SUPPLEMENT
`Volume 42
`
`Number 2S
`
`Technical Report No. 10
`
`':: .. 1n,- .
`<ii~~-- ........
`""!o.
`
`•
`
`\
`
`•• ., •
`
`Parenteral Formiiia-t1oiis of·
`Proteins and Peptides:
`Stability and Stabilizers
`
`i
`
`,•.•. .
`
`~
`
`.
`
`.
`
`.
`
`Journal.of Parenteral ,.
`.. .
`. ..
`Scien9e;a:nd Tech,nology
`
`"!-!• · i.j
`
`'
`
`.
`
`.
`
`'
`
`.
`:
`-
`::;.~-. .c--
`- •-
`.
`,...
`A publication of Th·e Parenteral Drug Association
`. :.. . :.t- ~~.rt~_~: ... ~ - ,.·.
`. .. ~-:
`..
`-
`
`
`
`
`MAIA Exhibit 1019
`MAIA V. BRACCO
`IPR PETITION
`
`

`

`Parenteral Formulations of Proteins and Peptides: Stability and Stabilizers
`
`YU-CHANG JOHN WANG• and MUSElTA A. HANSONt
`
`• Research Laboratories, Ortbo Pharmaceutical Corporation, Raritan, New Jersey. t Rorer Central Research, Fort Washington,
`Pennsyltania
`
`I. Introduction
`
`Recent developments in biotechnology will cause a tre(cid:173)
`mendous increase in the number of injectable products
`containing proteins or peptides, and there exists a definite
`need for skill in formulating such products. Scientists who
`have no difficulty in formulating chemical compounds
`into injectable products often find that dealing with pro(cid:173)
`teins and peptides is a completely different experience. To
`illustrate this point, the degradation of a chemical com(cid:173)
`pound, such as penicillin or aspirin, may be readily detect(cid:173)
`ed by use of HPLC or other assays indicative of stability,
`but degradation/denaturation of a protein is not a simple
`one-step reaction and products of denatured protein are
`not readily detectable. In predicting shelflife, formulators
`frequently determine the degradation rate of a chemical
`compound at an elevated temperature (50 °C or 37 °C),
`then extrapolate the result to storage at low temperatures.
`This approach is often not applicable to proteins because
`some may not be stable at 50 °C for any reasonable period
`of time, but may be extremely stable at 5 °C. The Arrhe(cid:173)
`nius equation often works only in a limited temperature
`range.
`This report reviews the scientific literature on degrada(cid:173)
`tion pathways, and on methods of stabilizing proteins and
`peptides, particularly methods that can be employed for a
`parenteral product. The peptides and proteins employed
`in parenteral formulations vary tremendously in their
`properties. The following list of selected market products
`used parenterally is arranged in ascending order of com(cid:173)
`plexity.
`
`Thymopentin*
`Oxytocin
`Leuprolide
`Gonadorelin
`Cyclosporin
`Glucagon
`Calcitonin
`
`Number
`of Amino Mol.
`Acids
`wt.
`5
`9
`9
`10
`11
`29
`32
`
`1007
`1209
`1182
`
`Comments
`
`orally active
`
`produced by
`chemical synthesis
`
`Insulin
`lnterleukin-2*
`Interferon
`Erythropoietin*
`Growth hormone
`
`51
`133
`165
`166
`191
`
`5808
`
`19,000
`35,000
`
`This material may be protected by Copyright law (Title 17 U.S. Code)
`
`Received September 16, 1987.
`This review was prepared under the auspices of The Parenteral Drug
`Association Research Committee.
`
`Comments
`
`umber
`of Amino Mot.
`Acids
`wt.
`267
`
`527
`
`150,000
`
`Urokinase
`Asparaginase
`Alteplase•
`Chymopapain
`RhoGAM®
`Atgam®
`Gamimune®
`Orthoclone OK~ 3
`Recombivax HB®
`
`polyclonal antibody
`intravenous infusion
`monoclonal antibody
`vaccine against
`hepatitis B virus
`* These products are not on the U.S. market as of July 1987.
`
`Information on how to stabilize proteins and peptides in
`products for parenteral use is not readily available to
`formulation scientists. In most cases, information con(cid:173)
`cerning the stabilizing of proteins is contained in journals
`or textbooks that are unfamiliar to formulation scientists.
`Scientists working in areas other than pharmaceutical
`chemistry have long been concerned with stabilizing pro(cid:173)
`teins in a hostile environment and/or for long-term stor(cid:173)
`age as, for example, proteases, carbohydrases in deter(cid:173)
`gents, hydrolytic enzymes in food processing, and en(cid:173)
`zymes used in analytical and diagnostic applications.
`Information generated in those areas is very useful, and
`this review seeks to assemble that information in one
`place, together with knowledge from existing parenteral
`products containing proteins and peptides, thus providing
`comprehensive information to the scientist setting out to
`design a stable and elegant injectable formulation con(cid:173)
`taining protein, peptide, or both.
`
`II. Degradation of Proteins and Peptides
`
`The degradation of proteins and peptides can be divided
`into two main categories. One involves a covalent bond
`and the other does not. The latter process is often referred
`to as denaturation.
`
`A. Covalent Bond Reactions
`
`Nine major reactions involve covalent bonds:
`a. Hydrolysis: The peptide (amide) linkage (RNH(cid:173)
`CO- R) is much more resistant to hydrolysis than is the
`ester linkage (R- O- CO-R) and amide bonds are consid(cid:173)
`ered stable unless hydrolysis is assisted by a neighboring
`group. The formulation factor that most influences the
`hydrolytic rate is solution pH. Examples of pH-stability
`studies on therapeutic peptides are: degradation of nafare(cid:173)
`lin (Johnson, 1986), thermal stability of D-Trp-LHRH
`
`S4
`
`Journal of Parenteral Science & Technology
`
`-S4-
`
`
`
`
`

`

`(Winterer, 1983}, acid hydrolysis of captopril, a quasi(cid:173)
`dipeptide (Timmins, 1983), heat stability of urokinase
`(Miwa, 1981 ), and summaries presented in "Analytical
`Profiles" (K. Florey, Ed.) for oxytocin (10,563), gramici(cid:173)
`din (8, 179) and bacitracin (9, 1). In general, these reports
`demonstrated tha.t peptides are stable for parenteral use.
`For instance, D-Trp-LHRH was stable at 60 °C for 5
`days and seven cycles (!f freezing and thawing; gramicidin,
`although used only for dermatological indications when
`dissolved in glycols, can be autoclaved. Information avail(cid:173)
`able on package inserts also shows a general trend of
`stability: for instance, leupr.olide and oxytocin injections
`have a shelf life of at least two years under refrigeration
`and three months at room temperature. LH-RH injection
`(Relefact~), made by Hoechst AG, does not even require
`storage under refrigeration. Oxytocin injection made by
`Sandoz was reported to be stable at room temperature for
`five years (Wolfert, 1975). Protirelin, a tripeptide
`(PyrGlu-His-Pro) is stable for 20 hours at 80 °Cat both
`pH 3.3 and pH 6 (Rao, 1987).
`Hydrolysis of peptides can become very complex, be(cid:173)
`cause the side chains on certain amino acids are capable of
`influencing the hydrolysis. The steric effect of amino acid
`side chains on hydrolysis can be exemplified by valine and
`leucine; their side chains, isopropyl and isobutyl, provide a
`hindrance to the hydronium ion in its attack on the peptide
`bond (Hill, I 965).
`The positive charge at the terminal amino group tends
`to repel hydronium ions, thus making the adjacent peptide
`bond resistant to acid hydrolysis (Hammel, 1954). How(cid:173)
`ever, the positive charge on t-ammonium, imidazolium, or
`guanidinium groups on the side chains does not decrease
`the rate of hydrolysis of peptide bonds significantly. The
`charge on the side chain of lysine, histidine, or arginine is
`too far removed from the peptide bonds to play an impor(cid:173)
`tant stabilizing role.
`The peptide bonds formed by the amino group of serine
`and threonine are more labile (Harris, 1956) (Eq. 1 ). The
`bonds formed by aspartyl residues are very susceptible to
`hydrolysis in dilute acid (Eq. 2). It has been suggested
`
`OH
`H20
`I
`9-yHz
`TH 2
`?t
`!
`~ HN-CHR-C-NH-CH-C-NH ~ _L___ ~ HN-CHR-C-N-CH-C-NH ~
`II
`11
`0
`0
`H,Ol
`9--TH'
`~ HN -CHR-C= 0
`l NH 2 O
`?H
`yH,
`
`H20
`
`HO,
`
`CH-C-NH ~
`II
`
`/
`
`Equation 1-Hydrolysls promoted by serine residue.
`
`~HN-CHR-C=O
`
`CH-C-NH~
`/
`II
`NH 2 O
`
`-A-8-C-Asp-D-E-Asp-r-G-
`
`i
`
`-A-8-C + D-E + r -G- + 2 Asp
`Equation 2-Hydrolysls promoted by aspartyl residue.
`
`0 0
`
`d
`
`O 0-PROTEIN
`
`OH
`~CHO
`
`H20
`
`00 C_)\
`~O
`Equation 3-Deglycosylatlon.
`
`t HO-PROHIN
`
`that the negatively charged carboxyl group of aspartic
`acid attracts hydrogen ions in dilute acid and thereby
`increases the !ability of neighboring peptide bonds; these
`bonds are severed and aspartic acid is liberated (Schultz,
`1961).
`Hydrolysis 0th.er than of peptide bonds can also be
`responsible for protein inactivation. For example, hydrol(cid:173)
`ysis of the tyrosine-O-sulfate moiety was responsible for
`inactivation of cholecystokinin (Wunsch, 1983). Acetal or
`hemi-acetal (Eq. 3) hydrolysis by acid is responsible for
`severing sugar moieties from glycoprotein (deglycosyla(cid:173)
`tion). Although reaction rates of deglycosylation have not
`been reported, data generated for the digitalis glycoside,
`digoxin (Sternson, 1978) can be useful as a reference.
`
`b. lmide Formation: The a-amino group of asparagine,
`glutamine, and aspartic and glutamic acids attacks the
`side-chain carbonyl carbon of these amino acid residues,
`thus forming aspartimides or glutarimides (Eq. 4). These
`cyclic imides are sensitive to hydrolysis and can be opened
`in two ways. One is to break the newly formed bond; this is
`the deamidation reaction when imides are formed by as(cid:173)
`paragine and glutamine. Another is to break the original
`C-N bond on the peptide backbone, leading to the forma(cid:173)
`tion of (3- (for Asn and Asp) or -y- (for Gin and Glu)
`carboxyl isomers, a process often referred to as isomeriza(cid:173)
`tion. The imides can also undergo racemization at the a(cid:173)
`carbon position.
`Geiger and Clarke (1987) conducted a comprehensive
`study on the rates of the aforementioned reactions, using a
`hexapeptide that is the residue 22- 27 of adrenocortico(cid:173)
`tropic hormone (Val-Tyr-Pro-Asn-Gly-Ala). The
`amounts of peptides obtained from all possible degrada(cid:173)
`tion routes are shown in Eq. S.
`Because early reports on deamidation and isomeriza(cid:173)
`tion were not analyzed with respect to other concurrent
`reactions, these reactions will be discussed separately in
`the following sections.
`
`c. Deamidation: Deamidation is the hydrolysi~ of the
`
`0
`II
`
`8
`
`CH"zc-o f ~
`I
`
`H~-CH-C~
`
`R'-t,0.4-C-( D[AWIOAllOH
`
`b ro• A1n • GI"
`✓ 0
`f fl
`I NH-CH-C ~
`----.!-
`R'- MH-C-c-o9
`g ThNSPtPTIOATIOH
`
`II
`CH;\,
`
`CHf\ f ~
`I H N-CH-C~
`R'-NH-~( IIAC[Ml?AllON
`II
`0
`Equation 4-lmlde formation and subsequent reactions.
`
`I WIOC roRMATION
`
`0
`If
`
`Vol. 42, Supplement 1988
`
`S5
`
`-S5-
`
`
`
`
`

`

`/
`
`L-OEAIAIDATED ( 15%)
`PEPTIDE
`
`L-Asn PEPTIDE -
`(Oll)
`
`(cid:141) L-IIAIOE ~ L-lso PEPTIDE (56%)
`( 0 . 1%)
`
`D-OEAIAIOATEO (7'!)
`D -I IAIDE ~ PEPTIDE
`•
`
`D- lso PEPT IDE (21")
`Equation 5-Proportions of reaction products of Val-Tyr-Pro-Asn-Gly(cid:173)
`Ala (Gelger, 1977).
`
`side chain amide on glutamine and asparagine (Eq. 3). lt
`is a major route of degradation for insulin (Rytting, 1982)
`and ACTH (Graf, 1971 ). Amnonia, if generated in signif(cid:173)
`icant amount upon hydrolysis, can accelerate hydrolysis
`of the peptide backbone (Gilbert, 1949). The deamidation
`rates were determined for 42 synthetic pentapeptides con(cid:173)
`taining glutarninyl or asparaginyl residues (McKerrow,
`1971; Robinson, 1973). Asparagine is more labile than
`glutamine, and is most labile in the presence of an adja(cid:173)
`cent glycine. The half-lives of these model compounds at
`37 °C in pH 7.2 phosphate buffer range from 20 days to
`nine years. Geiger and Clarke ( 1987) concluded that dea(cid:173)
`midation occurs almost exclusively through an imide
`pathway, and that direct solvent hydrolysis is insignifi(cid:173)
`cant. The heat of activation for deamidation of their L(cid:173)
`Asn hexapeptide is approximately 21 Kcal/ mole, which is
`higher than for unassisted hydrolytic reactions.
`
`d. lsomerization: Asparagine, glutamine, aspartic acid,
`and glutamic acid can cyclize back onto the peptide chain
`(Eq. 4). Subsequent hydrolysis of such cyclic imides re(cid:173)
`sulted in isomerization. Hydrolysis of asparagine and glu(cid:173)
`tamine is accelerated by low pH; that of aspartic acid and
`glutamic acid is accelerated by high pH (Hill, 1965).
`Isomerization in the sequence 3-4 represents one of the
`mechanisms for the inactivation of secretin in solution
`(Wunsch, I 983). lsomerization is often referred to as
`transpeptidation.
`
`e. Racemization: Racemization produces enantiomers
`in both acidic and alkaline solutions. Racemization of the
`serine residue in analogs of gonadotropin-releasing hor(cid:173)
`mone was studied by
`ishi (1980). A comprehensive re(cid:173)
`view of racemization in alkali-treated food proteins was
`provided by Masters and Friedman(! 980). Racemization
`is thought to proceed by abstraction of the a-proton from
`an amino acid in a peptide to give a negatively charged
`planar carbanion. A proton can then be returned to this
`optically inactive intermediate, thus producing a mixture
`ofD and L enantiomers (Eq. 6). An electron-withdrawing
`group in the side chain will stabilize the negatively
`charged intermediate. Thus, asparagine, tyrosine, serine,
`phenylalanine, etc., all promote racemization.
`In an imide-containing peptide, the nitrogen atom and
`the a- and ,B-carbonyl groups all contribute to the stable
`
`CHR O
`H 0
`I
`II
`I
`II
`~NH-C-C- .:::= ~ NH-C-C
`I
`CHR
`
`"
`
`CHR O
`I
`II
`~NH-C -
`C~ -
`I
`H
`'oH 8
`Equation 6-Racemization.
`
`S6
`
`R O
`R' 0
`R" 0
`I
`II
`I
`II
`I
`II
`NH -CH-C- NH-CH-C-NH-CH-C ~
`z.____,..,
`
`R" 0
`II
`NH 2-CH-C ~
`
`I
`
`Equation 7- Diketoplperazlne formation.
`
`5--S -
`
`5 - -- - S
`Equation 8-Disulfide exchange.
`
`s-s
`
`s-
`
`s
`
`resonance of a carbanion, thus racemization is facilitated
`through imide formation (Eq. 4). At neutral pH this path(cid:173)
`way is principally responsible for the racemization of D
`and L isomers. (Geiger and Clarke, 1987)
`
`f. Diketopiperazine Formation: Rearrangement of the
`N-terminal dipeptide will result in the splitting off of a
`cyclic diketopiperazine at high pH (Eq. 7). Praline and
`glycine in the N-terminal promote the reaction. Recently,
`it was reported (Bouvette and Digenis, 1987) that the
`cyclization of aspartame is minimal at pH 6 to 7, moder(cid:173)
`ate at 7.0- 8.5, and rapid at pH above 8.5 .
`
`g. Oxidation: Tryptophan, methionine, and cysteine
`are susceptible to oxidation by air. Disulfide bridging is
`the main course of reaction in somatostatin. Oxidation of
`the sulfhydryl group is promoted at both neutral and basic
`pH and the rate-pH profile for captopril showed an in(cid:173)
`crease in oxidation rate starting at pH 4 (Fig. I).
`
`h. Disulfide Exchange: The reaction (Eq. 8) is base
`(OH-)-catalyzed and promoted by mercaptoethanol, a
`sometimes employed antioxidant. The reaction is concen(cid:173)
`tration dependent, and oligomers are frequently formed as
`a result of disulfide bonds between peptide chains. Im(cid:173)
`proper linkages of disulfide bond were responsible for a
`
`so.-------------,
`
`40
`
`"' ~ 30
`
`~ ' .. >-.. ::?
`
`"'
`
`20
`
`•
`
`10
`
`•
`
`•
`
`•
`
`•
`
`0 2
`
`3
`
`4
`pH
`Figure 1-Rate-pH profile for oxidation of captopril at 50 °C (Tim(cid:173)
`mins, 1982).
`
`5
`
`Journal of Parenteral Science & Technology
`
`-S6-
`
`
`
`
`

`

`reduction in biological activity of interleukin-2 (Kenney,
`1986).
`i. Photodecomposition: Tryptophan is one of the most
`photosensitive amino acid residues in proteins (Holtz,
`1977). Decomposition of tryptopban, which liberates am(cid:173)
`monia and forms photoproducts of increased molecular
`weight, has been responsible for the discoloration of pro(cid:173)
`teins.
`
`B. Denaturation
`a. Protein Conformation: In considering the formula(cid:173)
`tion of a protein or peptide molecule for use in parenteral
`therapy, one must take into account conformation, i.e., the
`spatial three-dimensional structure of the molecule, how
`this structure can be disrupted (leading to loss of poten(cid:173)
`cy), and how this structure can be stabilized. The primary
`structure (amino acid sequence) and secondary structure
`(a-helix or ,8-sheet) contribute to the tertiary structure of
`a protein or peptide. It is this tertiary structure that must
`be stabilized against the forces that work to disrupt it. The
`tertiary structure of the protein suggests what forces can
`be used in stabilization. For example, the conformation of
`each peptide group is usually one of a stable energy state
`and directs the secondary and tertiary structures.
`Charged residues are generally on the surface, since it is
`energetically costly to bury them within the folded struc(cid:173)
`ture.
`These charged residues and their arrangement on the
`surface of the folded molecule contribute to an overall
`molecular dipole and determine how the molecule inter(cid:173)
`acts with solvent molecules in the solution. The nonpolar
`hydrophobic residues are usually found within the globu(cid:173)
`lar structure of these water-soluble proteins. Perhaps the
`exceptions to this rule are those proteins known to interact
`or associate with lipids. Elucidation of the structure of
`proteins with exposed hydrophobic areas suggests that
`lipid type molecules should contribute to overall stability.
`Finally, polar side-chain residues will also tend to be on
`the surface, hydrogen-bonding with each other to further
`stabilize the tertiary structure.
`Even this brief consideration of protein structure sug(cid:173)
`gests that the more a formulator knows about the struc(cid:173)
`ture of the protein, the better the chances are to stabilize
`the drug. It is also important to recognize the potentially
`denaturing forces that may be applied to the molecule
`during its processing, e.g., chemical stress from factors
`used in purification of the protein, such as pH, ionic
`strength or detergents, or physical stress from processes
`such as filtration and filling, in which the protein is subject
`to surface adsorption and shear. These factors must be
`considered in addition to "simple" solution stability once
`the product is in its final container. As biological poly(cid:173)
`mers, proteins can be unfolded or denatured from their
`thermodynamically preferred compact tertiary globular
`structure, usually with a concomitant loss of potency or
`activity. An understanding of how the preferred confor(cid:173)
`mation is determined and the mechanisms by which it can
`be unfolded is important in formulating the molecule as a
`stable drug.
`b. Multistate Mechanism: The denaturation process
`
`has been described variously as a two-state or multiple
`process and the kinetics of these transitions have been
`described (references in Table VII) . Three possible mech(cid:173)
`anisms can be entertained. First, the folded structure may
`simply unwind to a random coil without passing through
`an intermediate state (a real two-state process). Second,
`the folded molecule may pass through a series of energeti(cid:173)
`cally different, partially folded coils as intermediate
`states. (Schematic presentation of this process is shown in
`Fig. 2.) Third, there may be a core-organized intermedi(cid:173)
`ate structure that exists before complete unfolding to a
`random coil occurs.
`
`c. Effect of Solvent: Hydrogen bonds are perhaps most
`important in determining overall protein conformation,
`since they are the major stabilizers of the secondary a(cid:173)
`helix and ,8-sheet, as well as stabilizing the folded struc(cid:173)
`ture. Water, the solvent of most proteins, contributes to
`this hydrogen bonding. It is the nature of this water struc(cid:173)
`ture around a protein that further complicates studies of
`conformation. It is necessary to understand the solvent(cid:173)
`protein interaction because of the strong influence the
`solvent exerts on protein conformation. A disrupting sol(cid:173)
`vent effect is evident when even low levels of urea or
`guanidinium chloride are placed in solution with a pro(cid:173)
`tein, they completely disrupt the native structure of the
`protein molecule. On the other hand some solvent
`changes contribute to the stability of an ordered globular
`state.
`Because the hydrocarbon side chains are usually tucked
`inside the protein's globular structure, significant stabiliz(cid:173)
`ing effects result from their hydrophobic bonding. This
`conformation, too, is sensitive to solvent effects, especially
`if the solvent partitions the hydrocarbon sidechains be-
`
`N
`
`35A•
`
`A•
`
`ID
`
`SOA•
`
`RC
`
`~ G (kcal/moll
`12s•c, pH s-11
`
`N - Native conformation
`A* a Critically activated state
`ID - Incompletely disordered conformation
`RC - Random coil !fully denatured state)
`Figure 2-A schematic illustration of conformational transition for
`denaturatlon of protein (Kuwajima, 1977).
`
`Vol. 42, Supplement 1988
`
`S7
`
`-S7-
`
`
`
`
`

`

`tween a hydrocarbon phase and an aqueous phase. Mea(cid:173)
`suring the free energy of transfer of a side chain from the
`apolar interior to a solvent exterior, such as 8M urea,
`indicates that the change in stability of the globular struc(cid:173)
`ture caused by urea can be very great.
`. The ionic side chains of aspartic acid, glutamic acid,
`lysine, arginine, and histidine, usually on the surface of
`the molecule, contribute to the stability of the native
`structure by forming salt bridges. Again, the composition
`of the surrounding solvent will contribute either to stabili(cid:173)
`zation or to denaturation by virtue of its effects on these
`bonds. The pH of the solvent will determine the charge of
`the side groups and the extent of ionic bonding. Further,
`the dielectric constant of the solvent will influence the
`stength of salt bridges.
`
`d. Effect of Hydrophobic Interaction: Hydrophobic in(cid:173)
`teraction is often considered the driving force for den a tur(cid:173)
`a tion. When a protein molecule unfolds, and before it
`refolds to its natural conformation, hydrophobic interac(cid:173)
`tion can cause either of two different things to happen. In
`a concentrated protein solution, these hydrophobic groups
`interact with such groups from other molecules, resulting
`in protein aggregation. In dilute solutions of protein, how(cid:173)
`ever, intermolecular aggregation is unlikely; much more
`likely is an intramolecular hydrophobic interaction that
`results in a conformation different from the native one.
`Such an incorrect structure may show partial or complete
`loss of biological activity.
`
`Ill. Experimental Methods to Study Denaturation and
`Evaluate Stabilizers
`
`A. Cloud Point as a Model
`
`A number of techniques are available for studying the
`effect of such factors as heat, shear, surface phenomena,
`and solvent additions on the denaturation or stabilization
`of a protein's native globular active state. Horne (1971)
`describes the use of cloud points and synthetic polymers as
`model proteins. The cloud point is thought to arise from
`the dissolution of a molecule's hydration layer, exposing
`the hydrophobic character of its interior, and exclusion of
`the entire molecule (precipitation) from the aqueous sol(cid:173)
`vent. He studied polyvinyl methylether as a model poly(cid:173)
`mer and measured the effect of electrolytes on the stabil(cid:173)
`ity of the polymer in solution. Magnesium sulfate Lowered
`the cloud point markedly. The effect of alcohols, sugars,
`and urea was also measured. The author suggests some
`caution in using simple polymers as models for protein
`denaturation, but the effects of various factors on macro(cid:173)
`molecular stability in aqueous solution may be predictable
`by use of this system.
`
`B. Critical Mice/le Concentration as a Model
`
`Gratzer (I 969) studied the effect of protein denatur(cid:173)
`ants on breakage of hydrophobic bonds. One measure of a
`reagent's ability to destroy hydrophobic interactions is its
`ability to affect the critical micelle concentration (CMC)
`of a detergent. The UV absorption spectrum of a deter(cid:173)
`gent undergoes a perturbation upon formation of a mi(cid:173)
`celle. Plots of absorbance against volume of detergent
`
`solution added to a volume of solution showed a clear
`discontinuity at the CMC. The effect of agents such as
`sucrose or urea on this CMC were measured. Sucrose has
`no effect on CMC and is known to have no denaturing
`effect on proteins. However, urea and guanidinium salts,
`known to denature proteins, disrupt the formation of mi(cid:173)
`celles, and cause CMC to increase. Thus, in general, the
`effect of solutes on CMC seems to parallel their effect on
`denaturation of a protein by disruption of hydrophobic
`bonds.
`
`C. Charge-Transfer Absorption
`
`Coan (1975) described a method of looking at denatur(cid:173)
`ation by measuring charge-transfer~(CT) absorption of
`complexes between exposed tryptophan or tyrosine resi(cid:173)
`dues and l-alkyl-3-carbamido-pyridinium ions. A weak
`complex between the indole or phenol chromophores and
`the acceptor ion exhibit a broad CT absorption separated
`from the near UV absorption bands of the chromophore
`themselves. The authors studied denatured proteins with
`oxidized or reduced disulfide bonds and found that the
`aromatic residues are less exposed in denatured proteins
`with intact disulfide bonds. This finding suggests that
`these residues are in more structured, less random regions
`of the molecule and that the aromatic residues may func(cid:173)
`tion to act as refolding centers for renaturation of the
`molecule. Fully denatured proteins with no crosslinks con(cid:173)
`form to the full random coil model of denaturation.
`
`D. Thermal Analysis
`
`Ross (1974, 1984) described the use of a scanning mi(cid:173)
`crocalorimeter to measure energies of transition in solu(cid:173)
`tion: the energies of transition of 25-250 mJ of heat at
`temperature range from ambient to 90 °C. In the case of
`the gel-to-liquid transformation of dipalmitoyl-L-lecithin,
`a sharp transition occurred over three degrees and the
`energy of transition was in agreement with other reported
`values. The transition of poly(rA.rU) helix unwinding
`occurred over a nine-degree range and, again, the results
`agreed with reported values.
`Differential scanning calorimetry is gaining more wide(cid:173)
`spread use as a tool in investigating transitions of proteins
`as a function of temperature and, more importantly, the
`effect of potential stabilizing excipients on a protein solu(cid:173)
`tion. Yu (1984) used this technique to study the effects of
`stabilizers on human serum albumin. In the presence of
`stabilizers, transition temperature increased by several
`degrees (Fig. 3). By repeat scanning, one can determine
`whether the thermal denaturation is a reversible process.
`A stabilizer found effective in thermal denaturation may
`be effective for prolonging shelf-life at refrigeration tem(cid:173)
`peratures. Other research groups have also used thermal
`analysis as a tool in studying denaturation (Back, 1979;
`Gekko, 1982; Fujita, I 981; Uedaira, 1980).
`
`E. Fluorescence Spectroscopy
`
`A fluorescent probe can detect thermal denaturation
`and other conformational changes in proteins. A hydro(cid:173)
`phobic fluorescent compound, such as 8-anilino-1-naphth(cid:173)
`alenesulfonate, has the convenient property of being al-
`
`S8
`
`Journal of Parenteral Science & Technology
`
`-S8-
`
`
`
`
`

`

`4 mM c ap
`
`ciI 4 mM N- Ac-
`
`DL - trp +
`4 mM cap
`
`1/)
`I-
`
`I-<
`~
`...J
`...J
`
`:i:
`
`8 mM N- Ac -
`DL· t r p
`
`Albu mi n
`alone
`
`40 50 60 7 0 80 90
`TE MPERATUR E °C
`Figure 3-Thermograms of albumin with or without stabilizers: sodi(cid:173)
`um caprylate and sodium N-acetyl-OL-tryptophanate (Shrake, 1984).
`
`most nonfluorescent in aqueous solution, but fluorescing
`strongly when bound to hydrophobic sites on certain pro(cid:173)
`teins. As temperature increases to the point where thermal
`denaturation takes place, the fluorescence intensity in(cid:173)
`creases sharply, giving a well defined curve that resembles
`a thermogram in having a characteristic midpoint. Like
`thermal analysis, this technique provides a convenient
`method for evaluating potential stabilizers in terms of
`their ability to shift the mid-point to higher temperatures.
`Busby utilized this technique to examine the stabiliza(cid:173)
`tion of antithrombin Ill by heparin and lyotropic anions
`( 1981 ), and by sugar derivatives (1984) . Two irreversible
`thermal transitions were originated from two different
`domains on human complement (1987). A typical melt
`curve, as constructed by fluorescent intensity and tem(cid:173)
`perature is shown in Figure 4.
`
`F. Others
`
`O ther than the aforementioned methods, the following
`experimental techniques are useful in detecting denatur(cid:173)
`ation: UV (Hagerman and Baldwin, 1976), circular di-
`
`w
`() >-
`ffi ~ 100
`g .D
`w <
`a:
`0
`=>
`...J u.
`
`50
`
`0 .____.,_...__.____.,_,
`40 50 60 70 80
`TEMPERATURE (°C)
`Figure 4- Effect of sodium gluconate on thermal denaturation of
`antithrombln Ill. Protein, 1 mg/mL: fluorescent probe, 66 ,uM; and
`gluconate, 0(0 ), 1.0(A), 1.5((cid:143) ), and 2.0M(e ), respectively (Busby,
`1984).
`
`chroism (Galat, 1985), MR (Brems, I 984), size-exclu(cid:173)
`sion chromatography (Light, 1987), and gel electrophore(cid:173)
`sis (Goldenberg, 1984).
`
`IV. Use of Excipients to Stabilize Parenteral
`Formulations of Proteins and Peptides
`
`A search of the literature for methods to prepare formu(cid:173)
`lations of therapeutically useful proteins and peptides
`and, specifically, how proteins and peptides can be stabi(cid:173)
`lized, makes it apparent that this is an area in which trial
`and error plays a major role. Various types of molecules,
`such as sugars, amino acids, surfactants, and fatty acids,
`have served to stabilize protein and peptide products
`against degradation. In each class of stabilizer described
`below, an explanation of how stabilization was observed
`and what types of protein and peptide were stabilized is
`presented. Postulated mechanisms and typical concentra(cid:173)
`tions used in parenteral products are also described. Only
`additives for parenteral formulations are of concern in this
`review; other means of stabilizing protein, such as modifi(cid:173)
`cation of the primary peptide sequence, immobilization of
`proteins, and grafting protein to polymers, to name a few,
`are not discussed.
`
`A . Serum Albumin
`
`Serum albumin, regardless of its origin (rabbit, bovine,
`or human), has been extensively cited in patents (Table I)
`and literature (Table Ia) as a stabilizer for enzymes and
`other proteinaceous material. O ne of many reasons for the
`choice of albumin over other proteins is its stability. Albu(cid:173)
`min, even without stabilizers, can withstand heating to 60
`°C for IO hours. At pH 1- 2, the albumin molecule ex(cid:173)
`pands and elongates; it can return to its native configura(cid:173)
`tion reversibly (Peters, 1985). Also, albumin bas good
`solubility; unlike many other proteins, it is solu ble in con(cid:173)
`centrated salt solution; unlike globulin, it is soluble at the
`isoelectric point; un like protamine, it is soluble in diluted
`base.
`The mechanisms by which al bumin acts as a stabilizer
`can be any of the following:
`a. Inhibition of Surfa ce Adsorption: Although, in a
`strict sense, this is not a stabilization phenomenon, in
`many instances it is the reason why an albumin-contain(cid:173)
`ing preparation shows higher recovered enzyme activity
`than one lacking albumin. Numerous studies l\_avedemon(cid:173)
`strated that albumin and similar proteins effectively de(cid:173)
`crease the loss of insulin to the plastic bags or tubing used
`in intravenous devices and to siliconized glassware
`(Wang, I 984) . A recent patent application showed that
`adsorption of erythropoietin can be retarded by albumin
`at concentrations as low as 0.003% (CA l04:230,446V).
`In contrast, an increased loss of thyroid hormone on a
`filter was observed at high concentrations (2.5- 5%) of
`albumin (Law, 1983). These results should not be a sur(cid:173)
`prise, because albumin binds various molecules, including
`peptides, while it acts as a carrier in the blood.
`b. Substitution for a Nascent Complexing Protein:
`Wolf (J 972) postulated that in the cell, the binding of
`adenylyltransferase to another protein has a stabilizing
`
`Vol. 42, Supplement 1988
`
`S9
`
`-S9-
`
`
`
`
`

`

`~
`0
`
`TABLE I. Patents Describing Albumin as a Stabilizer
`
`Stabilizer
`
`HSA
`HSA
`BSA
`Hydrolyzed
`ovalbumin
`HSA
`HSA
`HSA
`
`Drug
`
`interleukin-2
`plasminogen activator
`uricase
`theophylline antibody
`
`tumor necrosis factor
`erythropoietin
`')'-interferon
`
`Concn.
`(%)
`
`I
`0.1
`I
`
`0.1
`
`0.1-1
`
`rreeze-dried
`+ glycine, + sugar, freeze-dried
`
`Conditions
`+ glutathione, pH 4, freeze-dried
`phosphate buffer, pH 7
`+ surfactant, + arginine, freeze-dried
`increased thermal stability
`
`Inventor/ Assignee
`
`Reference•
`
`Y. Mikura/Takeda
`A. Hasegawa/Asahi
`