Вы находитесь на странице: 1из 8

Characterization of protein aggregation: The case of

a therapeutic immunoglobulin
Barthlemy Demeule
a
, M. Jayne Lawrence
b
, Alex F. Drake
b
, Robert Gurny
a
, Tudor Arvinte
a,

a
Department of Pharmaceutics and Biopharmaceutics, School of Pharmaceutical Sciences, University of Geneva, University of Lausanne,
30 quai Ernest-Ansermet, CH-1211 Geneva 4, Switzerland
b
Department of Pharmacy, Kings College London, Franklin-Wilkins Building, 150 Stamford Street, London SE1 9NH, UK
Received 23 August 2006; accepted 12 October 2006
Available online 27 October 2006
Abstract
In this paper, a therapeutic immunoglobulin (Antibody A) has been characterized in two solutions: (1) 0.1% acetic acid containing 50 mM
magnesium chloride, a solution in which the immunoglobulin is stable, and (2) 10 mM sodium phosphate buffer pH 7. The protein solutions
were characterized by microscopy, asymmetrical flow field-flow fractionation (FFF), light scattering, circular dichroism, fluorescence and
fluorescence lifetime spectroscopy. The results show that Antibody A dissolved in 0.1% acetic acid containing 50 mM magnesium chloride exists
as 88% monomer, 2% low molecular weight aggregates and 10% high molecular weight aggregates (>1 million Dalton). In phosphate buffer,
Antibody A formed micrometre-sized aggregates that were best characterized by fluorescence microscopy. The aggregation of Antibody A in
phosphate buffer was shown to be concomitant with conformational changes in amino acid residue side chains. The aggregates formed in
phosphate buffer were easily disrupted during FFF analysis, indicating that they are formed by weak interactions. The combination of microscopy,
asymmetrical flow field-flow fractionation (FFF) and spectroscopy allowed a reliable assessment of protein self association and aggregation.
2006 Elsevier B.V. All rights reserved.
Keywords: Immunoglobulin; Aggregation; Characterization; Biopharmaceutical; Conformation
1. Introduction
The use of proteins and nucleic acids as biopharmaceutical
drugs has opened the way for new treatments of diseases.
Proteins are the most widely used biopharmaceutical drugs on
the market [1]. However, the efficacy of protein drugs can be
compromised by instability. From production to administration,
various factors such as pH, shear and thermal stress can
compromise the stability of therapeutic proteins [2]. One major
aspect of protein instability is self-association leading to
aggregation. Aggregates can reduce the efficacy of protein
drugs and can lead to immunological reactions [3,4] and toxicity
[5]. During the development of a protein formulation, a
combination of appropriate analytical methods must be used to
detect subtle changes in the state of a protein to ensure the
efficacy and safety [6]. Proteins can aggregate by forming
covalent or non-covalent bonds. Non-covalent protein bonds are
formed by interactions involving hydrogen bonding, electro-
static repulsion and attraction and factors related to hydro-
phobicity. The relative weakness of non-covalent protein bonds
can lead to aggregate disruption during the analytical process
[7]. Loose protein aggregates formed by non-covalent bonds
are therefore more difficult to detect and quantify. The size of
aggregates can be determined by static or dynamic light
scattering; however in the case of a multimodal distribution
(e.g. presence of monomers and aggregates), the results are often
biased towards the larger species. The precise composition of
protein aggregates in solutions can be measured with a
separation method coupled with static or dynamic light
scattering detection. Separation can be achieved by size-
exclusion chromatography although non-covalent aggregates
are potentially disrupted during the process [7]. The aggregation
state of a protein can be measured by static light scattering after
separation by asymmetrical flow field-flow fractionation (FFF).
Biochimica et Biophysica Acta 1774 (2007) 146153
www.elsevier.com/locate/bbapap
Abbreviations: CD, circular dichroism; FFF, asymmetrical flow field-flow
fractionation; RMS, root mean square; TCSPC, time-correlated single-photon
counting; IgG, immunoglobulin G

Corresponding author. Tel.: +41 22 379 63 39; fax: +41 22 379 65 67.
E-mail address: tudor.arvinte@pharm.unige.ch (T. Arvinte).
1570-9639/$ - see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.bbapap.2006.10.010
FFF has the ability to separate proteins and particles without an
interacting stationary phase and under much lower pressures
than in conventional chromatographic methods. Both weakly
bound protein assemblies and more strongly formed aggregates
can be separated by FFF in one run, under low shear stress [8].
In addition to the detection and characterization of protein
aggregates, the spectroscopic properties of proteins can be
investigated to provide a better understanding of the protein
conformation [9]. Intrinsic fluorescence, fluorescence aniso-
tropy and fluorescence lifetime, 90 light scattering and circular
dichroism are able to detect small changes in the protein
conformation. Staining with hydrophobic dyes such as Nile Red
can show the presence and extent of hydrophobic surfaces on
the protein population and enhances the detection of protein
aggregates by fluorescence microscopy [10].
A combination of separation and spectroscopic techniques is
presented in this paper. Antibody A, a therapeutic immunoglobu-
lin, was characterized in several buffers at lowconcentrations, with
a focus on weak molecular interactions. The combination of
various methods is necessary to allow a rational selection of
appropriate protein formulations during protein drug development.
2. Materials and methods
2.1. Materials
Antibody A is an IgG1 recombinant humanized monoclonal antibody
provided by Novartis Pharma AG, Basel, Switzerland. Antibody Awas stored at
4 C in a stock solution consisting of 0.1% acetic acid containing 50 mM
magnesium chloride (protein concentration: 193 mg/ml). The solvent constitu-
tion of the stock solution ensures the chemical and physical stability of the
antibody. All antibody solutions were prepared by dilution of the aforemen-
tioned stock solution in either 0.1% acetic acid containing 50 mM magnesium
chloride or 10 mM sodium phosphate buffer pH 7. The concentration of
Antibody A solutions was determined by UV spectroscopy at 280 nm, based
upon an absorbance of 1.6 for a 1 mg/ml solution in a 1-cm cell. The UV
measurements were performed at 25 C with a temperature-controlled Cintra 40
spectrophotometer (GBC, Melbourne, Australia). In the case of aggregated
solutions, the UV spectrum was corrected to account for the apparent
absorbance due to light scattering [11]. Nile Red (9-diethylamino-5H-benzo
[]phenoxazine-5-one) was purchased from Sigma (Buchs, Switzerland). Nile
Red was dissolved in ethanol, to produce a 100 M stock solution, which was
stored at 4 C, protected from light. All water used in the experiments was
deionized by a Milli-Q academic system (Millipore, Billerica, USA).
2.2. Fluorescence microscopy
Antibody Asamples were stained with Nile Red prior to observation. 0.5 l of
the Nile Red stocksolution was added to 50 l of the protein solution. Immediately
after staining, aliquots of the protein solution containing the dye were placed on
Kova Glasstic slides (Hycor, Garden Grove, USA) and observed by microscopy.
The observations were performed on an Axiovert 200 microscope (Zeiss,
Gttingen, Germany) equipped with a mercury discharge lamp. To visualize Nile
Red fluorescence, a Zeiss filter cube n15 was used (EX BP 546/12, BS FT 580,
EMLP 590). The images were acquired with a cooled Retiga 1300 C colour CCD
camera (QImaging, Burnaby, Canada) and processed with the Openlab version
3.1.7 software (Improvision, Coventry, UK). The observations were performed
using 10x, 20x and 40x A-Plan LD objectives (Zeiss, Gttingen, Germany).
2.3. Static and dynamic light scattering
Molecular size determinations were performed in cylindrical light scattering
cuvettes at room temperature on an ALV-5000 goniometer (ALV-Laser
Vertriebsgesellschaft mbH, Langen, Germany). The light source was a HeNe
laser (wavelength: 632.8 nm). Static light scattering intensity (yielding
molecular weight information) as well as dynamic light scattering (yielding
size information) were determined at various angles in 10 steps between 30 and
150. Toluene was used as the primary standard. A 10 l aliquot of the Antibody
Astock solution was diluted in 3 ml of either 0.1%acetic acid containing 50 mM
magnesium chloride or 10 mM phosphate buffer pH 6.9. The samples were not
filtered. Calculations were made only at angles which gave a <10% error in the
dynamic light scattering measurement. In the static light scattering experiments,
molecular weight was first approximated with the following formula:
K
1
c
R
90

1
M
2Bc
[12,13] where c is the concentration, R
90
is the Raleigh ratio of the sample
relative to benzene, M is the molecular weight of the scattering unit, B is the
second virial coefficient and K
1
is an optical constant defined as 4
2
n
2
(dn/dc)
2
/
(N
A

4
) where n is the solvent refractive index, dn/dc is the refractive index
increment, is the wavelength of incident light and N
A
is the Avogadro's
number. Arefractive index increment (dn/dc) of 0.185 ml/g was considered to be
a sufficient approximation [13]. Molecular weight values were confirmed by a
Zimm plot.
2.4. Asymmetrical flow field-flow fractionation (FFF)
Fractionation of the protein samples was performed in a trapezoidal channel,
26.5 cm in length and 350 m in height, connected to an Eclipse F system
(Wyatt Technology Europe, Dernbach, Germany). The bottom of the channel
was lined with a regenerated cellulose membrane with a 10 kDa cut-off
(Microdyn-Nadir GmbH, Wiesbaden, Germany). The quantity of Antibody A
injected into the system was between 5 and 10 g in 10 l when Antibody Awas
diluted in 0.1% acetic acid containing 50 mM magnesium chloride. When
Antibody Awas diluted in phosphate buffer, 1.25 to 2.5 g in 2 l were injected
into the channel. The elution medium was the same as the protein formulation,
e.g. if Antibody Awas dissolved in 10 mM phosphate buffer pH 6.9, the same
buffer was used as eluent. The channel flowwas set to 1 ml/min and the injection
flow to 0.2 ml/min. For the analysis of Antibody A in 0.1% acetic acid
containing 50 mM magnesium chloride, the cross flow was initially set to
2.5 ml/min for 10 min, then lowered to 0.25 ml/min for 3 min and finally left at
0.25 ml/min for 1.5 min. To analyse the fragile aggregates of Antibody A in
phosphate buffer, the separation started with a focus flow of 0.5 ml/min for
1.5 min, followed by a cross flow raised from 0.5 to 1.8 ml/min for 0.2 min,
followed by 1.8 ml/min for 7.8 min, then from 1.8 to 0.18 ml/min for 3 min and
finally 0.18 ml/min for 1.5 min.
A Dawn EOS multi-angle light scattering detector (Wyatt Technology, Santa
Barbara, USA) and a UV detector (Agilent Technologies Schweiz AG, Basel,
Switzerland) were coupled in-line with the FFF channel. The light scattering
detector was equipped with a GaAs laser (wavelength: 690 nm) and eighteen
detectors. Scattered light was collected at defined angles between 14 and 163.
Sample concentration was determined by UVabsorbance at 280 nm. Data were
collected and analysed with the Astra version 4.90.08 software. A refractive
index increment (dn/dc) of 0.185 ml/g was considered to be a sufficient
approximation [13]. A second virial coefficient of 110
4
molml/g
2
was used,
assuming small repulsive forces between the proteins as seen in low ionic
strength protein solutions [14,15]. The use of theoretical values for dn/dc and
second virial coefficient do not compromise the ability of FFF to distinguish
monomers from aggregates. The results are given in molecular weight or root
mean square (RMS) radius.
2.5. Circular dichroism
Circular dichroism and absorption spectra were measured with a Jasco J720
and a Jasco J600 spectropolarimeter (Tokyo, Japan). Near-UV spectra were
measured between 240 and 450 nm in cylindrical 1 cm cells; far-UV spectra
were measured between 205 and 240 nm in cylindrical 0.5 mm cells. Spectra
were recorded at a speed of 10 nm/min, with a step size of 0.2 nm, a response
time of 4 s, and a spectral bandwidth of 2 nm. The immunoglobulin stock
solution was diluted to a concentration of 0.7 mg/ml in 0.1% acetic acid
147 B. Demeule et al. / Biochimica et Biophysica Acta 1774 (2007) 146153
containing 50 mM magnesium chloride, and in 10 mM sodium phosphate
buffer pH 6.9. The spectra of the buffer solutions were measured and
subtracted from the spectra of the protein solutions. Samples were placed as
close to the detector as possible to maximise the collection of scattered light
[16].
2.6. Fluorescence measurements
The steady-state fluorescence and fluorescence anisotropy measurements
were performed with a Fluoromax spectrofluorometer (Spex, Stanmore, UK)
at 25 C in a thermostatted cuvette holder. The tryptophan fluorescence was
monitored between 250 and 450 nm, with an excitation wavelength of
280 nm. The spectra were recorded with a 0.01-s integration time per 1 nm
increment. The excitation and emission slits were set to 1 and 0.3 mm,
respectively. Nile Red fluorescence was monitored between 575 and
750 nm, with an excitation wavelength of 560 nm. The spectra were
recorded with a 0.01-s integration time per 1 nm increment. The excitation
and emission slits were set to 1 and 2 mm, respectively. Fluorescence
spectra of the buffers without Antibody A were subtracted from the protein
emission spectra. Steady-state tryptophan anisotropy measurements were
performed using prism polarizers and the anisotropy (A) was calculated
from the equation:
A I
0;0
G I
0;90
=I
0;0
2G I
0;90

where I
m,n
is the fluorescence intensity at a given wavelength, and the
subscripts indicate the position of the polarizers in the excitation (m) and
emission (n) beams, relative to the vertical axis. G is a correction factor,
G=I
90,0
/I
90,90
The anisotropy value was calculated from fluorescence spectra
(background corrected) between 325 and 345 nm, using an excitation
wavelength of 280 nm, with 4-s integration time per 1 nm increment. The
excitation and emission slits were both set to 1 mm.
2.7. 90 light scattering
Light scattering intensity of protein solutions was measured with the
Fluoromax spectrofluorometer (Spex, Stanmore, UK) between 500 and 750 nm,
with the excitation and emission monochromators synchronized. The spectra
were recorded with 0.01-s integration time per 1 nm increment. The excitation
and emission slits were set to 1 mm. The aggregation was monitored by the area
under the curve between 500 and 750 nm. The light scattering intensity is
expressed as counts per second (cps).
2.8. Fluorescence lifetime measurements
Fluorescence lifetimes were measured using time-correlated single-photon
counting (TCSPC) on an IBH 5000U fluorescence lifetime spectrophotometer
(Glasgow, United Kingdom), fitted with a 279 nm and a 560 nm NanoLED
excitation sources and a monochromator at the emission side. Emission
wavelength was determined by steady-state fluorescence spectroscopy. Data
analysis was performed using the DAS6 software (IBH, Glasgow, United
Kingdom). After reconvolution of the intensity decay with the instrument
response function, the calculated data were analysed by linear and non-linear
least-squares modeling. The average lifetimes were calculated using the
equation:
s
P
i
a
i
s
2
i
P
i
a
i
s
i
[17] where is the fluorescence decay time and the normalized pre-
exponential factor. Rotational correlation times were determined by measuring
intensity decays with film polarizers in the optical path. Intensity decays were
collected until a difference of 10000 counts was achieved between the crossed
and uncrossed polarizer orientations. Rotational correlation times were
calculated using an impulse reconvolution model. The fundamental anisotropy
A
0
was determined using the Perrin equation [18,19]:
A
0
A
1
s
#
where is the rotational correlation time. Fundamental anisotropy values allow
the determination of the angle between the excitation and emission transition
electric dipoles [18,19]:
A
0

3cos
2
b 1
5
The instrument response function was measured using a dilute suspension of
colloidal silica (Ludox, Aldrich, Milwaukee, USA). The antibody concentration
was 0.7 mg/ml.
3. Results
3.1. Comparison of Antibody A aggregation in two buffers by
microscopy, static and dynamic light scattering
Fluorescence microscopy after staining with Nile Red has
been shown to efficiently detect protein aggregates. The
aggregation of Antibody A in phosphate buffer near
physiological pH has been reported, whereas no aggregates
have been observed by microscopy in 0.1% acetic acid
containing 50 mM magnesium chloride [10]. Fig. 1 shows
Antibody A solutions in 0.1% acetic acid containing 50 mM
magnesium chloride (94 mg/ml) as well as in phosphate buffer
(0.8 mg/ml). The particles formed in phosphate buffer have a
mean diameter of 3.18 m as determined by light microscopy
[10].
Characterization of Antibody A in 0.1% acetic acid con-
taining 50 mM magnesium chloride was performed by static
light scattering. Static light scattering allows the determination
of molecular weight. Potential disruption of aggregates by
filtration or other purification steps was avoided by measure-
ment without prior filtration. Dynamic light scattering measure-
ments were simultaneously made on the samples. The error in
the dynamic light scattering measurements was used as a quality
indicator and for each experiment, only the light scattering
Fig. 1. Fluorescence photomicrographs of Antibody A dissolved in 0.1% acetic
acid containing 50 mM magnesium chloride (A) and in 10 mM phosphate buffer
pH 7.1 (B). No aggregates were visible in 0.1% acetic acid containing 50 mM
magnesium chloride, even though protein concentration was high (94 mg/ml).
Antibody A solution in phosphate buffer (0.8 mg/ml) showed many spherical
aggregates, with a mean diameter of 3.18 m.
148 B. Demeule et al. / Biochimica et Biophysica Acta 1774 (2007) 146153
angles showing a dynamic light scattering error <10% were
considered. Even though the data could not be perfectly fitted,
both the first molecular weight approximation (see Materials
and methods) and the Zimm plot indicated a molecular weight
around 2 million Daltons, far above the expected 150 kDa for a
monomeric IgG.
Characterization of Antibody A in phosphate buffer was
performed by dynamic light scattering. The apparent diameter
of the aggregates increased from 1 to >2.0 m in the first few
min after dilution of the stock solution.
3.2. Assessment of the aggregate content by FFF
Antibody A was dissolved in 0.1% acetic acid containing
50 mM magnesium chloride at a concentration of 0.5 mg/ml
and analysed by asymmetrical flow field-flow fractionation
(FFF). A typical chromatogram is presented in Fig. 2. When
diluted in 0.1% acetic acid containing 50 mM magnesium
chloride, Antibody A showed two main peaks: a monomer
peak of 160 kDa, representing 88% of the protein mass, and a
second peak corresponding to a molecular weight between 1
and 2 million Daltons (mean: 1.5 million Daltons), represent-
ing 10% of the protein mass. A small peak representing 2%
of the protein mass was observed between the two main
peaks, probably corresponding to a dimer. Upon prolonging
the focussing time, the proportion of monomers and
aggregates remained identical (data not shown). Prolonging
the focussing time causes a concentration of the antibody in
the FFF channel and a higher shear stress, which did not
disrupt the aggregates. Interestingly, when the eluent used in
the FFF system was not the same as the protein solution, (i.e.
10 mM phosphate buffer pH 7.1 rather than 0.1% acetic acid
containing 50 mM magnesium chloride), the second main
peak disappeared (Fig. 3), indicating that the high molecular
weight aggregates were formed by weak interactions,
stabilized by acetic acid containing magnesium chloride.
Antibody Awas also analysed in 10 mMphosphate buffer pH
7.1 at a concentration of 0.6 mg/ml. Two factors hampered the
analysis of the aggregates that were observed by fluorescence
microscopy (Fig. 1): the fragility of the aggregates, disrupted by
the cross-flow, and the considerable difference in size between
the monomers and aggregates. Nevertheless, a very mild
separation method allowed the rapid elution of the larger
aggregates in the steric hyperlayer mode, followed by the elution
of the monomer (Fig. 4). In steric hyperlayer mode, large
particles are positioned far fromthe channel walls and elute prior
to smaller species [8]. The aggregates, representing 20% of the
protein mass showed a wide size distribution, with particles
>200 nm in diameter. The monomer (170 kDa) represented 80%
of the protein mass.
3.3. Changes in Antibody A conformation monitored by
circular dichroism
Differences in circular dichroism spectra were observed
between Antibody A dissolved in 0.1% acetic acid containing
50 mM magnesium chloride (pH5) and Antibody A dissolved
in phosphate buffer pH 6.9 at the same concentration (0.7 mg/
ml) (Fig. 5). The two formulations showed differences in the
near-UVand, to a much lesser extent, in the far-UV spectra. In
0.1% acetic acid containing 50 mM magnesium chloride,
Antibody A exhibited a circular dichroism spectrum typical of
immunoglobulins in the near-UV (Fig. 5A) with a distinct
positive peak at 295 nm (tryptophan) and a negative CD at
Fig. 2. Antibody A solution in 0.1% acetic acid containing 50 mM magnesium
chloride (0.5 mg/ml) analysed by FFF. Separation started at 5 min elution time.
Molecular weights (thick lines, left scale) are superimposed on the UV (dots)
and 90 light scattering (dashes) signals. The monomer peak at 12.5 min shows a
molecular weight of 170 kDa, whereas the aggregates peak at 18.5 min exhibits
molecular weight ranging from 1 to 2 million Daltons. Few antibody fragments
can be seen at 9 min. The inset provides a magnified view of the chromatogram
area situated between the main peaks, revealing a smaller peak that can
reasonably be interpreted as a dimer. At 15 min, the low signal-to-noise ratio
prohibits a precise determination of the species molecular weight.
Fig. 3. Antibody A solution in 0.1% acetic acid containing 50 mM magnesium
chloride (0.5 mg/ml) analysed by FFF, using phosphate buffer as eluent.
Separation started at 4 min elution time. Molecular weights (thick line, left scale)
are superimposed on the UV (dots) and 90 light scattering (dashes) signals. The
aggregate's peak, expected at 17.5 min, is not present on the chromatogram. The
monomer peak remained unchanged, but the contact of the protein sample with
phosphate buffer during elution was sufficient to modify the properties of the
sample.
149 B. Demeule et al. / Biochimica et Biophysica Acta 1774 (2007) 146153
270 nm (tryptophan and tyrosine) [16]. Upon dilution in
phosphate buffer, the positive peak at 295 nm disappeared and
the CD is clearly distorted with light scattering effects giving a
CDcontribution where there is no absorption (450325 nm); the
associated absorption shows turbidity and absorption flattening
typical of aggregated species. The 0.1% acetic acid phosphate
buffer induced changes were found to be reversible. In the far-
UV (Fig. 5B), the CD spectra in both buffers are dominated by a
strong -sheet component. At 228 nm, where the second
tryptophan absorption envelope is known to peak, a slight
difference was detectable between the spectra of Antibody A
dissolved in 0.1% acetic acid containing 50 mM magnesium
chloride and Antibody A dissolved in phosphate buffer. The
differences observed in the near- and far-UV spectra occurred at
wavelengths of tryptophan absorption [16]. Therefore, the CD
spectral changes can be best rationalized as being due to small
local conformational changes associated with aromatic amino
acid side chains. The global secondary structure of Antibody A
remains relatively unchanged, dominated by -sheet in both
solvents.
3.4. Characterization of Antibody A by fluorescence
spectroscopy
Tryptophan fluorescence and fluorescence lifetime measure-
ments were performed on Antibody A dissolved in 0.1% acetic
acid containing 50 mM magnesium chloride as well as in
10 mM phosphate buffer pH 7.1. Both formulations had a
concentration of 0.7 mg/ml, as in the circular dichroism
Fig. 5. Circular dichroism spectra of 0.7 mg/ml Antibody A in 0.1% acetic acid containing 50 mM magnesium chloride (pH 5, solid line) and in 10 mM
phosphate buffer pH 6.9 (dotted line). (A) Near-UV circular dichroism and absorbance spectra (1-cm cuvette). In phosphate buffer (dotted line), the positive
tryptophan peak at 295 nm disappeared. (B) Far-UV circular dichroism and absorbance spectra (0.5-mm cuvette). The far-UV circular dichroism spectrum is
indicative of a dominant -sheet structure. At 228 nm, small differences can be observed between the two solutions confirming local conformational changes
associated with tryptophan side chains.
Fig. 4. Antibody A solution in 10 mM phosphate buffer, pH 7.1 (0.6 mg/ml),
analysed by FFF. Separation started at 3.5 min elution time. R.M.S. radii (thick
dots, left scale) are superimposed on the UV (dots) and 90 light scattering
(dashes) signals. The first species to elute (in steric hyperlayer mode) are the
aggregates at 4.5 min, showing radii of 100 nm and higher. The largest
aggregates, visible by fluorescence microscopy, could not be measured by FFF.
The monomer peak eluted at 9.5 min. Monomers constitute 80% of the sample.
150 B. Demeule et al. / Biochimica et Biophysica Acta 1774 (2007) 146153
measurements. The tryptophan fluorescence intensity was
higher in phosphate buffer, indicating a more hydrophobic
environment (Fig. 6). As expected, the 90 light scattering
signal was higher in the aggregating phosphate buffer solution
than in 0.1% acetic acid containing 50 mM magnesium chloride
(Fig. 7). Mean fluorescence lifetimes were similar in phosphate
buffer and in 0.1% acetic acid containing 50 mM magnesium
chloride (Table 1). Anisotropy and rotational correlation time
measurements provided insight into the extent and speed of
depolarization of Antibody A fluorescence in both buffers.
Fig. 8 shows that Antibody A dissolved in 0.1% acetic acid
containing 50 mM magnesium chloride exhibited a lower
anisotropy than the aggregated solution in phosphate buffer.
The lower anisotropy indicates a more significant depolariza-
tion and hence a greater mobility of the tryptophan residues in
0.1% acetic acid containing magnesium chloride. The rotational
correlation times were 5.4 ns in 0.1% acetic acid containing
50 mM magnesium chloride and 3.0 ns in phosphate buffer
(Table 2). These small correlation times indicate that the
fluorophore (tryptophan) has an independent motion from the
rest of the immunoglobulin. If the observed correlation time was
related to the motion of the whole antibody, it would be >50 ns,
due to the high molecular weight of the protein [18]. The
fundamental anisotropy values were 0.14 in acetic acid
containing magnesium chloride and 0.21 in phosphate buffer,
corresponding to angles of 40.9 and 34.5 respectively
between the excitation and emission transition electric dipoles
(Table 2). These data show that in acetic acid containing
magnesium chloride the tryptophan residues have a greater
mobility than in phosphate buffer; the amplitude of depolariza-
tion is higher but occurs at a slower rate. In acetic acid
containing magnesium chloride, most of the depolarization
occurs between the fluorophore excitation and the fluorescence
emission, as indicated by the lower fundamental anisotropy. In
phosphate buffer, the tryptophan residues exhibit smaller but
Fig. 6. Steady-state fluorescence emission spectra of 0.7 mg/ml Antibody A in
0.1% acetic acid containing 50 mM magnesium chloride (pH 5, solid line) and
in 10 mM phosphate buffer pH 7.1 (dotted line). Background fluorescence was
subtracted from both spectra. Antibody A exhibited a stronger fluorescence in
phosphate buffer, indicating a more hydrophobic environment of the tryptophan
residues.
Fig. 7. 90 light scattering. Antibody A aggregation in 0.1% acetic acid
containing 50 mM magnesium chloride (solid line) and in 10 mM phosphate
buffer pH 7.1 (dotted line). As expected from the microscopy data, Antibody A
showed a strong aggregation in phosphate buffer.
Table 1
Fluorescence lifetime parameters for Antibody A dissolved in two different
buffers: (A) 0.1% acetic acid containing 50 mM magnesium chloride and (B)
10 mM phosphate buffer pH 7.1
Antibody sample
1
a

1
(ns)
2

2
(ns)
3

3
(ns)
2

b
(ns)
(A) 0.29 0.51 0.45 1.77 0.26 4.48 1.04 3.17
(B) 0.29 0.60 0.48 1.49 0.24 4.65 1.02 3.16
Concentration was 0.7 mg/ml.
a
The best fit to fluorescence lifetime data was obtained with a three-
exponential model where
1
,
2
and
3
are the decay times and
1
,
2
and
3
the pre-exponential factors.
b
Average fluorescence lifetime (see Materials and methods).
Fig. 8. Steady-state fluorescence anisotropy of Antibody A in 0.1% acetic acid
containing 50 mM magnesium chloride (solid line) and in 10 mM phosphate
buffer pH 7.1 (dotted line). Excitation was set at 280 nm. Even though the
difference between the samples was small (0.01 at 336 nm), tryptophan residues
were clearly more mobile (lower anisotropy) in acetic acid containing
magnesium chloride.
151 B. Demeule et al. / Biochimica et Biophysica Acta 1774 (2007) 146153
faster movements and are located in a more hydrophobic
environment.
4. Discussion
The work presented in this paper shows that proteins can form
several types of aggregates through non-covalent bonds.
Analytical methods have been adapted to study the variety of
aggregates that can be found in a protein sample. In the case of
Antibody A, two types of aggregates formed by weak
interactions were characterized by specific methods. Table 3
summarizes the results obtained during the characterization of
Antibody A. In phosphate buffer, Antibody A formed large
aggregates resulting in a relatively turbid solution. The large and
fragile aggregates formed by Antibody A in phosphate buffer
were efficiently detected by fluorescence microscopy after
staining with Nile Red (Fig. 1) and by dynamic light scattering.
Fluorescence microscopy showed aggregates 3.18 m in
diameter matching the dynamic light scattering values of
2.2 m. The difference may be due to the sampling that was
made at different stages of protein aggregation. Dynamic light
scattering measurements were done immediately after sample
preparation and presented a light scattering mean value.
Microscopy measurements were performed several minutes
after sample preparation and better represent the aggregation
steady-state of individual aggregates. The easy disruption of
Antibody A aggregates in phosphate buffer mitigates against
their characterization with separation methods such as FFF.
Even though Fig. 4 shows the presence of aggregates species
>200 nm in diameter after 4 min of elution, the FFF data do not
showthe 3 maggregates observed by fluorescence microscopy.
Shear stresses and possibly contact with the cellulose membrane
may have contributed to the disruption of the largest aggregates.
Consequently, a complete evaluation of aggregate population
was difficult, since a part of the 80% monomers shown in Fig. 4
could be generated by the disruption of looser aggregates. Fig. 1
(b) shows a relatively high Nile Red fluorescence intensity in the
background, suggesting that smaller aggregates are present in
the solution. If there were only large aggregates, the background
would be dark since free Nile Red is not fluorescent in water.
The 12 million Dalton aggregates present when Antibody
A was dissolved in 0.1% acetic acid containing magnesium
chloride are too small to be detected by fluorescence
microscopy. Static light scattering detected 2 million Dalton
aggregates even though the solution was perfectly clear by eye.
The percentage of aggregated protein could however not be
determined without a prior separation step due to the dominant
light scattering signal of the larger species, masking the
monomer signal. Asymmetrical flow field-flow fractionation
(FFF) provided a separation step allowing the subsequent
analysis of the fractions by static light scattering. FFF confirmed
the presence of 12 million Dalton protein aggregates,
representing 10% of the total protein mass (Fig. 2). The
stability of the 12 million Dalton aggregates in acetic acid
containing magnesium chloride was demonstrated by the FFF
experiments performed with a long focussing time. Despite
being submitted to an additional stress during focussing, the
Antibody A solution in acetic acid containing magnesium
chloride remained unchanged. The different types of aggregates
formed by a single immunoglobulin demonstrate that the
combination of several analytical methods is essential for the
characterization of protein solutions.
Protein aggregation often involves conformational changes.
Spectroscopic characterization of protein samples by fluores-
cence and circular dichroism spectroscopy assesses local
environment and conformational changes in the protein [9].
CD indicates that there is little or no change in the global
-sheet secondary structure on aggregation of Antibody A.
Antibody A showed an increase in tryptophan fluorescence in
phosphate buffer (Fig. 6), which implies an increase in
hydrophobicity around tryptophan residues. Tryptophan
Table 2
Fluorescence parameters for Antibody A dissolved in two different buffers: (A)
0.1% acetic acid containing 50 mM magnesium chloride and (B) 10 mM
phosphate buffer pH 7.1
Antibody sample
em
max (nm) F
a

b
(ns)
c
(ns) A
d
A
0
e

f
(A) 336 1 3.17 5.40 0.090 0.143 40.9
(B) 336 1.18 3.16 2.96 0.100 0.208 34.5
Concentration was 0.7 mg/ml.
a
Normalized steady-state fluorescence intensity at the emission maximum.
b
Average fluorescence lifetime (see Materials and methods).
c
Rotational correlation time.
d
Steady-state anisotropy at the emission maximum.
e
Fundamental anisotropy.
f
Displacement angle between the excitation and emission transition electric
dipoles.
Table 3
Presentation of the main methods used to characterize Antibody A
Analytical method Information obtained for Antibody A dissolved in (A) Information obtained for Antibody A dissolved in (B)
Nile Red staining, fluorescence microscopy Absence of large aggregates. Presence of large aggregates.
Asymmetrical flow field-flow fractionation (FFF) Monomers constitute 88% of the sample.
Presence of 10% aggregates (12 million Daltons).
Aggregates were stable.
Presence of aggregates.
The large aggregates observed by microscopy are fragile
and were disrupted.
Circular dichroism Dominant beta-sheet component. Small conformational changes observed when
compared to (A).
Tryptophan fluorescence Tryptophan residues were in a more hydrophilic
environment than in (B).
Tryptophan residues were in a
more hydrophobic environment than in (A).
Tryptophan fluorescence anisotropy and
time-resolved fluorescence
Compared to (B), tryptophan residues were more
mobile but exhibited a lower rate of depolarization.
Compared to (A), tryptophan residues were less
mobile but exhibited a higher rate of depolarization.
The most important information provided by each method is described. (A) 0.1% acetic acid containing 50 mM magnesium chloride and (B) 10 mM phosphate buffer.
152 B. Demeule et al. / Biochimica et Biophysica Acta 1774 (2007) 146153
residues may be situated in a hydrophobic pocket, possibly at
the interface between aggregated proteins. The change in
conformation around tryptophan residues was further confirmed
by circular dichroism (Fig. 5). Anisotropy and time-resolved
fluorescence measurements (Table 2) were also able to detect
differences between Antibody A dissolved in phosphate buffer
and Antibody A dissolved in 0.1% acetic acid containing
50 mM magnesium chloride. The higher rigidity of tryptophan
residues and their presence in a more hydrophobic environment
could be explained by the involvement of tryptophan residues in
the interactions between aggregated proteins.
Measurements of fluorescence anisotropies, lifetimes and
rotational correlation times of Antibody A enabled the
calculation of the fundamental fluorescence anisotropy value
of the antibody in each buffer. The fundamental fluorescence
anisotropy values were different in the two buffers used. The
amplitude (anisotropy) and speed (rotational correlation time)
of tryptophan mobility were therefore different. Caution is
needed when interpreting time-resolved fluorescence data,
especially if only partial measurements are available (e.g.
only fluorescence lifetime and anisotropy).
In order to successfully assess protein aggregation, changes
in the local environment of the proteins need to be minimized.
In particular, the solvent or buffer used during the analytical
experiment should be the same as the protein formulation
buffer. Field-flow fractionation analysis underlined the impor-
tance of performing aggregate characterization with minimal
changes in the local environment of the protein. One of the
advantages of using FFF for the analysis of protein solutions is
that the eluent can easily be adapted to the studied formulation.
Fig. 3, compared to Fig. 2, shows that the aggregate content of
Antibody A can change depending upon the eluent.
Assessment of protein aggregation should be performed by
combining the results of several analytical methods, tailored to
fit the particular needs of aggregate characterization.
Acknowledgement
We thank Novartis Pharma AG for providing Antibody A.
References
[1] G. Walsh, Pharmaceutical biotechnology products approved within the
European Union, Eur. J. Pharm. Biopharm. 55 (2003) 310.
[2] J.F. Carpenter, B.S. Kendrick, B.S. Chang, M.C. Manning, T.W. Randolph,
Inhibition of stress-induced aggregation of protein therapeutics, Methods
Enzymol. 309 (1999) 236255.
[3] A. Braun, L. Kwee, M.A. Labow, J. Alsenz, Protein aggregates seem to
play a key role among the parameters influencing the antigenicity of
interferon alpha (IFN-alpha) in normal and transgenic mice, Pharm. Res.
14 (1997) 14721478.
[4] S. Hermeling, D.J. Crommelin, H. Schellekens, W. Jiskoot, Structure
immunogenicity relationships of therapeutic proteins, Pharm. Res. 21
(2004) 897903.
[5] M. Bucciantini, E. Giannoni, F. Chiti, F. Baroni, L. Formigli, J. Zurdo, N.
Taddei, G. Ramponi, C.M. Dobson, M. Stefani, Inherent toxicity of
aggregates implies a common mechanism for protein misfolding diseases,
Nature 416 (2002) 507511.
[6] T. Arvinte, Analytical methods for protein formulations, in: W. Jiskoot, D.J.
Crommelin (Eds.), Methods for Structural Analysis of Protein Pharma-
ceuticals, AAPS Press, Arlington, VA, 2005, pp. 661666.
[7] D.K. Clodfelter, M.A. Nussbaum, J. Reilly, Comparison of free solution
capillary electrophoresis and size exclusion chromatography for quantitat-
ing non-covalent aggregation of an acylated peptide, J. Pharm. Biomed.
Anal. 19 (1999) 763775.
[8] W. Fraunhofer, G. Winter, The use of asymmetrical flow field-flow
fractionation in pharmaceutics and biopharmaceutics, Eur. J. Pharm.
Biopharm. 58 (2004) 369383.
[9] W. Colon, Analysis of protein structure by solution optical spectroscopy,
Methods Enzymol. 309 (1999) 605632.
[10] B. Demeule, R. Gurny, T. Arvinte, Detection and characterization of
protein aggregates by fluorescence microscopy, Int. J. Pharm. (2006).
doi:10.1016/j.ijpharm.2006.08.024.
[11] A.F. Drake, The measurement of electronic absorption spectra in the
ultraviolet and visible, in: C. Jones, B. Mulloy, A.H. Thomas (Eds.),
Microscopy, Optical Spectroscopy and Macroscopic Techniques, vol. 22,
Humana Press Inc, Totowa, NJ, 1994, pp. 173182.
[12] J.D. Ingle, S.R. Crouch, Molecular scattering methods, in: J.D. Ingle, S.R.
Crouch (Eds.), Spectrochemical Analysis, Prentice-Hall, Upper Saddle
River, 1988, pp. 494524.
[13] J. Demeester, S.S. De Smedt, N.N. Sanders, J. Haustraete, Light
scattering, in: W. Jiskoot, D.J. Crommelin (Eds.), Methods for Structural
Analysis of Protein Pharmaceuticals, AAPS Press, Arlington, VA, 2005,
pp. 245275.
[14] P.M. Tessier, A.M. Lenhoff, Measurements of protein self-association as a
guide to crystallization, Curr. Opin. Biotechnol. 14 (2003) 512516.
[15] H. Bajaj, V.K. Sharma, D.S. Kalonia, Determination of second virial
coefficient of proteins using a dual-detector cell for simultaneous
measurement of scattered light intensity and concentration in SEC-
HPLC, Biophys. J. 87 (2004) 40484055.
[16] T. Arvinte, T.T. Bui, A.A. Dahab, B. Demeule, A.F. Drake, D. Elhag, P.
King, The multi-mode polarization modulation spectrometer: part 1:
simultaneous detection of absorption, turbidity, and optical activity, Anal.
Biochem. 332 (2004) 4657.
[17] Time-domain lifetime measurements, in: J.R. Lakowicz (Ed.), Principles of
Fluorescence Spectroscopy, Kluwer Academic/Plenum Publishers, New
York, 1999, pp. 95140.
[18] Fluorescence anisotropy, in: J.R. Lakowicz (Ed.), Principles of Fluores-
cence Spectroscopy, Kluwer Academic/Plenum Publishers, New York,
1999, pp. 291319.
[19] C.R. Cantor, P.R. Schimmel, Biophysical Chemistry Part 2: Techniques for
the Study of Biological Structure and Function, W.H. Freeman, New York,
1980.
153 B. Demeule et al. / Biochimica et Biophysica Acta 1774 (2007) 146153

Вам также может понравиться