6 Physical Instability of
`Peptides and Proteins
`
`Marco van de Weert and Theodore W. Randolph
`
`CONTENTS
`6.1
`Introduction .................................................................................................. 107
`6.2 Protein Structure ........................................................................................... 108
`6.2.1 Peptides, Polypeptides, and Proteins ................................................ 108
`6.2.2 Protein Structure: Primary, Secondary, Tertiary, and
`Quaternary Structure ........................................................................ 108
`6.3 Protein Folding: Why Do Proteins Fold? ..................................................... 109
`6.3.1 Role of Water and Stabilizing Interactions ....................................... 109
`6.3.2 The Energy Landscape of a Protein Molecule ................................. 111
`6.4 Protein Physical Degradation ....................................................................... 114
`6.4.1 Protein Unfolding ............................................................................. 114
`6.4.2 Adsorption ........................................................................................ 117
`6.4.3 Protein Aggregation .......................................................................... 118
`6.4.3.1 Aggregation Mechanisms and Kinetics ............................. 119
`6.4.3.2 Fibrillation: A Special Case of Protein Aggregation ......... 120
`6.4.4 Protein Precipitation ......................................................................... 121
`6.5 Stabilization Strategies ................................................................................. 122
`6.6 Concluding Remarks .................................................................................... 125
`References .............................................................................................................. 126
`
`INTRODUCTION
`6.1
`The biological function of peptides and proteins is highly dependent on their three-
`dimensional structure. Changes in that structure, which may arise due to chemical
`or physical processes, may alter or abolish that function, or even result in toxicity.
`Thus, it is of importance that a pharmaceutical formulation of therapeutic peptides
`and proteins retains the normal (native) structure of those peptides or proteins, or
`that any changes are fully reversible upon administration to the patient.
`A major difference between proteins and low molecular weight drugs is the com-
`plexity of the three-dimensional structure and concomitant sensitivity toward exter-
`nal stress factors. The three-dimensional structure of proteins is mostly held together
`by noncovalent interactions, such as hydrogen bonds, salt bridges, and van der Waals
`forces. Any stress factor may alter these noncovalent interactions, possibly leading to
`new intra- or intermolecular interactions which may not be reversible upon removing
`the stress factor.
`
`107
`
`Ex. 2004-0001
`
`

`
`108
`
`Pharmaceutical Formulation Development of Peptides and Proteins
`
`In this chapter, we will discuss the noncovalent interactions that result in the for-
`mation of the specific three-dimensional fold of most proteins, the most important
`stress factors that may cause changes in that protein fold, and the resulting physical
`instability of the protein. It should be noted that we will discuss this physical instabil-
`ity as a separate issue to the chemical instability discussed in Chapter 5. In reality, the
`two are highly interdependent, as chemical destabilization may lead to physical desta-
`bilization, and vice versa. This chapter will end with potential stabilization strategies
`to prevent physical instability of proteins in pharmaceutical formulations. Several of
`these strategies will be discussed with more specific examples in other chapters.
`Throughout the chapter, a number of semantic issues will be discussed, which are of
`importance when reading the literature. The commonly used terminology within the
`field of protein structure, folding, and stability is not always strictly defined, and defini-
`tions may differ over time and depending on the context. Unfortunately, the definitions
`used in a particular scientific paper are often not explicit, which may lead to confusion
`when the reader is insufficiently aware of the different descriptions that are in use.
`
`6.2 PROTEIN STRUCTURE
`
`6.2.1 PePtides, PolyPePtides, and Proteins
`All peptides, polypeptides, and proteins are considered condensation polymers of
`amino acids, resulting in a linear backbone of alternating amide, C–C, and C–N
`bonds. However, the distinction between peptide, polypeptide, and protein is rather
`diffuse. One may find at least three different and partly overlapping descriptions,
`rather than definitions, of the difference between peptide and protein alone. The cur-
`rently most common description refers to any peptide chain of more than 50 amino
`acids as a protein. Others refer to peptides as proteins whenever the peptide has
`a biological function. This could then even include several simple dipeptides (i.e.,
`two amino acids linked together), which can have a biological function. Finally, the
`absence or presence of a well-defined tertiary structure has been used to distinguish
`peptides from proteins. Also, this distinction is not without problems; there are pro-
`teins that are referred to as “natively unfolded,” so called because they do not have
`a specific tertiary structure. In addition, some “peptides” can form fully reversible
`multimeric structures, such as glucagon (forming trimers) (Formisano et al., 1977),
`which involves the formation of a defined three-dimensional structure. The term
`“polypeptide” generally overlaps with that of “peptide” and “protein.” The reader
`may thus encounter all three terms used in connection with the same biological
`compound. For practical purposes, we have used the first definition, calling every
` compound with more than 50 amino acids a protein.
`
`6.2.2
`
` Protein structure: Primary, secondary,
`tertiary, and Quaternary structure
`The three-dimensional structure of proteins is often subdivided into four types of
`structure, referred to as the primary, secondary, tertiary, and quaternary structure.
`The primary structure refers to the amino acid sequence within the polymer chain.
`
`Downloaded by [Jeff Bird] at 14:03 27 June 2015
`
`Ex. 2004-0002
`
`

`
`Physical Instability of Peptides and Proteins
`
`109
`
`Intra- or interchain cross-links are also prevalent, usually through cysteine residues
`(forming a cystine or S–S bridges).
`The secondary structure refers to the folding of this backbone into specific struc-
`tures, which are defined by the bond angles and hydrogen bonding pattern of the
`amide bond. The secondary structure can roughly be subdivided into four classes:
`helical structures such as the alpha-helix, pleated structures such as the beta-strand
`and beta-sheet, turn structures such as the beta-turns, and loop structures. The latter
`are often referred to as “random” structures.
`The three-dimensional alignment of the secondary structural elements is known
`as the tertiary structure of a protein. This alignment often results in an almost ideal
`close packing of the amino acids, particularly in the core of the protein molecule.
`Protein tertiary structures can be described by commonly appearing architectures
`such as barrels or alpha-helix bundles (Orengo and Thornton, 2005).
`Some proteins exist under physiological conditions as specific multimeric pro-
`teins linked through noncovalent interactions. This multimerization is known as the
`quaternary structure of a protein. Examples of proteins with a quaternary structure
`include hemoglobin, alpha-crystallin, and HIV-1 protease. In general, the biologi-
`cal function of such multimeric proteins depends on this multimerization, but some
`proteins may also form specific (and reversible) multimers that are not biologically
`active. Perhaps the best known example of the latter is insulin; insulin forms dimers
`and hexamers at elevated concentration, especially in the presence of certain diva-
`lent metal ions, but is only active as a monomer (Uversky et al., 2003).
`The ultimate fold of the protein is usually referred to as the “native” structure.
`In principle, the latter refers to the functional structure of the protein. However,
`many proteins change structure during their biological function, which would sug-
`gest there are multiple “native” structures. Furthermore, artificially created proteins
`(e.g., fusion proteins created by genetic engineering techniques) may assemble into
`well-defined folds but have unknown levels of function. It may therefore be easier to
`describe the protein structure under physiological conditions (in terms of pH, ionic
`strength, etc.) as the native structure. The mechanism of protein folding is discussed
`in the following section.
`
`6.3 PROTEIN FOLDING: WHY DO PROTEINS FOLD?
`
`6.3.1 role of Water and stabilizing interactions
`The observation that most proteins are folded into a specific structure in simple
`aqueous solutions suggests that folding is a thermodynamically favorable process.
`Many decades of research have been aimed at elucidating why, and how, proteins
`fold (Anfinsen, 1973). Although there are still several limitations, it is now possible
`to predict with reasonable accuracy how a protein will fold using computational
`methods (Kryshtafovych and Fidelis, 2009). In this section, we will discuss the driv-
`ing forces for folding, starting with a protein in the gas phase before moving to the
`more complex situation of a protein in solution.
`For a single protein molecule in the gas phase, there are four fundamental forces
`to take into account. The first is the entropy of the amino acid chain, which tends
`
`Downloaded by [Jeff Bird] at 14:03 27 June 2015
`
`Ex. 2004-0003
`
`

`
`110
`
`Pharmaceutical Formulation Development of Peptides and Proteins
`
`to disfavor folding. That is, folding of the amino acid chain into a specific structure
`reduces the degrees of freedom for that chain, which results in a loss of entropy. In
`contrast, hydrogen bonding and van der Waals forces favor folding. Electrostatic
`interactions may either favor or disfavor folding, depending on the sign of the charges.
`Experiments with peptides in the gas phase have shown that folding can be spontane-
`ous (Chin et al., 2006); hence, at least in some circumstances the entropy loss upon
`folding can be overcome by the enthalpy gain from electrostatic interactions, hydro-
`gen bonding, and/or van der Waals forces. Often electrostatic interactions are an
`important driving force for the folding process in the gas phase (Chin et al., 2006).
`Because proteins are typically found in an aqueous environment, these gas-phase
`experiments offer only limited insights to the understanding of protein folding under
`solution conditions. The high dielectric constant of water means that the strength of
`electrostatic interactions is significantly reduced, and therefore is a much less impor-
`tant driving force for folding, if at all. Moreover, the peptide chain now has the ability
`to form hydrogen bonds with water, as well as to interact with water molecules through
`van der Waals forces. Thus, intramolecular interactions like van der Waals forces and
`hydrogen bonds also are not immediately apparent driving forces for folding.
`And yet, proteins do fold in water. An important driving force of this folding is
`the negative effect of solute–water interactions on the interaction between the water
`molecules themselves. Pure water may be viewed as a collection of oxygen atoms
`suspended in a sea of hydrogen atoms. On average, four hydrogen atoms surround
`one oxygen atom in a (imperfect) tetrahedral shape, with two of those hydrogen
`atoms close enough to describe the bond as covalent and two slightly further away,
`forming a hydrogen bond. This is, however, a highly dynamic system, and there
`will be a constant exchange between covalently bound and hydrogen bond-linked
`hydrogen atoms. In essence, any solute will negatively affect this dynamic system;
`this is known as the hydrophobic effect (Dill et al., 2005). Whether a solute dissolves
`in water, and how much, is a matter of accounting: as long as there is a negative
`change in Gibbs free energy for the system as a whole upon dissolution of the solute,
`the compound will dissolve. Thus, the negative energetic contribution by distorting
`the dynamic water network needs to be counterbalanced by the positive contribution
`of the solute dissolving, which includes increased entropy of the solute upon dis-
`solution as well as hydrogen bonding and van der Waals interactions with the water
`molecules. Due to the ability to form hydrogen bonds and significant van der Waals
`interactions, polar (hydrophilic) compounds dissolve to a much larger extent in water
`than nonpolar (hydrophobic) compounds.
`Most proteins contain a significant amount of nonpolar amino acid residues and
`their dissolution in an aqueous environment would be energetically unfavorable.
`In contrast, the dissolution of the polar amino acids would be a favorable process.
`By folding of the amino acid chain such that the hydrophobic amino acids are hid-
`den from the aqueous surroundings, a protein significantly reduces the hydrophobic
`effect by the nonpolar residues, while maintaining the positive interaction between
`the polar residues and the water molecules. The hydrophobic effect and the result-
`ing “hiding” of nonpolar amino acids in the core of the protein is believed to be the
`main driving force for folding (Dill, 1990; Kauzmann, 1959). Further folding and
`specificity of the fold are then governed by other interactions like hydrogen bonding,
`
`Downloaded by [Jeff Bird] at 14:03 27 June 2015
`
`Ex. 2004-0004
`
`

`
`Physical Instability of Peptides and Proteins
`
`111
`
`salt bridge formation, and van der Waals interactions between tightly packed resi-
`dues (Rose and Wolfenden, 1993). Finally, the ultimate fold may be stabilized by the
`formation of cystines.
`The importance of the hydrophobic amino acids for protein folding is also sug-
`gested by the relatively conservative changes of the core amino acids for similar pro-
`teins across species. That is, even though the overall amino acid sequence may be
`significantly different for a given protein isolated from various species, the differences
`in the (usually hydrophobic) amino acids forming the core of the protein are usually the
`smallest (Mirny and Shakhnovich, 2001), and mutations in these amino acids are more
`likely to yield an inactive protein (Guo et al., 2004). As a result, even proteins with a
`mere 30–40% similarity in amino acid sequence can yield very similar protein folds.
`Considering the above, it should be no surprise that natively unfolded proteins
`generally do not contain such a core of hydrophobic amino acids. In fact, it is likely
`the absence of a significant amount of hydrophobic amino acids, along with many
`charged residues, that allows these proteins to have little tertiary fold (Uversky and
`Dunker, 2010). However, they often do have a specific secondary structure, which
`suggests that for amino acid chains, the intrachain hydrogen bonding is more favor-
`able than hydrogen bonding to water.
`
`6.3.2 the energy landscaPe of a Protein molecule
`As discussed above, proteins may spontaneously fold in aqueous solution. That means
`that the change in Gibbs free energy upon folding is negative, that is, ΔGf < 0, and
`thus the change in Gibbs free energy of unfolding is positive (ΔGu > 0). However,
`due to the complex interaction between protein and solvent, the Gibbs free energy is
`not a simple linear function of temperature (Privalov, 1990; Robertson and Murphy,
`1997). Let us first examine a simple two-state reversible folding process between a
`protein in its unfolded state (U) and in a folded state (N) (Scheme 6.1):
`
`
`
`U
`
`N(cid:31)
`
`(Scheme 6.1)
`
`The change in Gibbs free energy for this folding process can be approximated
`using a modified form of the Gibbs–Helmholtz equation (Equation 6.1), in which the
`temperature dependence of ΔH and ΔS are approximated by a constant difference
`in heat capacity between the native and unfolded stated of the protein, ΔCp. In this
`equation, Tm is a temperature where ΔG is zero; ΔHf is the enthalpy change upon
`folding at this temperature, and ΔCp,f is the change in heat capacity upon folding.
`
` (6.1)
`
`
`
` 
`
`T T
`
`m
`
`−(
`T T
`m
`
`) −
`
`T
`
`ln
`
` 
`
`+
`
`∆
`C
`
`p,f
`
` 
`
`T T
`
`m
`
`1
`
`−
`
` 
`
`∆
`G
`
`f
`
`=
`
`∆
`H
`
`f
`
`
`
`Data on Tm, ΔCp, and ΔHf can be obtained, for example, using differential scan-
`ning calorimetry (DSC)*. Plotting this data using Equation 6.1 will yield a parabola
`
`* Note that in a typical DSC experiment the protein is folded at the start of the experiment. Thus, the ΔH
`and ΔCp obtained are those for the unfolding process.
`
`Downloaded by [Jeff Bird] at 14:03 27 June 2015
`
`Ex. 2004-0005
`
`

`
`112
`
`Pharmaceutical Formulation Development of Peptides and Proteins
`
`Myoglobin
`
`40
`
`20
`
`0
`
`–20
`
`ƊGf (kJ/mol)
`
`–40
`
`240
`
`Lysozyme
`320
`
`360
`
`280
`T (K)
`FIGURE 6.1 Graphical representation of the thermodynamic stability of two model pro-
`teins as a function of temperature as derived from the modified Gibbs–Helmholtz equation
`(Equation 6.1). Figure created using numerical data from Anjum et al. (2000), with Tm =
`340.4 K, ΔHf = −343 kJ/mol, and ΔCp,f = −11.45 kJ/mol for myoglobin in a pH 6.1 buffer;
`Tm = 335.7 K, ΔHf = −372 kJ/mol, and ΔCp,u = −6.52 kJ/mol for lysozyme in a pH 4.8 buffer.
`(Data from Anjum et al., Biochim. Biophys. Acta 1476, 2000.)
`
`(Figure 6.1), which is known as the protein stability curve. It has been observed that
`many proteins have their highest thermodynamic stability around 283 K, indepen-
`dent of their melting temperatures (Rees and Robertson, 2001). This is significantly
`below physiological temperatures for many organisms, probably because some struc-
`tural flexibility is required for activity.
`Figure 6.1 shows there are two crossings where ΔG = 0, suggesting that proteins
`can unfold due to both increased as well as decreased temperatures. The latter, cold
`denaturation (Privalov, 1990), usually occurs at temperatures below 270 K and thus
`is less likely to be observed in standard analytical techniques due to ice formation.
`Furthermore, the kinetics of unfolding slows down with decreased temperature,
`which may result in kinetic trapping of the protein in its folded structure. Finally, it
`is important to note that under physiological conditions, the magnitude of ΔGf,max is
`relatively small, typically ca. 10–50 kJ/mol. This is a rather weak stabilizing interac-
`tion, considering that a typical hydrogen bond contributes about 5–30 kJ/mol.
`The pathway from unfolded to folded state is, for many proteins, likely not as sim-
`ple as suggested by Scheme 6.1. Unfolded proteins may assume an enormous number
`of conformational states; indeed, a simple calculation shows that in a typical sample
`of unfolded protein molecules, each molecule is likely to be found in a different con-
`formational state.* Yet, proteins can spontaneously fold to their native conformation
`within a second. This suggests that each protein molecule must necessarily follow a
`slightly different pathway to the folded state. This complex folding process can be
`conceptualized in terms of a biased random walk, wherein proteins fold via a large
`number of small conformational changes, with the likelihood of any conformational
`change occurring being biased toward those that lower the overall free energy of the
`protein (Bryngelson et al., 1995). The collection of all possible conformational trajec-
`tories and associated free energies forms an “energy landscape.” To better visualize
`
`* Take, for example, a protein of 100 amino acids, and allow each amino acid only two different confor-
`mations. This already yields 2100 = 1030 different potential conformations. This exceeds the number of
`molecules of a specific protein on Earth.
`
`Downloaded by [Jeff Bird] at 14:03 27 June 2015
`
`Ex. 2004-0006
`
`

`
`Physical Instability of Peptides and Proteins
`
`113
`
`the high-dimensional space represented by this enormous collection of conformational
`states and energies, the energy landscape is often conceptualized as a “folding fun-
`nel,” wherein the vertical position on the funnel is representative of the free energy
`of a given conformation, and the circumference of the funnel is representative of the
`number of states having a given free energy (Bryngelson et al., 1995). Thus, the large
`number of unfolded states would be found at the top of the funnel, and the singular
`native state conformation would be found at the funnel bottom (Figure 6.2a).
`
`Entropy
`
`(a)
`
`Gibbs free energy
`
`Misfolded
`
`Oligomer
`
`Folding
`intermediate
`
`Gibbs free energy
`
`Native state(s)
`
`Amorphous
`aggregates
`
`Fibrils
`(b)
`FIGURE 6.2 Energy landscape of a protein. (a) (See color insert.) An idealized folding fun-
`nel for a single protein molecule. At the high Gibbs free energy end (top of picture), the protein
`molecule can adopt many different conformations; the width of the funnel can be viewed as a
`measure of the conformational entropy. At the bottom of the funnel a singular folded state with
`very limited conformational entropy is present. (b) A more realistic two-dimensional representa-
`tion of the energy landscape of a protein. On the left-hand side, the folding of a single protein is
`shown; as depicted, there may be several folds with almost the same low Gibbs free energy. Also,
`there may be a folding intermediate(s) and misfolded species with higher energy which can be
`significantly populated due to kinetic barriers. In the middle and on the right-hand side possible
`energy states are shown for ensembles of protein molecules, resulting in various aggregated
`species (oligomers, fibrils, and amorphous aggregates). These may have lower Gibbs free energy
`than the native protein, but may also be populated due to a large kinetic barrier toward refolding.
`
`Downloaded by [Jeff Bird] at 14:03 27 June 2015
`
`Ex. 2004-0007
`
`

`
`114
`
`Pharmaceutical Formulation Development of Peptides and Proteins
`
`The energy landscape of proteins can probably be better conceptualized as a
`jagged funnel (Figure 6.2b), where the protein fold, or a subpopulation of protein
`molecules, may be kinetically trapped in local minima, rather than in the thermo-
`dynamically lowest energy state. Note that the aggregated states, in particular the
`fibrillar state, are shown in Figure 6.2b as being equally or even more thermody-
`namically stable than the native state. That is, the protein fold observed for native
`proteins may well be considered a metastable state, with kinetic barriers preventing
`rapid population of the aggregated and/or fibrillar state (Baldwin et al., 2011). In
`the human body, various regulatory processes have developed that are designed to
`degrade and eliminate improperly folded proteins, and thus prevent and/or reduce
`the rate of fibril formation. In a pharmaceutical formulation, there are no processes
`that remove misfolded protein, meaning that aggregation and fibrillation can occur
`for proteins that are not known to aggregate or fibrillate in vivo, or can occur much
`faster than observed in vivo. Aggregation and fibrillation is further discussed in
`Section 6.4.3.
`The jagged funnel depicted in Figure 6.2b should not be seen as static; changing
`the solution conditions will alter the relative magnitudes of the local minima relative
`to the global minimum, possibly resulting in a new global minimum. This may also
`occur upon binding of a protein to a ligand or to its receptor, if this involves signifi-
`cant changes in protein structure. The energy barriers between the various states will
`likely also change, and may either increase or decrease. This will affect the kinetics
`of the physical degradation processes taking place. As a result, even small changes in
`solution conditions can have a major impact on the main degradation route.
`
`6.4 PROTEIN PHYSICAL DEGRADATION
`The physical degradation of proteins refers to any loss in bioactive protein that does
`not involve formation or breakage of chemical bonds and is sometimes also referred
`to as denaturation. It can be subdivided in four, often interrelated processes: unfold-
`ing, adsorption, aggregation, and precipitation.
`
`6.4.1 Protein unfolding
`In the previous section, the spontaneous folding of a protein into a specific three-
`dimensional structure was discussed. Here, we look at the reverse process: the
`spontaneous unfolding of a protein, sometimes also referred to as denaturation. As
`discussed above, under physiological conditions, the most thermodynamically stable
`state for a single protein molecule (usually) is the folded state. Any deviation from
`physiological conditions, for example, a change in temperature, pH, or ionic strength,
`will change the intramolecular interactions within the protein, as well as the inter-
`actions between protein and water. Thus, one may expect a change in the protein
`folding stability upon changing the environment of the protein. As long as those
`changes are fully reversible upon removing the stress factor or upon administration
`to the patient, this may appear irrelevant for a therapeutic protein in a formulation.
`However, as will be discussed in more detail in Section 6.4.3, (partial) unfolding is
`commonly the first step in protein aggregation, which is often irreversible. Moreover,
`
`Downloaded by [Jeff Bird] at 14:03 27 June 2015
`
`Ex. 2004-0008
`
`

`
`Physical Instability of Peptides and Proteins
`
`115
`
`(partial) unfolding followed by subsequent refolding may trap the protein in a non-
`native and thus inactive conformation. Finally, unfolded proteins are often more sus-
`ceptible to chemical degradation. Protein unfolding is thus, in general, detrimental
`to a protein’s physicochemical stability. At the same time, it should be realized that a
`pure thermodynamic treatment of Scheme 6.1, which assumes a fully reversible pro-
`cess, suggests that there will always be a population of unfolded protein molecules.
`This is exemplified in Equation 6.2, where the equilibrium constant K for the fold-
`ing process is given by the population of the native state (N) divided by that of the
`unfolded state (U). Although under conditions in which the native protein is stable,
`U will be significantly smaller than N, it is never zero, and hence there will always
`be a number of unfolded protein molecules.
`
`(6.2)
`
`
`
`N U
`
`∆G
`
`f = −
`
`ln
`RT K
`
`= −
`
`RT
`
`ln
`
`
`
`As may be obvious from Figure 6.1, temperature is an important factor in deter-
`mining the thermodynamic stability of a protein. Increasing the temperature from
`physiological conditions results in increasing levels of unfolded protein due to an
`increasingly smaller ΔGf. Ultimately, one will reach a temperature where ΔG = 0.
`At this temperature, the melting temperature Tm, half of all protein molecules are
`unfolded. However, some proteins may contain domains that behave as indepen-
`dent units, and such proteins can thus have multiple melting temperatures. Also, the
`behavior of many proteins in solution does not comply with the required reversibility
`for Scheme 6.1 and Equation 6.2. For example, some proteins may be kinetically
`locked into a conformation, thus requiring more energy (and hence higher tempera-
`ture) than the equilibrium thermodynamics calculations would suggest (Sanchez-
`Ruiz, 2010). Rapid aggregation upon unfolding will also affect any measurements of
`the thermodynamics by depleting the solution of the unfolded species. These pitfalls
`are important to take into account when evaluating protein unfolding data obtained
`from various analytical methodologies.
`Unfolding as a result of thermal stress (but also other stresses) may sometimes
`proceed through a distinct intermediate, as exemplified in Scheme 6.2:
`
`N
`
`(Scheme 6.2)
`
`
`
`(cid:31) (cid:31)
`U
`I
`This intermediate is one of the local minima in the energy funnel (Figure 6.2b),
`and often shows almost completely native secondary structure, but a much less well-
`defined tertiary structure. It is commonly referred to as a molten globule, and is
`likely a main starting point for aggregation (Wang et al., 2010).
`As noted earlier, cooling a protein solution may also cause unfolding (Privalov,
`1990). In most cases, the cold denaturation temperature is much lower than 0°C, and
`the whole solution will have turned to ice well before that temperature is reached.
`In addition to the decrease in conformational stability caused by low temperature,
`proteins may be destabilized as the formation of ice causes both the protein and
`any cosolutes to become more concentrated (see also Chapter 10). Finally, as also
`discussed later, proteins may adsorb at the ice–water interface, with concomitant
`partial unfolding.
`
`Downloaded by [Jeff Bird] at 14:03 27 June 2015
`
`Ex. 2004-0009
`
`

`
`116
`
`Pharmaceutical Formulation Development of Peptides and Proteins
`
`Changes in pH are another common cause of protein unfolding. Lowering or
`increasing the pH, away from its isoelectric point (pI, the pH at which the net charge
`on the protein is zero) can alter the charge states of ionizable amino acids, resulting
`in more electrostatic repulsion, which in turn can lead to unfolding. Also, here the
`unfolding process may proceed through an intermediate, although likely a different
`intermediate than that observed for thermal unfolding.
`The influence of ions and ionic strength on protein unfolding is more complex.
`An increase in ionic strength will decrease any intramolecular repulsion that may
`be present, but will also negatively affect favorable salt bridges on the surface of the
`protein. Thus, both stabilizing and destabilizing effects may occur. For example,
`sufficiently high concentrations of certain anions at low pH have been shown to
`counteract the destabilizing effect of the lowered pH (Goto et al., 1990).
`Salts may also exert an indirect effect on protein thermodynamic stability, pos-
`sibly by affecting water structuring. There is an empirical relationship between the
`type of salt and its (de)stabilizing behavior (cf. Boström et al., 2005; Fesinmeyer
`et al., 2009; Sedlák et al., 2008; Tadeo et al., 2007; Yang et al., 2010), known as the
`Hofmeister series (Figure 6.3) (Hofmeister, 1888) (translated version in Kunz et al.
`[2004]). Originally based on their effect on salting-in and salting-out of proteins,
`the Hofmeister series salts also often follow roughly the same sequence in a variety
`of other phenomena. However, solution conditions can have a major impact on the
`sequence of the series and the Hofmeister series must therefore be used with caution
`(cf. Boström et al., 2005, 2011). Most importantly, a significant Hofmeister series
`effect of these salts does not show up until relatively high concentrations (>200 mM),
`which usually are not encountered in a typical pharmaceutical formulation.
`Adding further complexity to the effect of ions is the possible presence of specific
`binding sites. For example, ribonuclease A has a binding site for phosphate, while
`many other proteins, such as insulin, and various factors in the blood coagulation
`process contain metal-binding sites. Thiocyanate, a destabilizing anion according to
`the Hofmeister series (Figure 6.3), binds to the zinc- and phenol-containing insulin
`hexamer, thus increasing its thermodynamic stability (Huus et al., 2006). As a result
`of this complexity, it is often difficult to predict the effect of adding salts on protein
`stability; this needs to be studied on a case-by-case basis.
`Not only organic solutes like ethanol or acetonitrile but also typical preservation
`agents like benzyl alcohol, phenol, and metacresol will generally decrease the fold-
`ing stability of a protein at physiological temperature and may thus lead to unfolding
`or misfolding. This destabilization is mostly due to the reduction in polarity of the
`solvent, which in turn reduces the hydrophobic effect. However, here also care needs
`
`Decreased protein stability; increased salting-in of protein
`
`SCN–
`I–
`Cl–
`–
`2–
`2–
`Anions
`NO3
`SO4
`HPO4
`Ca2+
`Gdn+
`Na+
`+
`Mg2+
`K+
`Cations
`NH4
`FIGURE 6.3 The Hofmeister series of anions and cations. From left to right high concentrations
`(>0.2–0.3 M) of these ions will result in salting-in of the protein, as well as decreased protein
`therm odynamic stability. Gdn+ is the guanidinium cation. (From Boström et al., 2005; Fesinmeyer
`et al., 2009; Sedlák et al., 2008; Tadeo et al., 2007; Yang et al., 2010; Zhang and Cremer, 2006.)
`
`Downloaded by [J