Protein drug stability: a formulation challenge. Nat Rev Drug Discov 4:298-306

Abstract

The increasing use of recombinantly expressed therapeutic proteins in the pharmaceutical industry has highlighted issues such as their stability during long-term storage and means of efficacious delivery that avoid adverse immunogenic side effects. Controlled chemical modifications, such as substitutions, acylation and PEGylation, have fulfilled some but not all of their promises, while hydrogels and lipid-based formulations could well be developed into generic delivery systems. Strategies to curb the aggregation and misfolding of proteins during storage are likely to benefit from the recent surge of interest in protein fibrillation. This might in turn lead to generally accepted guidelines and tests to avoid unforeseen adverse effects in drug delivery.

Full text

29 8 |AP RIL 20 05 |VOLUM E 4 www.nat ure.com/reviews/drug disc
R E V I E W S
During the past two decades,recombinant DNA tech-
nology has led to a significant increase in the number
of approved biotechnology medicines and a shift away
from the production of biologically active materials
(biologics) on the basis of animal or human material,
and towards cloning and fermentation.A recent survey
lists 324 biotechnology medicines, either in human
clinical trials or under review by regulatory agencies.
These drugs cover nearly 150 diseases,including cancer,
infectious diseases, autoimmune diseases and AIDS/
HIV1. A substantial number of these substances are
proteins.This development will expand the list of medi-
cines and open the way for more and better treatments
to the benefit of patients.
This general trend accords with the vision of several
analysts, including IBM Business Consulting Services2.
According to this view,many of the new medicines will
be based on the recombinant DNA expression of pro-
teins rather than organic chemistry,because biologics in
general are expected to be less toxic and to behave more
predictably in vivo. Biotechnological medicines could
therefore potentially reach the market faster than chemi-
cal entities developed through traditional methods.On
the other hand, the decoding of the human genomic
and proteomic maps is anticipated to lead to discoveries
in molecular medicine that could revitalize the develop-
ment of small-molecular synthetic drugs3,4. In many
respects,the efficacy and safety requirements ofbiologics
are similar to the requirements for small, chemically
synthesized drugs. However, owing to the biological
origin and macromolecular structure of biologics, there
is particular focus on contamination of biologics with
other biological impurities, such as viruses, as well as
conformational changes introduced either during pro-
duction of the bulk substance or the final formulation.
Well-documented and validated biological, physical and
chemical methods are important tools for securing the
quality and safety of biologics.
Several challenges confront pharmaceutical scientists
involved in the development of biotechnological medi-
cines, such as proteins. The successful formulation of
proteins depends on a thorough understanding of their
physico-chemical and biological characteristics,includ-
ing chemical and physical stability, immunogenicity
and pharmacokinetic properties.The therapeutic activity
of proteins is highly dependent on their conformational
structure.However, the protein structure is flexible and
sensitive to external conditions,which means that pro-
duction, formulation and handling of proteins needs
special attention in optimizing efficacy and safety,
including minimized immune responses.The primary
focus of this review is on the challenges related to qual-
ity and safety issues that arise in the development of
protein-based medicines.
The chemical and physical stability of proteins can be
compromised by external factors such as pH,temperature
PROTEIN DRUG STABILITY:
A FORMULATION CHALLENGE
Sven Frokjaer* and Daniel E.Otzen‡
Abstract | The increasing use of recombinantly expressed therapeutic proteins in the pharma-
ceutical industry has highlighted issues such as their stability during long-term storage and means
of efficacious delivery that avoid adverse immunogenic side effects. Controlled chemical
modifications, such as substitutions, acylation and PEGylation, have fulfilled some but not all of
their promises, while hydrogels and lipid-based formulations could well be developed into generic
delivery systems. Strategies to curb the aggregation and misfolding of proteins during storage are
likely to benefit from the recent surge of interest in protein fibrillation. This might in turn lead to
generally accepted guidelines and tests to avoid unforeseen adverse effects in drug delivery.
*Department of
Pharmaceutics,
The Danish University of
Pharmaceutical Sciences,
Universitetsparken 2,
DK-2100 Copenhagen O,
Denmark.
‡Department ofLife
Sciences, Aalborg University,
Sohngaardsholmsvej 49,
DK-9000 Aalborg,
Denmark.
Correspondence to D.E.O.
e-mail: dao@bio.aau.dk
doi:10.1038/nrd1695

NATU RE REV IEWS |D RUG DI SCO VERY VOLUME 4 |APR IL 200 5 |299
R E V I E W S
PARENTERAL
A substance that is introduced
into the body anyway except
by mouth.
PHARMACOKINETIC
The study of the absorption,
distribution,metabolism,
excretion and interactions
of a drug.
PEGylation
The covalent binding of
polyethylene glycol (PEG)
to a protein.
Protein analogues. The development of rapid-acting
monomeric insulin analogues provides an excellent
example of the analogue approach12. The basic idea
behind this development was that the transport rate
from the subcutaneous injection site of administration
across the biological membrane to the systemic circu-
lation would be increased if the self-association charac-
teristics of insulin were shifted from hexameric insulin
to di- or monomeric insulin (FIG.2). By mutating amino
acids involved in self-association, this approach was
proven to be feasible. Monomeric insulin analogues
with a more rapid time-action profile than human
insulin are now on the market (for example, insulin
lispro (Eli Lilly) and insulin aspart (NovoRapid; Novo
Nordisk)).
Recombinant human interleukin-2 (IL-2) is another
example in which the therapeutic effect is improved
through molecular design.The commercially available
non-glycosylated IL-2 analogue aldesleukin (Proleukin;
Chiron) is a des-alanyl-1 analogue of human IL-2 in
which cysteine in position 125 has been replaced by
serine. The experimental IL-2 analogue BAY 50-4798
(Bayer) is another example of a genetically engineered
protein with potentially improved therapeutic effect.
Acylation. Chemical attachment of fatty acids to exposed
residues on the protein surface can in some cases increase
the affinity of the protein to serum albumin sufficiently
to increase its circulation time in the blood13,14.The effect
of acylation is naturally greater for smaller proteins and
peptides because of the greater relative increase in
hydrophobicity. For example, the acylated insulin
analogue [Lys.sup.B29]-tetradecanoyl des (B30) human
insulin, insulin detemir (Novo Nordisk A/S) is long-
acting and has recently been approved by the regulatory
agencies in some countries.The principle of acylation
has also been applied successfully to other proteins —
for example, glucagon-like peptide 1 (GLP1) (REF. 15)
and interferon-α16,as well as for peptides such as desmo-
pressin17.Whereas the acylation of insulin and GLP1 is
site-specific, the acylation of interferon is less specific.
This potential heterogeneity of acylated interferon could
eventually prove to be a problem both from an efficacy
and a safety point of view.
and surface interaction, as well as by contaminants
and impurities for excipients, and so on5,6. Several
recent reviews discuss the importance of protein stability
in formulation development and the immunogenicity
of proteins from a pharmaceutical perspective5,7. In
addition, several recent reviews focus on conditions
leading to undesirable changes in the chemical structure
of proteins8,9.
Delivery challenges
Oral administration of medicines is the preferred and
most widely used route of administration.However,this
route is generally not feasible for the delivery of macro-
molecules such as proteins.The inherent instability of
proteins in the gastro-intestinal tract,as well as the low
permeability across biological membranes due to the
high molecular mass and polar surface characteristics of
proteins, implies that proteins for systemic treatment
should be administered PARENTERALLY;however,efforts are
being made to improve bioavailability through alterna-
tive routes of administration,for instance,by the nasal
or pulmonary route10,1 1. Nevertheless, bioavailability
through the various non-parenteral routes of adminis-
tration is low and normally insufficient for an effective
systemic effect.The obstacles for efficient delivery to the
site of action can broadly be categorized as either vari-
ous enzymatic barriers that the protein encounters in
moving from the administration site to the site of
action,or physical barriers to effective transport, such as
epi- and endothelial cell linings (FIG.1).
The pharmaceutical and PHARMACOKINETIC properties
of proteins can be optimized by different approaches —
for example,by mutagenesis,chemical modification or
by designing specific drug-delivery systems.However,
most protein-based medicines today are still formulated
as suspensions or aqueous solutions either in a ready-to-
use form or as a lyophilized product for reconstitution.
Substitution and chemical modification
Proteins for therapeutic use can be chemically modi-
fied in several ways — for example,by mutating one or
more amino acids (that is,creating a protein analogue)
or by acylation or PEGylation. These modifications can be
used to optimize the pharmacokinetic properties of
the protein, but care must always be taken not to
reduce their biological efficacy.
Cell barrier
Site
of
action
Enzymes
Protein
Release from
dosage form
Figure 1 | Barriers to biomacromolecular drug transport. The major obstacles for efficient
transport to the site of action are enzymatic degradation and biological membranes.
Hexamer
Biological membrane
Dimer Monomer
Figure 2 | Transport of insulin across biological membranes.
Insulin monomer has the highest flux.

30 0 |AP RIL 20 05 |VOLUM E 4 www.nat ure.com/reviews/drug disc
R E V I E W S
harmful manufacturing conditions appropriate for
sensitive drug molecules such as proteins26.
Examples of lipid-based formulations with the
potential for serving as delivery systems for macromole-
cules such as proteins include liposomes, solid lipid
nanospheres and water-in-oil emulsions27–29. The incor-
poration and release of proteins from liposomes, for
instance, can be controlled by the physico-chemical
characteristics of the protein — that is, the hydrophobic
surface area, which can be modified, for example, by
acylation30 or by constructing lipid bilayers that are
sensitive to triggered degradation by enzymes such as
secretory phospholipase A2 that are present in inflam-
matory tissues and certain cancerous fluids31 . Rep-
resentative examples of modifications and delivery
systems are given in TABLE 1.
Immunogenicity of therapeutic proteins. The ability to
produce highly purified proteins that are identical or
nearly identical to endogenous human proteins is often
considered as the key to avoiding T- and B-lymphocyte
reactivity during treatment,because patients are expected
to be immune tolerant to their own proteins. However,
although the level and incidence of immune responses is
low, most of these proteins have been shown to be
immunogenic,and in some cases have even led to serious
safety problems and inhibition of the therapeutic effect
due to neutralizing antibodies.An example is antibody
formation after long-term treatment with recombinant
interferons in patients with multiple sclerosis,which indi-
cates that the presence of neutralizing antibodies against
interferons reduces the clinical effect of the drug32,33.
Recombinant human erythropoietin is another example
in which the neutralizing antibodies react against ery-
thropoietin and cause pure red-cell aplasia34.It has been
suggested that the reason for anti-erythropoietin anti-
body production seen in patients might be related to the
production process of the hormone35. A formulation
change introduced in the late-1990s has also been pro-
posed as a plausible reason for the induction of anti-
bodies that neutralize the endogenous erythropoietin.
A recent study indicates that this formulation contains
micelle-associated erythropoietin,which could be a risk
factor for the development of antibodies36.The difficul-
ties in detecting structural changes in proteins in more
complex pharmaceutical formulations is a major chal-
lenge for formulation scientists responsible for the devel-
opment and quality of biopharmaceutical medicines37.
In general,it is the protein structure that can trigger
an immune response during treatment. Proteins are
complex and flexible molecules, and even a small
change at a particular site can result in a major change
in the overall structure (TABLE 2).For instance, recombi-
nant mutated allergens with a modified surface topog-
raphy, but that retain an α-carbon-backbone folding
pattern, have been shown to retain the capacity to
induce allergen-specific immune responses, but with
different anaphylactic potential38. This shows that
although the surface structures positioned outside the
mutated regions are retained, the introduction of muta-
tions that change both the topography and the charge
PEGylation. In general, PEGylation reduces the
plasma clearance rate by reducing the metabolic
degradation and receptor-mediated uptake of the
protein from the systemic circulation. PEGylation also
improves the safety profile of the protein by shielding
antigenic and immunogenic EPITOPES18. However,it is
important to realize, both from an efficacy and safety
point of view,that PEG polymers consist of a mixture
of polymers with different molecular mass; in addi-
tion, larger proteins will have several sites that are
accessible to PEGylation. Owing to this potential
product heterogeneity,the clinical and pharmaceutical
documentation required for drug approval could be
even more demanding than for non-modified pro-
teins. PEG-interferon-αis a commercial example in
which the pharmacokinetic profile is improved by
PEGylation, which enables less frequent administra-
tion and results in improved efficacy with a similar
side-effect profile19. On the other hand, mono-
PEGylated epidermal growth factor (EGF) with PEG
3400 at Lys28 and Lys48 was significantly less active
than an EGF isomer PEGylated at the amino terminus
in an in vitro assay for mitogenic activity20. Several
excellent reviews are available on these specific types
of derivative.A recent issue of Advanced Drug Delivery
Reviews is devoted to PEGylation21.
Drug-delivery systems. Entrapment and encapsula-
tion are the most widely used formulation principles
adopted for protein-delivery systems.Polymeric drug-
delivery systems,such as hydrogels,nanocapsules and
microspheres, and lipid-based drug-delivery systems
such as liposomes and solid lipid nanoparticles, are all
examples of protein-delivery systems (FIG.3).
Hydrogels are crosslinked hydrophilic polymers
that form three-dimensional networks, which can
either be synthesized to degrade at a certain rate or to
respond to several physiological stimuli present in the
body, such as pH, ionic strength and temperature.
Both principles are used for the controlled release of
proteins22–25. These hydrogels can be formulated either
as microspheres or as in situ forming delivery systems.
The in situforming systems could have several advan-
tages over conventional microspheres, such as ease of
administration and less complicated and potentially
EPITOPE
An alternative term for an
antigenic determinant. These
are particular chemical groups
on a molecule that elicit a
specific immune response.
Change in
external
conditions
Polymer
Protein
Figure 3 | Schematic illustration of protein release from a polymer sensitive to external
stimuli, such as change in pH, ionic strength and temperature. Swelling of the polymer
releases the entrapped protein.

NATU RE REV IEWS |D RUG DI SCO VERY VOLUME 4 |APR IL 200 5 |301
R E V I E W S
rate limiting, aggregation kinetics can be pseudo-first
order47,48, whereas there is an inverse dependence for
processes involving air–water interfaces49. Initial dimer
formation is seen with colony stimulating factor50,
whereas lysozyme aggregation follows higher-order
kinetics51. Aggregation often shows high activation
barriers, which make the process slow, irreversible and
kinetically controlled.This means that the handling his-
tory of a sample can have a disproportionate role in the
outcome.Irreversible aggregation,caused by disulphide-
shuffling or stable hydrophobic association,can have a
direct impact on drug potency,immunogenicity47 and
the unfolded protein response52,53.Reversible aggrega-
tion, however, can also be problematic during drug
administration,ifdissociation is slow on the physiological
time scale. Slow dissociation kinetics can be a conse-
quence of crowding effects in the body, which favour the
aggregated state even after dilution54,55.
The protein chain is chemically complex and can
assume an astronomical number of possible conforma-
tions, the relative stabilities of which are very sensitive to
pH,ionic strength and temperature.This creates a large
number of ‘conformational end stations’— including
both those that are monomeric and oligomeric — that
depend on environmental conditions (FIG. 5).The aggre-
gates are not just featureless heaps;it has been known for
a long time that aggregation is generally quite specific56,
involving well-defined oligomerization interfaces.
Probably the most important — and certainly most
well-characterized — aggregation state is the amyloid
fibril,associated with neurodegenerative diseases such as
Alzheimer’s disease,Parkinson’s disease,Huntington’s
disease and many others. Curiously, fibrillation is
exploited in a wide range of organisms, ranging from
biofilm in Escherichia coli57 to melanosome formation in
humans5 8. It has been suggested that all proteins can
form fibrils with the same structural characteristics,
namely a cross-βstructure and parallel β-helices59, irre-
spective of the native structure and primary sequence60.
This presents a potential hazard during production
and/or storage.Fibrillation is particularly encouraged
under moderately destabilizing conditions,such as pH
below 3 or above 10, temperatures above 40o C and/or
intermediate denaturant concentrations, which allow
the protein better access to other conformations while
retaining some structure.
distribution can be sufficient to affect the immuno-
genicity of the protein (FIG. 4). However,it is normally
difficult to relate a particular change in protein structure
to a change in immunogenicity.
Two is a crowd: coping with aggregation
During processing and formulation of the drug product,
the protein is exposed to conditions that could have
significant effects on its chemical and physical stability,
and lead to aggregation and ultimately precipitation9,39–41.
It is therefore important to understand the circumstances
by which protein stability is compromised. The main
factors are shear/shaking, temperature,pH and protein
concentration (see REF.39 for a detailed overview).Shear
forces encountered by vortexing can partition proteins
to the air–water interface, which encourages partial
unfolding on exposure to the more hydrophobic air
phase41,42. This will be even more pronounced if the
second phase involves an organic solvent such as chlo-
roform4 3. Elevated temperatures44 , changes in pH or
intermediate denaturant conditions45 can also favour
the formation of such states. Small, single-domain
proteins usually require extreme conditions to unfold,
but for large, multidomain proteins, a few ‘weak links’
unravelling under relatively gentle conditions can be
sufficient to initiate aggregation. Partially unfolded
states are much more susceptible to aggregation than
the native or unfolded state,due to the exposure of con-
tiguous hydrophobic regions that are buried in the
native state or absent in the denatured state40,46.
Although aggregation is fundamentally bi-molecular,
the dependence on concentration will reflect the mecha-
nism of the process. If unfolding of the native state is
Table 1 | Modified proteins and protein-delivery systems approved for marketing
Product (company) Drug Modification/delivery system Administration route
Proleukin (Chiron) Aldesleukin Analogue Intravenous
Humalog (Eli Lilly) Insulin lispro Analogue Subcutaneous
NovoRapid (Novo Nordisk) Insulin aspart Analogue Subcutaneous
Neulasta (Amgen) PEGinterferon α-2a Mono-pegylated Subcutaneous
Pegasys (Roche) Pegfilgrastim Mono-pegylated Subcutaneous
Somavert (Pharmacia) Pegvisomat Multi (4–6)-pegylated Subcutaneous
Levemir (Novo Nordisk) Insulin detemir Mono-acylated Subcutaneous
Nutropin Depot (Genentech) Human growth hormone PLGA microspheres Subcutaneous
InductOs (Wyeth (MDT)) Bone morphogenic protein 2 Absorbable collagen sponge Implanatable medical device
PLGA, poly(lactide-co-glycolic acid)
Table 2 | Potentially immunogenic protein modifications
Modification Effect
Engineered modifications
Amino-acid sequence Human versus analogues and non-human proteins
Chemical modification Acylation, PEGylation
Pharmaceutical formulation Lyophilization, micro-encapuslation
Unwanted modifications during processing, production and storage
Chemical degradation Deamidation, oxidation
Physical degradation Denaturation, aggregation, fibrillation, misfolding
PEG, polyethylene glycol.

30 2 |AP RIL 20 05 |VOLUM E 4 www.nat ure.com/reviews/drug disc
R E V I E W S
detection in the lag phase.For polyglutamine sequences
involved in Huntington’s disease,the nucleus is actually a
monomer79,revealing that the rate-limiting step is not the
assembly of an oligomeric species but a conformational
change occurring at the monomer level.
In cases in which seeding has been shown to accelerate
the process, such as for the Aβpeptide80,it might provide
some relief to the biotechnological community to know
that spurious fibrillar contaminants of another protein
need not wreak havoc.The seeding process is very specific
and demands very close sequence81 and enantiomer82
correspondence between the soluble monomer and the
fibril;even point mutations can completely abolish seed-
ing ability83. Nevertheless, it should also be noted that
fibrillating proteins can form fibrils that differ signifi-
cantly at the molecular level,simply by varying growth
conditions such as stirring/aggregation84, temperature or
ionic strength (J.S. Pedersen and D.E.O.,unpublished
observations), and these differences can be propagated
from one generation to the next. This polymorphism is
Fibrillation is generally modelled as a nucleation-
dependent process, in which the nuclei accumulate
during the lag phase without any bulk conformational
changes, followed by the rapid accumulation of fibril-
lated protein as the nuclei are extended61.The lag phase
is typically very long,mainly because the higher-order
equilibria leading to nuclei are unfavourable at low pro-
tein concentrations. This makes the process very sensitive
to protein concentration; for sickle-cell haemoglobin
fibrillation, concentration dependencies of up to 50
have been predicted and observed62.Simulating aggre-
gation in silico63,64 or setting up mathematical models65
can be useful to predict and prevent problems in protein
production,but it remains challenging to incorporate all
the environmental factors that impinge on the process.
Aggregation is fundamentally a deviation from the
‘productive’folding pathway,and it has been proposed
that the individual energy landscapes or funnels describ-
ing protein folding should really be represented as ‘double
funnels’, the second of which represents intermolecular
interactions ofpartially folded states (FIG.6)66.
Intermolecular β-strand formation can be a powerful
driving force for aggregation,and the propensity to form
β-strands generally correlates positively with aggrega-
tion, in contrast to that of α-helices67. However,it has
recently been suggested that α-helices can contribute to
aggregation through the formation of coiled-coil con-
tacts68, although this is unlikely to be a general mecha-
nism.It is now possible to predict absolute aggregation
rates and the effect ofdifferent mutations on aggregation
using simple physico-chemical approaches69–71,such as
changes in secondary structure propensities,hydropho-
bicity,patterns of alternating hydrophobic–hydrophilic
residues,pH, ionic strength, concentration,and so on.
For example, the sequential location of hydrophobic
amino acids controls fibrillation of the amyloid β(Aβ)
peptide72 and insulin73. Aggregation propensities can
even be predicted at the level of individual residues to
pinpoint aggregation hot spots69. Key ‘gatekeeper’
residues are sometimes involved in stabilizing/destabi-
lizing such interactions,so a few mutations can easily tip
the balance towards the monomeric form74,75. However, it
is characteristic of the complexity of the reactions leading
to aggregation that a simple parameter such as stirring
has yet to be incorporated into these models.
The lag time is probably the most important parame-
ter from a formulation point of view.Unfortunately,it is
very difficult to predict,not only because of the stochastic
nature of the nucleation processes that can give rise to a
lag phase,but also because of the complexity of the mech-
anism involved.It is common to assume that lag times
represent a nucleation process61,but this is not always the
case76.Lag phases can also be observed even with a supply
of preformed fibril seeds76 (J.S. Pedersen and D.E.O.,
unpublished observations).In such cases,fibrillation is
more likely to be promoted by small fibrils fragmenting
to smaller entities that can capture more monomers
before they fragment again,leading to the exponential
growth of monomer-ensnaring fibrils77,78. Here,the lag
phase does not have a simple physical interpretation,but
simply reflects that the amount ofamyloid is too low for
(N28,K32)
(N28,K32)
E45 (T28,Q32)
(T28,Q32)
(N28,K32) (T28,Q32)
P108 G108
S45
Figure 4 | Protein structure and immunogenicity.
Comparison of molecular surfaces coloured according to
electrostatic potential. Introduction of amino-acid mutations
change both the topography and the polarity locally,
whereas surface structures outside the mutation sites are
retained. Adapted, with permission, from REF. 38 © (2004)
American Association of Immunologists.

NATU RE REV IEWS |D RUG DI SCO VERY VOLUME 4 |APR IL 200 5 |303
R E V I E W S
Because of the grave consequences of protein fibrilla-
tion in neurodegenerative diseases, the prevention of
aggregation in vivo is a subject of intense research93.Many
of the medically relevant fibrillating proteins have been
subjected to extensive screening procedures,resulting in
promising small-molecule leads for α-synuclein94 and
tau95,for example.Fragments of the fibrillating peptide,
which can bind to the growing edges of the fibrils by
hydrogen bonding but prevent further elongation due to
their small size (or by the inclusion of‘β-breaker’residues
such as proline), also have an effect96.A combination of
these two approaches has led to the use of small molecules
with hydrogen-bonding patterns complementary to the
exposed β-strand,but not to new incoming β-strands97.
An even more ingenious ‘Trojan horse’approach has been
to couple the fibril-binding dye to a peptide,which,when
in the cell, binds to the FK506-binding protein (FKBP)
chaperone,thereby increasing its steric bulk and blocking
further fibrillation98.Inhibiting aggregation under appli-
cation-relevant conditions in vitrois obviously also very
important.Mutagenesis to reduce aggregation is problem-
atic because it can entail extensive clinical trials to docu-
ment lack of other adverse effects.A more direct approach,
therefore, is to alter the formulation conditions of the
protein,but this is also challenging.Many therapeutic pro-
teins are required in high doses,administered as part of
frequent-dosing regimens,which means that it is impor-
tant to keep them stably dissolved for extended periods of
time at concentrations of tens of mg per ml. Determing
the solubility of a protein is a complex task,because of the
propensity of proteins to interact with themselves,surfaces
and solutes.Of the many techniques used for concentrat-
ing proteins, none are without problems99,and most are
only appropriate at certain concentrations of protein.
At present, the prevention of aggregation remains
largely empirical, due to a lack ofinsight into the molec-
ular details of the aggregation process (see REF. 39 for an
excellent overview of different approaches).One popular
approach is to stabilize the protein and thereby reduce
access to partially folded conformations favouring aggre-
gation by hydrophobic contacts,for example.A typical
strategy is to add sugars or salts to a protein solution.
These solutes are thought to be preferentially excluded
from the surface of the protein, therefore favouring a
compact state100,101. However, because aggregation
buries even more surface area per protein molecule,
over-stabilization can ultimately lead to aggregation.
Other stabilizers include polyols, PEGs and other poly-
mers that sterically hinder protein–protein interactions
and limit diffusion.Free amino acids are also often used;
arginine is particularly good at preventing aggregation
during the refolding of proteins from inclusion bodies102.
A more sophisticated approach is to add 50 mM each of
arginine and glutamate,which leads to a marked increase
in the long-term stability of the sample,and also prevents
aggregation, precipitation and proteolysis103.It has been
proposed that the basis for this remarkable effect lies in
the capacity of the pair to neutralize opposite charges
(which would otherwise lead to intermolecular associ-
ation) combined with the ability to cover adjoining
hydrophobic areas through aliphatic tails103.
to be expected,simply because there has not been evo-
lutionary pressure to evolve one particular kind of fibril,
and fibrillogenic proteins are typically flexible, giving
access to many conformations.Calorimetric85 and struc-
tural86 studies also indicate that fibrils are more porous
than conventional globular proteins.
Peptides are more of a liability than proteins because
of their greater flexibility and access to fibrillogenic con-
formations.The peptide hormone glucagon,which does
not fibrillate in the body,can nevertheless fibrillate fairly
easily if mishandled by,for example, prolonged storage
at 2.5 mg per ml or higher at 37o C87.It is possible to select
for less aggregation-prone peptide variants by fusing the
peptide in question to green fluorescent protein (GFP),
which does not show fluorescence when it accumulates in
inclusion bodies88 and thereby indicates loss of biological
structure and function.
Formulating longevity: additives and storage
Recent work has revealed a strong correlation between
different types of kinetic stability,as evaluated by unfold-
ing in denaturant conditions and detergents,as well as by
proteolysis resistance89,and has found oligomeric β-sheet
proteins to be particularly robust. These proteins are
more rigid than their α-helix or mixed α/βcounterparts.
Analogously, proteins become both rigid and kinetically
stable in very high concentrations of organic solvent90,91,
probably because the lack of bulk water acting as a lubri-
cant reduces the energetic driving force for partial and
global unfolding.Kinetic stability can be increased by
introducing hydrophobic mutations,disulphide bonds,
salt bridges and metal ions at the protein surface to sta-
bilize and rigidify regions involved in local unfolding92.
Amyloid fibril
Prefibrillar speciesDegraded fragments
Intermediate NativeUnfoldedSynthesis
Disordered aggregate Oligomer Fibre
Crystal
Disordered aggregate
Disordered aggregate
Figure 5 | The many conformational choices for a polypeptide chain. Adapted, with
permission, from REF.138 © (2003) Macmillan Magazines Ltd.

30 4 |AP RIL 20 05 |VOLUM E 4 www.nat ure.com/reviews/drug disc
R E V I E W S
to seven glucose rings113,although larger polymers also
occur114.Cyclodextrins suppress aggregation of therapeu-
tically relevant proteins,such as insulin115–117 and growth
hormone118,119, as well as several other proteins120,121 .
This derives from their ability to bind to aromatic
residues122, which can lead to preferential stabilization
of the unfolded state123 and a reduction in folding rates124.
The ability to suppress aggregation has also led to their
use as simple chaperones for the refolding of proteins
in combination with SDS10 7,125. The commonly used
β-cyclodextrin is not approved for human consump-
tion in an unmodified form because it is capable of
extracting biomembrane components126, and forms a
highly insoluble complex with cholesterol,which could
lead to toxic effects after prolonged exposure127.However,
careful derivatization ofβ-cyclodextrin alters its selectiv-
ity towards biomembrane components and mini-
mizes its toxicity127. Drug formulations containing
hydroxypropyl-β-cyclodextrin (for the peptides leucine
enkephelin and a neuromedinB-receptor antagonist)
and sulphobutylether-β-cyclodextrin (for the small-
molecule drugs ziprasidone (Geodon; Pfizer) and
voriconazole (Vfend; Pfizer) have been approved for
parenteral drug administration.
Lyophilization. Storing proteins in solution at low
temperatures generally extends their shelf-life. Proteins
can be denatured at low temperatures in the liquid phase,
but, in general, cold denaturation is reversible,unlike
most high-temperature denaturation,and so offers no
practical problems128. Freezing obviously provides
access to even lower temperatures,but repeated cycles of
freeze–thawing can strongly stimulate aggregation by
providing nucleation surfaces at the ice–water inter-
faces105, as well as leading to pH changes and phase sepa-
ration129.Another approach is therefore to restrict protein
mobility by lyophilization or freeze-drying,for example.
Freeze-drying is the separation of liquid water from a
solution frozen to ice by vacuum sublimation,leaving the
solutes in an anhydrous or almost anhydrous state. As
described by Franks130,131,the process is not just trial-and-
error,but can be controlled by well-established physico-
chemical and engineering principles,taking into account
parameters such as the composition and concentration of
the product,the type of container, the equipment and
process cycle (including primary and secondary drying).
The method is not without challenges,because differen-
tial precipitation ofbuffers or other solvent compositions
during freezing can lead to pH changes that irreversibly
inactivate proteins.This can be counteracted by excipi-
ents, such as certain carbohydrates,soluble polymers,
salts and volatile compounds, which form glasses during
freezing,which in effect preserves liquid properties in the
solid state and prevents unwanted crystallization and
chemical side reactions. However, their effect on the
glassing temperature must also be taken into account.
Temperature and chamber pressure both affect subli-
mation rates and can be combined judiciously to avoid
overheating and structural collapse of the lyophilized
solute network. Assessing the stability of the freeze-
dried product is not straightforward;accelerated storage
Detergents and other amphiphiles. Non-ionic detergents
are often used to reduce the effect of shear,as they out-
compete proteins at the air–water interface104,but this can
sometimes be counterbalanced by accelerated aggrega-
tion under long-term motionless shelf storage105. They
are sometimes106,but not always107,able to prevent aggre-
gation induced by heat. Ionic surfactants are generally
denaturing, but reversible unfolding has been used to
prevent aggregation of heat-denatured RNase108. They
tend to stabilize α-helical conformations that do not
form intermolecular contacts with the same alacrity as
β-strands.However,just as the water–chloroform inter-
face can be a powerful stimulant of aggregation43,recent
work indicates that anionic surfaces encourage protein
aggregation and fibrillation — for example, through
negatively charged phospholipid vesicles109,110, anionic
surfactants111 and carboxylate-modified polystyrene
beads112. The phospholipid environment encountered
in vivo is chemically diverse,providing an opportunity for
both hydrophobic interactions in the transmembrane
region as well as polar and electrostatic contacts through
the head group oflipids in the membrane.It has been sug-
gested that the lipid surface acts to concentrate and align
protein molecules,possibly through clusters of cationic
residues on the protein.Another mechanism could be for
the negative surface to stabilize a fibrillogenic monomer
conformation,which through intermolecular interactions
forms the nucleus for subsequent fibrillation112.
Cyclodextrins. A class of substances that has been found
to have significant potential in reducing aggregation is the
cyclodextrins. These are circular polymers of typically five
Energy
Aggregation
Intermediate
Aggregate Native state
Folding
Unfolded states
Figure 6 | The double funnel of folding and aggregation. Folding is a journey through
conformational space, whose architecture is sensitive to environmental conditions, including
protein concentration. Whether the protein chooses the folding or aggregation funnel will depend
on these conditions. Note that the aggregation funnel is less jagged than the native funnel,
emphasizing that there is space for more ‘conformational largesse’ due to the lower restrictions
on aggregate conformations. Adapted, with permission, from REF.66 © (2004) Elsevier Science.

NATU RE REV IEWS |D RUG DI SCO VERY VOLUME 4 |APR IL 200 5 |305
R E V I E W S
chemical and physical stability, and their efficacy and
safety profile. To get new, safe therapeutic proteins to
the market more quickly, it is important to obtain a
detailed understanding of what kind of protein modifi-
cation might be acceptable from a safety and efficacy
point of view early in the development process.
However, it is equally important to understand the
mechanisms by which the protein structure can be com-
promised during bulk processing and production of the
final product.This includes strategies to reduce or pre-
vent chemical degradation, denaturation, aggregation
and other structural changes that could prove to be pro-
hibitive for successful development of the drug. It is the
joint responsibility of academia, the pharmaceutical
industry and the regulatory authorities to establish the
scientific background for the safe, fast testing and
assessment of promising new biopharmaceuticals to
the benefit of patients and society.
testing by product stressing can be misleading,because
the Arrhenius kinetics used to extrapolate to low tem-
peratures are not appropriate for temperatures around
the glass transition132.It should also be noted that pro-
teins can undergo reversible conformational changes in
the lyophilized state that expose otherwise buried
regions, making them susceptible to undesirable side
reactions, such as disulphide bond shuffling in the pres-
ence of trace moisture.Examples include insulin133 and
β-galactosidase134 (see also REF.135 and references therein).
Other crosslinking reactions include insulin transami-
dation136 and dityrosine formation occurring under
oxidative conditions or ultraviolet stress137.
Conclusions
The successful development of protein-based medicines
depends on an intimate understanding of their phys-
ico-chemical and biological characteristics, including
1. Holmer, A. F. Survey: Medicines In Development for
HIV/AIDS(Pharmaceutical Research and Manufacturers
Association, Washington DC, 2004).
2. IBM Consulting Services. Pharma 2010: The Threshold of
Innovation (IBM, 2003).
3. Dobson, C. M. Chemical space and biology. Nature 432,
824–828 (2004).
4. Pavlou, A. K. & Reichert, J. M. Recombinant protein
therapeutics — success rates, market trends and values to
2010. Nature Biotechnol. 22, 1513–1519 (2004).
5. Krishnamurthy, R. & Manning, M. C. The stability factor:
importance in formulation development. Curr. Pharm.
Biotech. 3, 361–371 (2002).
6. Wang, W. Instability, stabilization, and formulation of liquid
protein pharmaceuticals. Int. J. Pharm.185, 129–188 (1999).
7. Hermelin, S., Crommelin, D. J. A., Schellekenes, H. &
Jiskoot, W. Structure–immunogenicity relationship of
therapeutic proteins. Pharm. Res. 21, 897–903 (2004).
8. Ahern, T. J. & Manning, M. C. (eds) Stability of Protein
Pharmaceuticals — Part A: Chemical and Physical
Pathways of Protein Degradation Pharmaceutical
Biotechnology Series Vol. 2 (Plenum, New York, 1992).
9. Cleland, J. L., Powell, M. F. & Shire, S. J. The development of
stable protein formulations: a close look at protein
aggregation, deamidation and oxidation. Crit. Rev. Ther. Drug
Carrier Syst.10, 307–377 (1993).
10. Alpar, H. O., Somavarapu, S., Atuah, K. N. & Bramwell, V. W.
Biodegradable mucoadhesive particulates for nasal and
pulmonary antigen and DNA delivery. Adv. Drug Deliv. Rev.
57, 411–430 (2005).
11. Hussain, A., Arnold, J. J., Khan, M. A. & Ashan, F. Absorption
enhancers in pulmonary protein delivery.
J. Control. Release 94, 15–24 (2004).
12. Brange, J., Owens, D. R., Kang, S. & Vølund, A. Monomeric
insulins and their experimental and clinical implications.
Diabetes Care13, 923–954 (1990).
13. Kurtzhals, P. et al. Albumin binding of insulins acylated with
fatty acids: characterization of the ligand–protein interaction
and correlation between binding affinity and timing of the
insulin effect in vivo. Biochem. J.312, 725–731 (1995).
14. Markussen, J. et al. Soluble, fatty acid acylated insulins bind
to albumin and show protracted action in pigs. Diabetologia
39, 281–288 (1996).
15. Knudsen, L. B. et al. Potent derivatives of glucagon-like
peptide-1 with pharmacokinetic properties suitable for
once daily administration. J. Med. Chem. 43, 1664–1669
(2000).
16. Foldvari, M. et al. Palmitoyl derivatives of interferon α: potential
for cutaneous delivery. J. Pharm. Sci.87, 1203–1208 (1998).
17. Wang, J., Shen, D. & Shen, W. C. Preparation, purification,
and characterization of a reversible lipidized desmopressin
with potential anti-diurectic activity. Pharm. Res.16,
1674–1679 (1999).
18. Bhadra, D., Bhadra, S., Jain, P. & Jain, N. K. Pegnology:
a review of PEG-ylated systems. Pharmazie57, 5–29 (2002).
19. Matthews, S. J. & McCoy, C. Peginteferon α2a: a review of
approved and investigational uses. Clin. Ther.26,
991–1025 (2004).
20. Lee, H.-J. & Park, T. G. Preparation and characterization of
mono-PEGylated epidermal growth factor: evaluation of in
vitrobiological activity. Pharm. Res. 19, 845–851 (2002).
21. Veronese, F. & Harris, J. M. Introduction and overview of
peptide and protein pegylation. Adv. Drug Deliv. Rev. 54,
453–459 (2002).
22. Dailey, L. A., Wittmar, M. & Kissel, T. The role of branched
polyesters and their modifications in the development of
modern drug delivery vehicles. J. Control. Release101,
137–149 (2005).
23. Kopecek, J. Smart and genetical engineered biomaterials and
drug delivery systems. Eur. J. Pharm. Sci.20, 1–16 (2003).
24. Crommelin, D. J. et al. Nanotechnological approaches for the
delivery of macromolecules. J. Control. Release 87, 81–88
(2003).
25. Peppas, N. A., Bures, P., Leobandung, W. & Ichikawa, H.
Hydrogels in pharmaceutical formulations. Eur. J. Pharm.
Biopharm.50, 27–46 (2000).
26. Packhaeuser, C. B., Schnieders, J., Oster, C. G. & Kissel, T. In
situforming parenteral drug delivery systems: an overview.
Eur. J. Pharm. Biopharm. 58, 445–455 (2004).
27. Metselaar, J. M., Mastrobattista, E. & Storm, G. Liposomes
for intravenous drug targeting: design and applications. Mini
Rev. Med. Chem. 4, 319–329 (2002).
28. Muller, R., Radtke, M. & Wissing, S. A. Nanostructured lipid
matrices for improved microencapsulation of drugs. Int. J.
Pharm. 248, 121–128 (2002).
29. Bjerregaard, S. et al. Parenteral water/oil emulsions containing
hydrophilic compounds with enhanced in vivo retention:
formulation, rheological characterization and study of in vivo
fate using whole body gammascintography. Int. J. Pharm.
215, 13–27 (2001).
30. Pedersen, T. B., Sabra, M. C., Frokjaer, S., Mouritsen, O. G. &
Jorgensen, K. Association of acylated cationic decapeptides
with dipalmitoylphosphatidylserine-
dipalmitoylphosphatidylcholine lipid membranes. Chem.
Phys. Lipids113, 83–95 (2001).
31. Jørgensen, K., Davidsen, J. & Mouritsen, O. G. Biophysical
mechanisms of phosholipase A2 activation and their use in
liposome-based delivery. FEBS Lett.531, 23–27 (2002).
32. Ross, C. et al. Immunogenicity of interferon-βin multiple
sclerosis patients: influence of preparation, dosage, dose
frequency, and route of administration. Ann. Neurol. 48,
706–712 (2000).
33. Sorensen, P. S. et al. Clinical importance of neutralising
antibodies against interferon-βin patients with relapsing-
remitting multiple sclerosis. Lancet 203, 1184–1191
(2003).
34. Casadevall, N. et al. Pure red-cell aplasia and antierythropoitin
antibodies in patients treated with recombinant erythropoietin.
N. Engl. J. Med.346, 469–475 (2002).
35. Casadevall, N. Antibodies against rHuEPO: native and
recombinant. Nephro. Dial. Transplant. 17 (Suppl. 5), 42–47
(2002).
36. Hermelin, S., Schellekenes, H., Crommelin, D. J. & Jiskoot,
W. Micelle-associated protein in epoetin formulations: a risk
factor for immunogenicity? Pharm. Res.20, 1903–1907
(2003).
37. Jørgensen, L., Van de Weert, M., Vermehren, C.,
Bjerregaard, S. & Frokjaer, S. Probing structural changes of
proteins incorporated into water-in-oil emulsions. J. Pharm.
Sci.93, 1847–1859 (2004).
38. Holm, J. et al. Allergy vaccine engineering: epitope
modulation of recombinant Bet v 1 reduces IgE binding
but retains protein folding pattern for induction of
protective blocking-antibody responses. J. Immunol.173,
5258–5267 (2004).
39. Wang, W. Protein aggregation and its inhibition in
biopharmaceutics. Int. J. Pharm. (in the press).
40. Fink, A. L. Protein aggregation: folding aggregates, inclusion
bodies and amyloid. Folding Des.3, R9–R29 (1998).
41. Carpenter, J. F., Kendrick, B. S., Chang, B. S.,
Manning, M. C. & Randolph, T. W. in Methods in Enzymology
(ed. Wetzel, R.) 236–255 (Academic, San Diego, 1999).
42. Gidalevitz, D., Zhengoing, H. & Rice, S. A. Protein folding
at the air–water interface studied with X-ray reflectivity.
Proc. Natl Acad. Sci. USA 96, 2608–2611 (1999).
43. Nichols, M. R., Moss, M., Reed, D. K., Hoh, J. H. &
Rosenberry, T. L. Rapid assembly of amyloid-βpeptide at a
liquid/liquid interface produces unstable β-sheet fibers.
Biochemistry 44, 165–173 (2005).
44. Kato, A. & Takagi, T. Formation of intermolecular β-sheet
structure during heat denaturation of ovalbumin. J. Agric.
Food Chem. 36, 1156–1159 (1988).
45. Pedersen, J. S., Christiansen, G. & Otzen, D. E. Modulation
of S6 fibrillation by unfolding rates and gatekeeper residues.
J. Mol. Biol.341, 575–588 (2004).
46. Fields, G. & Alonso, D. Theory for the aggregation of proteins
and copolymers. J. Phys. Chem. B96, 3674–3981 (1992).
47. Chi, E. Y., Krishnan, S., Randolph, T. W. & Carpenter, J. F.
Physical stability of proteins in aqueous solution: mechanism
and driving forces in non-native protein aggregation. Pharm.
Res.20, 1325–1336 (2003).
48. Kendrick, B. S., Carpenter, J. F., Cleland, J. L. &
Randolph, T. W. A transient expansion of the native state
precedes aggregation of recombinant human interferon-γ.
Proc. Natl Acad. Sci. USA 95, 14142–14146 (1998).
49. Treuheit, M. J., Kosky, A. A. & Brems, D. N. Inverse
relationship of protein concentration and aggregation. Pharm.
Res.19, 511–516 (2002).
50. Krishnan, S. et al. Aggregation of granulocyte colony
stimulating factor under physiological conditions:
characterization and thermodynamic inhibition. Biochemistry
41, 6422–6431 (2002).
51. Hevehan, D. & De Bernardez-Clark, E. Oxidative renaturation
of lysozyme at high concentrations. Biotechnol. Bioeng. 54,
221–230 (1997).
52. Gass, J. N., Gunn, K. E., Sriburi, R. & Brewer, J. W.
Stressed-out B cells? Plasma-cell differentiation and the
unfolded protein response. Trends Immunol. 25, 17–24
(2004).
53. Rutkowski, D. T. & Kaufman, R. J. A trip to the ER: coping
with stress. Trends Cell Biol. 14, 20–28 (2004).
54. Ellis, R. J. Macromolecular crowding: an important but
neglected aspect of the intracellular environment. Curr. Opin.
Struct. Biol.11, 114–119 (2001).

30 6 |AP RIL 20 05 |VOLUM E 4 www.nat ure.com/reviews/drug disc
R E V I E W S
55. Minton, A. P. Influence of excluded volume upon
macromolecular structure and associations in ‘crowded’
media. Curr. Opin. Biotech. 8, 65–69 (1997).
56. London, J., Skrzynia, C. & Goldberg, M. E. Renaturation of
Escherichia coli tryptophanase after exposure to 8M urea.
Evidence for the existence of nucleation centers. Eur. J.
Biochem. 47, 409–415 (1974).
57. Chapman, M. R. et al. Role of Eschericia colicurli operons
in directing amyloid fiber formation. Science 295, 851–855
(2002).
58. Huff, M. E., Balch, W. E. & Kelly, J. W. Pathological and
functional amyloid formation orchestrated by the secretory
pathway. Curr. Op. Struct. Biol. 13, 674–682 (2003).
59. Sunde, M. et al. Common core structure of amyloid fibrils by
synchrotron X-ray diffraction. J. Mol. Biol. 273, 729–739
(1997).
60. Dobson, C. M. Protein misfolding, evolution and disease.
Trends Biochem. Sci. 24, 329–332 (1999).
61. Harper, J. D. & Lansbury, P. T. Models of amyloid seeding in
Alzheimer’s disease and scrapie: mechanistic truths and
physiological consequences of the time-dependent solubility of
amyloid proteins. Annu. Rev. Biochem. 66, 385–407 (1997).
62. Cao, Z. & Ferrone, F. A. A 50th order reaction predicted and
observed for sickle hemoglobin nucleation. J. Mol. Biol. 256,
219–222 (1996).
63. Patro, S. Y. & Przybycien, T. M. Simulations of kinetically
irreversible protein aggregate structure. Biophys. J.66,
1274–1289 (1994).
64. Istrail, S., Schwartz, R. & King, J. Lattice simulations of
aggregation funnels for protein folding. J. Comput. Biol.6,
143–162 (1999).
65. Ferrone, F. Analysis of protein aggregation kinetics. Meth.
Enzymol. 309, 256–274 (1999).
66. Clark, P. L. Protein folding in the cell: reshaping the folding
funnel. Trends Biochem. Sci. 29, 527–534 (2004).
67. Chiti, F. et al. Kinetic partitioning of protein folding and
aggregation. Nature Struct. Biol.9, 137–143 (2002).
68. Kunjithapatham, R. et al. Role for the α-helix in aberrant
protein aggregation. Biochemistry 44, 149–156 (2005).
69. Fernandez-Escamilla, A. M., Rousseau, F., Schymkowitz, J.
& Serrano, L. Prediction of sequence-dependent and
mutational effects on the aggregation of peptides and
protein. Nature Biotechnol. 22, 1302–1306 (2004).
70. Chiti, F., Stefani, M., Taddei, N., Ramponi, G. & Dobson, C. M.
Rationalization of the effects of mutations on peptide and
protein aggregation rates. Nature424, 805–808 (2003).
71. DuBay, K. F. et al. Predicting absolute aggregation rates of
amyloidogenic polypeptide chains. J. Mol. Biol. 341,
1317–1326 (2004).
72. Fraser, P. E. et al. Conformation and fibrillogenesis of
Alzheimer Aβpeptides with selected substitution of charged
residues. J. Mol. Biol.244, 64–73 (1994).
73. Frokjaer, S. et al. Probing the mechanism of insulin fibril
formation with insulin mutants. Biochemistry 40,
8397–8409 (2001).
74. Otzen, D. E., Kristensen, P. & Oliveberg, M. Designed protein
tetramer zipped together with an Alzheimer sequence:
a structural clue to amyloid assembly. Proc. Natl Acad. Sci.
USA 97, 9907–9912 (2000).
75. Otzen, D. E. & Oliveberg, M. Salt-induced detour through
compact regions of the protein folding landscape. Proc. Natl
Acad. Sci. USA 96, 11746–11751 (1999).
76. Ferrone, F. Analysis of protein aggregation kinetics. Meth.
Enzymol. 309, 256–274 (1999).
77. Hall, D. & Edskes, H. Silent prions lying in wait: a two-hit
model of prion/amyloid formation and infection. J. Mol. Biol.
336, 775–786 (2004).
78. Masel, J. & Jansen, V. A. Designing drugs to stop the
formation of prion aggregates and other amyloids. Biophys.
Chem.88, 47–59 (2000).
79. Chen, S., Ferrone, F. & Wetzel, R. Huntington’s disease age-
of-onset linked to polyglutamine aggregation nucleation.
Proc. Natl Acad. Sci. USA 99, 11884–11889 (2002).
80. Jarrett, J. T., Berger, E. P. & Lansbury, P. T. The carboxy
terminus of the β amyloid protein is critical for the seeding of
amyloid formation: implications for the pathogenesis of
Alzheimer’s Disease. Biochemistry 32, 4693–4697 (1993).
81. Krebs, M. R., Morozova-Roche, L. A., Daniel, K.,
Robinson, C. V. & Dobson, C. M. Observation of sequence
specificity in the seeding of protein amyloid fibrils. Protein Sci.
13, 1933–1938 (2004).
82. Wadai, H. et al. Stereospecific amyloid-like fibril formation by a
peptide fragment of β2-microglobulin. Biochemistry 44,
157–164 (2004).
83. O’Nuallain, B., Williams, A. D., Westermark, P. & Wetzel, R.
Seeding specificity in amyloid growth induced by
heterologous fibrils. J. Biol. Chem.279, 17490–17499 (2004).
84. Petkova, A. T. et al. Self-propagating, molecular-level
polymorphism in Alzheimer’s β-amyloid fibrils. Science 307,
262–265 (2005).
85. Kardos, J., Yamamoto, K., Hasegawa, K., Naiki, H. & Goto, Y.
Direct measurement of the thermodynamic parameters of
amyloid formation by isothermal titration calorimetry. J. Biol.
Chem.279, 55308–55314 (2004).
86. Perutz, M. F., Finch, J. T., Berriman, J. & Lesk, A. Amyloid
fibers are water-filled nanotubes. Proc. Natl Acad. Sci. USA
99, 5591–5595 (2002).
87. Onoue, S. et al. Mishandling of the therapeutic peptide
glucagon generates cytotoxic amyloidogenic fibrils. Pharm.
Res.21, 1274–1283 (2004).
88. Wurth, C., Guimard, N. K. & Hecht, M. H. Mutations that
reduce aggregation of the Alzheimer’s Aβ42 peptide: an
unbiased search for the sequence determinants of Aβ
amyloidogenesis. J. Mol. Biol.319, 1279–1290 (2002).
89. Manning, M. & Colón, W. Structural basis of protein kinetic
stability: resistance to sodium dodecyl sulfate suggests a
central role for rigidity and a bias toward β-sheet structure.
Biochemistry 43, 11248–11254 (2004).
90. Klibanov, A. M. Why are enzymes less active in organic
solvents than in water? Trends Biotechnol.15, 97–101 (1997).
91. Partridge, J., Moore, B. D. & Halling, P. J. α-chymotrypsin
stability in aqueous-acetonitrile mixtures: is the native enzyme
thermodynamically or kinetically stable under low water
conditions? J. Mol. Catal., B Enzym.6, 11–20 (1999).
92. Machius, M., Declerck, N., Huber, R. & Wiegand, G. Kinetic
stabilization of Bacillus licheniformisα-amylase through
introduction of hydrophobic residues at the surface. J. Biol.
Chem.278, 11546–11553 (2003).
93. Murphy, C. M. Peptide aggregation in neurodegenerative
disease. Annu. Rev. Biomed. Eng. 4, 155–174 (2002).
94. Conway, K. A., Rochet, J.-C., Bieganski, R. M. &
Lansbury, P. T. Kinetic stabilization of the α-synuclein
protofibril by a dopamine–α-synuclein adduct. Science
294, 1346–1349 (2001).
95. Taniguchi, S. et al. Inhibition of heparin-induced tau filament
formation by phenothiazines, polyphenols, and porphyrins.
J. Biol. Chem. 280, 7614–7623 (2005).
96. Tjernberg, L. O. et al. Arrest of β-amyloid fibril formation by a
pentapeptide ligand. J. Biol. Chem. 271, 8545–8548 (1996).
97. Rzepecki, P. et al. Prevention of Alzheimer’s disease-
associated Aβaggregation by rationally designed nonpeptidic
β-sheet ligands. J. Biol. Chem. 279, 47497–47505 (2004).
98. Gestwicki, J. E., Crabtree, G. R. & Graef, I. A. Harnessing
chaperones to generate small-molecule inhibitors of amyloid β
aggregation. Science 306, 865–869 (2004).
99. Shire, S. J., Shahrokh, Z. & Liu, J. H. Challenges in the
development of high protein concentration formulations.
J. Pharm. Sci. 93, 1390–1402 (2004).
100. Timasheff, S. Protein hydration, thermodynamic binding, and
preferential hydration. Biochemistry 41, 13473–13482 (2002).
101. Baldwin, R. L. How Hofmeister ion interactions affect protein
stability. Biophys. J.71, 2056–2063 (1996).
102. De Bernardez-Clark, E., Schwarz, E. & Rudolph, R. Inhibition
of aggregation side reactions during in vitro protein folding.
Meth. Enzymol. 309, 217–236 (1999).
103. Golovanov, A. P., Hautbergue, G. M., Wilson, S. A. & Lian, L.-Y.
A simple method for improving protein solubility and long-
term stability. J. Am. Chem. Soc. 126, 8933–8939 (2004).
104. Webb, S. D., Cleland, J. L., Carpenter, J. F. & Randolph, T. W.
A new mechanism for decreasing aggregation of recombinant
human interferon-γby a surfactant: slowed dissolution of
lyophilized formulations in a solution containing 0.03%
polysorbate 20. J. Pharm. Sci. 91, 543–558 (2002).
105. Kerwin, B. A., Heller, M. C., Levin, S. H. & Randolph, T. W.
Effects of Tween 80 and sucrose on acute short-term stability
and long-term storage at –20 ºC of a recombinant
hemoglobin. J. Pharm. Sci. 87, 1062–1068 (1998).
106. Arakawa, T. & Kita, Y. Protection of bovine serum albumin from
aggregation by Tween 80. J. Pharm. Sci.89, 646–651 (2000).
107. Rozema, D. & Gellman, S. H. Artificial chaperone-assisted
refolding of carbonic anhydrase B. J. Biol. Chem. 271,
3478–3487 (1996).
108. Tsai, A. M., van Zanten, J. H. & Betenbaugh, M. J. II:
Electrostatic effect in the aggregation of heat-denatured
RNase A and implications for protein additive design.
Biotechnol. Bioeng. 59, 281–285 (1998).
109. Zhao, H., Tuominen, E. K. J. & Kinnunen, P. K. J. Formation of
amyloid fibers triggered by phosphatidylserine-containing
membranes. Biochemistry43, 10302–10307 (2004).
110. Knight, J. D. & Miranker, A. D. Phospholipid catalysis of
diabetic amyloid assembly. J. Mol. Biol.341, 1175–1187
(2004).
111. Necula, M., Chirita, C. N. & Kuret, J. Rapid anionic micelle-
mediated α-synuclein fibrillization in vitro. J. Biol. Chem.
278, 46674–46680 (2003).
112. Chirita, C. N. & Kuret, J. Evidence for an intermediate in tau
filament formation. Biochemistry 43, 1704–1714 (2004).
113. Frömming, K.-H. & Szejtli, J. Cyclodextrins in Pharmacy
(Kluwer Academic, Dordrecht, 1994).
114. Larsen, K. L. Large cyclodextrins. Biol. J. Armenia 53,
9–26 (2001).
115. Banga, A. K. & Mitra, R. Minimization of shaking-induced
formation of insoluble aggregates of insulin by cyclodextrins.
J. Drug Target. 1, 341–345 (1993).
116. Dotsikas, Y. & Loukas, Y. L. Kinetic degradation study of
insulin complexed with methyl-βcyclodextrin. Confirmation
of complexation with electrospray mass spectrometry and
1H-NMR. J. Pharm. Biomed. Anal. 29, 487–494 (2002).
117. Tokihiro, K., Irie, T., Uekama, K. & Pitha, J. Potential use of
maltosyl-β-cyclodextrin for inhibition of insulin self-association
in aqueous solution. Pharm. Sci.1, 49–53 (1995).
118. Otzen, D. E., Knudsen, B. R., Aachmann, F. L., Larsen, K. L.
& Wimmer, R. Structural basis for cyclodextrins’ suppression
of human growth hormone aggregation. Protein Sci. 11,
1779–1787 (2002).
119. Hagenlocher, M. & Pearlman, R. Use of a substituted
cyclodextrin for stabilization of solutions of recombinant
human growth hormone. Pharm. Res. 6, S30 (1989).
120. Sharma, L. & Sharma, A. Influence of cyclodextrin ring
substituents on folding-related aggregation of bovine carbonic
anhydrase. Eur. J. Biochem. 268, 2456–2463 (2001).
121. Sigurjonsdottir, J. F., Loftsson, T. & Masson, M. Influence of
cyclodextrins on the stability of the peptide salmon calcitonin
in aqueous solution. Int. J. Pharm. 186, 205–213 (1999).
122. Aachmann, F. L., Otzen, D. E., Larsen, K. L. & Wimmer, R.
Structural background of cyclodextrin–protein interactions.
Protein Eng.16, 1–8 (2003).
123. Cooper, A. Effect of cyclodextrins on the thermal stability of
globular proteins. J. Am. Chem. Soc. 114, 9208–9209 (1992).
124. Otzen, D. E. & Oliveberg, M. A simple way to measure
protein refolding rates in water. J. Mol. Biol. 313, 479–483
(2001).
125. Karuppiah, N. & Sharma, A. Cyclodextrins as protein folding
aids. Biochem. Biophys. Res. Commun. 211, 60–66 (1995).
126. Easton, C. J. & Lincoln, S. F. Modified Cyclodextrins:
Scaffolds and Templates for Supramolecular Chemistry
(Imperial College Press, London, 1999).
127. Szejtli, J & Osa, T (eds). Comprehensive Supramolecular
Chemistry(Elsevier Science, Oxford, 1996).
128. Franks, F. Protein destabilization at low temperatures. Adv.
Protein Chem.46, 105–139 (1995).
129. Wang, W. Lyophilization and development of slid protein
pharmaceuticals. Int. J. Pharm. 203, 1–60 (2000).
130. Franks, F. Freeze-drying of bioproducts: putting principles into
practice. Eur. J. Pharm. Biopharm.45, 221–229 (1998).
131. Franks, F. Long-term stabilization of biologicals.
Biotechnology (NY)12, 253–256 (1994).
132. Levine, H. & Slade, L. Water as a plasticizer: physico-chemical
aspects of low-moisture polymeric systems. Water Sci. Rev.
3, 79–185 (1988).
133. Costantino, H. R., Langer, R. & Klibanov, A. M. Moisture-
induced aggregation of lyophilized insulin. Pharm. Res. 11,
21–29 (1994).
134. Yoshioka, S., Aso, Y., Izutsu, K. & Terao, T. Aggregates
formed during storage of β-galactosidase in solution and in
the freeze-dried state. Pharm. Res. 10, 687–691 (1993).
135. Klibanov, A. M. & Schefiliti, J. A. On the relationship between
conformation and stability in solid pharmaceutical protein
formulations. Bioctechnol. Lett. 26, 1103–1106 (2004).
136. Brange, J., Andersen, L., Laursen, E. D., Meyn, G. &
Rasmussen, E. Toward understanding insulin fibrillation.
J. Pharm. Sci. 86, 517–525 (1997).
137. Malencik, D. A. & Anderson, S. R. Dityrosine as a product of
oxidative stress and fluorescent probe. Amino Acids25,
233–247 (2003).
138. Dobson, C. M. Protein folding and misfolding. Nature 426,
884–890 (2003).
Acknowledgements
D.O. is supported by the Technical Science Research Foundation
and the Villum Kann Rasmussen Foundation.
Competing interests statement
The authors declare no competing financial interests.
Online links
DATABASES
The following terms in this article are linked online to:
Entrez Gene:
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene
GLP1 | IL-2
OMIM: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=OMIM
Alzheimer’s disease | Huntington’s disease | Parkinson’s disease
FURTHER INFORMATION
FoldX — a force field for energy calculations in proteins:
http://foldx.embl.de
Tango — computer algorithm for prediction of aggregating
regions in unfolded polypeptide chains: http://tango.embl.de
Access to this interactive links box is free online.