipase from Humicola lanuginose, Tween 20 addition caused the formation of
`large, insoluble non-native aggregates (Kreilgaard et al., 1999).
`
`SURFACTANT EFFECTS ON PROTEINS DURING FREEZING,
`FREEZE-DRYING AND RECONSTITUTION
`
`The processes of freezing, drying, and reconstitution of protein solutions
`present a numberof stresses that may denature proteins. Many ofthese stresses
`are associated with surfaces: new ice-water and ice-glassy solid interfaces are
`formed during freezing, drying replaces ice-glass interfaces with air-glass inter-
`faces, and reconstitution exposes the glassy solid surfaces to aqueous solution.
`In each of these steps protein adsorption to surfaces is potentially damaging.
`The ice-water interface has been implicated as a source of damage to proteins
`(Strambini and Gabellieri, 1996), as has the solid-air interface (Hsuetal., 1995).
`Addition of nonionic surfactants can reduce this damage, presumably by com-
`peting with the protein for the ice-water interface (Chang, 1996). For example,
`addition of Tween 80 to solutions of recombinant hemoglobin reduced aggrega-
`tion seen during freeze thaw studies (Kerwin etal., 1998). Interestingly, Tween
`80 did not offer protection against methemoglobin formation or hemoglobin
`aggregation during long-term frozen storage.
`Non-ionic surfactants also have been shownto affect the recovery of native
`protein from lyophilized formulations. Sarciaux et al. (1999) showed that the
`addition of Tween 80 to the formulation solution or the reconstitution medium
`for lyophilized formulations resulted in reduced levels of aggregates. Likewise,
`Zhanget al. (1995, 1996) have demonstrated that, following long-term storage,
`surfactants in the reconstitution medium can affect protein recovery. We have
`recently shown (Webbetal., 2000) that addition of Tween 20 to the reconstitu-
`tion medium for lyophilized preparations of recombinant human interferon-y
`results in decreased levels of aggregates. The mechanism for such reduced
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 11 of 17
`Page 11 of 17
`
`

`

`170
`
`Theodore W. Randolph and LaToya S. Jones
`
`aggregation was shownto be a surfactant effect on dissolution rates. Addition of
`Tween 20 slowedthe dissolution of the lyophilized solid, allowing protein that
`partially unfolded during freeze-drying to refold before aggregating. Interest-
`ingly, Tween 20 in aqueoussolutions destabilizes recombinant human interferon-
`Y against urea-induced unfolding, impedes refolding during rapid dilution from
`urea solutions, and actually increases aggregation during agitation.
`
`ENZYMATIC DEGRADATION OF NON-IONIC SURFACTANTS
`
`Although most non-ionic surfactants are thought of as chemically inert com-
`ponents of a formulation, specific chemical interactions between proteins can
`occur. For example, some enzymes show hydrolytic activity toward Tweens.
`Smegmatocin, an esterase from Mycobacterium smegmatis, shows a broad
`thermal and pHstability in its activity against Tween 80 (Tomioka, 1983). The
`byproduct of Tween 80 hydrolysis, oleic acid, is toxic to some bacteria. Similar
`bacteriocins, which require Tween their expression, have been ascribed to other
`mycobacteria (Saito et al., 1983). It is not clear how widespread esterase activ-
`ity is against Tweens. However,it is clear that caution should be used whenfor-
`mulating proteins with esterase activities in Tween solutions.
`
`RECOMMENDATIONS FOR PROTEIN FORMULATION
`
`Clearly, it is desirable to minimize the addition of any excipient to a for-
`mulation. This rule of thumb is even more pertinent for surfactants, because there
`is ample evidence that high concentrations of surfactants can be destabilizing to
`protein structure. On the other hand, small amounts of surfactant often provide
`benefits in preventing aggregation that greatly outweigh any conformationally
`destabilizing effect. How then should surfactant levels be chosen for optimal for-
`mulation? The answer appears to depend on the mechanism(s) by which a par-
`ticular protein is protected from damage by surfactant addition. In cases where
`surfactants act to stabilize the native state of a protein by bindingto the protein,
`a specific surfactant: protein stoichiometry may need to be maintained in order
`to provide optimal protection. In these cases, changes in the protein concentra-
`tion within a formulation will dictate proportional changes in surfactant concen-
`tration to maintain a fixed molar ratio. This appears to be the case for recombinant
`human growth hormone, where protection against agitation-induced damagecor-
`related with the molar ratio of surfactant to protein rather than to the surfactant’s
`CMC(Bam etal., 1998). In the case of specific binding, the choice of nonionic
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 12 of 17
`Page 12 of 17
`
`

`

`Surfactant-Protein Interactions
`
`171
`
`surfactant may be important. In the case of recombinant human growth hormone,
`for example, different binding stoichiometries and degrees of protein stabiliza-
`tion were seen for a variety of common surfactants (Bametal., 1995). A general
`recommendation for proteins that show specific binding to the native state of the
`protein is to formulate so that the ratio of surfactant to protein is slightly above
`the binding stoichiometry for a particular surfactant. The choice of surfactant may
`be dictated by the degree of stabilization (which should correlate with the degree
`of binding) provided to a protein by a particular surfactant.
`In contrast, if no specific binding is seen, then maximum levels of protec-
`tion generally correlate with the CMC of the surfactant. In this case, surfactant
`should be addedat levels slightly above the CMC. The choice of surfactant is
`often dictated by a trade-off: surfactants with lower CMC’s will require less sur-
`factantin solution to saturate surfaces and reduce surface-induced damageto pro-
`teins. However, surfactants with low CMC’s are much moredifficult to remove
`from solution (e.g., by dialysis) if necessary, and also tend to be less soluble than
`surfactants with higher CMC’s, raising the possibility of undesirable phase
`separation during processes such as freezing or lyophilization.
`
`_
`
`REFERENCES
`
`Adler, M., and Lee, G., 1999. Stability and surface activity of lactate dehydrogenase in
`spray-dried trehalose. Journal of Pharmaceutical Sciences 88:199.
`Anson,M., 1939. The denaturation ofproteins by detergents and bile salts. Science 90:256.
`Arakawa,T., Bhat, R. et al., 1990. Preferential interactions determine protein solubility in
`three-componentsolutions: the MgCl, system. Biochemistry 29:1914.
`Arakawa, T., Bhat, R. et al., 1990. Why preferential hydration does not always stabilize
`the native structure of globular proteins. Biochemistry 29:1924.
`Arakawa,T., and Kita, Y., 2000. Protection of bovine serum albumin from aggregation by
`Tween 80. J. Pharm. Sci. 89:646.
`
`Arakawa,T., and Timasheff. S.N., 1982. Preferential interactions of proteins with salts in
`concentrated solutions. Biochemistry 21:6545.
`Arakawa, T., and Timasheff, S.N., 1982. Stabilization of protein structure by sugars.
`Biochemistry 21:6536.
`Arakawa,T., and Timasheff, S.N., 1983. Preferential interactions of proteins with solvent
`components in aqueous amino acid solutions. Arch. Biochem. Biophys. 224:169.
`Arakawa,T., and Timasheff, S.N., 1984a. Mechanism ofprotein salting in and salting out
`by divalent cation salts: balance between hydration and salt binding. Biochemistry
`23:5912.
`
`Arakawa, T., and Timasheff, S.N., 1984b. Protein stabilization and destabilization by
`guanidinium salts. Biochemistry 23:5924.
`Arakawa,T., and Timasheff, S.N., 1985a. Mechanism of poly(ethylene glycol) interaction
`with proteins. Biochemistry 24:6756.
`
`Apotex Exhibit 1027
`Apotex Exhibit 1027
`Page 13 of 17
`Page 13 of 17
`
`

`

`172
`
`Theodore W. Randolph and LaToyaS. Jones
`
`Arakawa, T., and Timasheff, S.N., 1985b. The stabilization of proteins by osmolytes.
`Biophys. J. 47:A11.
`Arakawa, T., and Timasheff, S.N., 1985c. Theory of protein solubility. Meth. Enzymol.
`114:49.
`
`Bam, N.B., Randolph, T.W., and Cleland, J.L., 1995. Stability of protein formulations:
`Investigation of surfactant effects by a novel EPR spectroscopic technique. Pharm.
`Res, 12:2.
`
`Bam, N.B., Cleland, J.L. et al., 1996. Molten globule intermediate of recombinant human
`growth hormone:stabilization with surfactants. Biotech. Prog. 12:801.
`Bam, N.B., Cleland, J.L.et al., 1998. Tween protects recombinant human growth hormone
`against agitation-induced damage via hydrophobic interactions. J. Pharm. Sci.
`87:1554.
`
`Bam,N.B., Randolph, T.W.et al., 1995. Stability of protein formulations: investigation of
`surfactant effects by a novel EPR spectroscopic technique. Pharm. Res. 12:2.
`Broadhead, J., Rouan, S.K. et al., 1994. The effect of process and formulation variables
`on the properties of spray-dried beta-galactosidase. J. Pharmaceut. Pharmacol.
`46:458.
`
`Caessens, P.W., De Jongh, H.et al., 1999. The adsorption-induced secondary structure
`of B-casein and of distinct parts of its sequence in relation to foam and emulsion
`properties. Biochim. Biophys. Acta 1430:73.
`in:
`Carpenter, J.F., and Chang, B.S., 1996. Lyophilization of protein pharmaceuticals,
`Biotechnology and Biopharmaceutical Manufacturing, Processing, and Preservation,
`K_E. Avis and V.L. Wu, eds., Interpharm Press, Buffalo Grove, IL.
`Carpenter, J.F., Pikal, M.J., Chang, B.S., and Randolph, T.W., 1997. Rational design of
`stable lyophilized protein formulations: some practical