Sedimentation velocity, circular dichroism spectroscopy, light scattering, and dynamic light scattering are effective methods for demonstrating the comparability of protein conformation and aggregation between product lots: 1) Sedimentation velocity precisely measures sedimentation coefficients, which are highly sensitive to small conformational differences, and can detect aggregates as low as 0.2%; 2) Circular dichroism spectroscopy provides spectral fingerprints of secondary and tertiary structure that are sensitive to conformational changes, and can quantify thermal stability; 3) Light scattering coupled with size-exclusion chromatography identifies aggregate masses and distinguishes conformational variants; 4) Dynamic light scattering readily detects large aggregates above 0.01% or very large

1. 1
Some Biophysical Methods to Demonstrate Comparability of Conformation and
Aggregation
John S. Philo* and Tsutomu Arakawa, Alliance Protein Laboratories
The bioactivity of protein or RNA-based pharmaceuticals depends on their conformation (secondary and
tertiary structure). They also often form inappropriate intermolecular interactions (aggregates), and such
formation of oligomers/aggregates may alter activity, PK, immunogenicity, etc. Therefore
demonstrating comparability requires good methods to assess both conformation and aggregation.
Sedimentation velocity can be a powerful method for demonstrating comparability. Sedimentation
coefficients are highly sensitive to small differences in molecular conformation, and thus provide an
excellent quantitative measure of comparability. Sedimentation velocity is also a powerful tool for
assessing heterogeneity, such as the presence of aggregates. Data for antibodies will show that recently
developed sedimentation velocity techniques can detect and quantitate aggregates at levels well below
1%, while covering a far wider range of masses than is possible by SEC.
Circular dichroism (CD) spectroscopy is another useful tool for measuring biomolecular conformation.
Protein spectra in the far-UV region are sensitive to secondary structure, while their near-UV spectra
depend on the tertiary structure around aromatic residues and disulfides. Thus both spectral regions
provide a spectral ‘fingerprint’ for showing comparability of conformation. As we will show, however,
the often-neglected near-UV spectrum can be more sensitive to subtle conformational differences. CD is
also very useful for measuring thermal unfolding transitions, which can be measured quantitatively for
lot-to-lot comparisons.
Light scattering techniques provide another comparability tool. When combined with SEC, classical
light scattering identifies the mass of each aggregate peak, in an absolute manner, independent of
molecular shape. Dynamic light scattering directly determines the hydrodynamic size of
biopharmaceuticals, which depends on both mass and conformation. It is also an excellent method for
detecting trace amounts (<0.01%) of very large aggregates.

2. 2
Sedimentation velocity
• The most quantitative way to compare molecular
conformation
– sedimentation coefficients are highly sensitive to small differences
in conformation
– they can be measured to a precision of 0.5% or better
• An excellent tool for measuring heterogeneity, such as that
due to aggregates
– aggregates can be detected at levels of ~0.2%
– covers a much broader range of sizes than any SEC column
• Can usually be done directly in formulation buffers

3. 3
Comparability testing by sedimentation velocity: two manufacturing processes
for a monoclonal antibody give main peaks with identical conformation, but
aggregate levels vary lot-to-lot
2 4 6 8 10 12
0.0
0.1
0.2
0.3
lot B is 94.0% main peak
(96.7% by SEC)
g(s*)(AU/Svedberg)
sedimentation coefficient (Svedbergs)
0.0
0.1
0.2
0.3
0.4
0.5
equal sedimentation coefficients
prove conformation of main peak
is the same for both lots
data
single species fit
lot A is 98.0% main peak
(99.0% by SEC)

4. 4
A new high-resolution analysis method is excellent for
resolving and quantifying minor components
0 2 4 6 8 10 12 14 16 18 20
0.0
0.4
0.8
1.2
1.6
2.0normalizedc(s)
sedimentation coefficient (Svedbergs)
0 2 4 6 8 10 12 14 16 18 20
0.00
0.05
0.10
0.7%
1.1%1.7%
4.7%
20X expanded

5. 5
0 2 4 6 8 10 12 14 16 18 20 22 24
0.0
0.2
0.4
0.6
0.8
1.0
heptamer,0.1%
hexamer,0.4%
pentamer1.4%
tetramer5.3%
trimer14.6%
dimer30.6%
main peak (monomer), 45.5%
?HLhalfmolecule,0.8%
?freelightchain,1.4%
c(s),normalized(totalarea=1)
sedimentation coefficient (Svedbergs)
This highly-stressed antibody sample shows a well-resolved series
of oligomers up to heptamer, as well as fragments.

6. 6
Circular dichroism (CD) spectroscopy
• Spectra provide a ‘fingerprint’ for comparing conformation
– ‘far-UV’ spectra (190-240 nm) are sensitive to secondary structure
– ‘near-UV’ spectra (240-340 nm) are sensitive to local tertiary
structure around aromatic residues and disulfide bonds
• In our experience near-UV spectra are often more sensitive
to subtle conformational differences than far-UV spectra
• Temperature-dependent spectra can quantitate and
compare thermal stability
– variations in conformation will lead to variation in the thermal
unfolding profile
– CD studies can usually be done at lower concentration than DSC,
reducing problems with irreversible unfolding and thus giving data
that is more relevant

7. 7
Differences in far-UV CD for natural vs. recombinant forms
of an enzyme (which also have different enzymatic activity)
210 220 230 240 250 260
-12000
-10000
-8000
-6000
-4000
-2000
0
MolarEllipticity
W avelength (nm)
natural
recombinant

8. 8
Differences in near-UV circular dichroism between stable and
unstable lots of a monoclonal antibody
240 250 260 270 280 290 300 310 320 330 340
-200
-150
-100
-50
0
50MolarEllipticity
Wavelength (nm)
lot 1
lot 2

9. 9
Quantitative analysis of thermal stability from CD data using
proprietary A.P.L. software
30 40 50 60 70
-80
-75
-70
-65
-60
-55
-50
-45
-40
melting
end (95%)
60.1 ± 0.5 °C
melting
onset (5%)
36.6 ± 0.4 °C
Tm
= 47.94 ± 0.14 °C,
∆H = 51.4 ± 3.2 kcal/mol
data
fit (2-state model)
molarellipticity@220nm
Temperature (°C)

10. 10
Classical light scattering used on-line with SEC
• identifies solution mass of individual peaks from SEC
• mass determination is absolute, independent of molecular shape
or interactions with the column matrix
• for comparability studies primary use is demonstrating that the
mass of aggregates is comparable (same oligomers are present)
• can measure degree of conjugation for PEGylated proteins
• can distinguish multiple conformations with same mass
– can distinguish different forms of dimer (see next slide)
– can distinguish dimer from unfolded monomer (which usually elute at
similar positions)
light scattering
detector
absorbance
detector
refractive index
detectorsize-exclusion
column
injectorpumpsolvent

11. 11
0.0
5
1.0x10
5
2.0x10
5
3.0x10
5
4.0x10
12.0 13.0 14.0 15.0 16.0
MolecularWeight
Volume (mL)
81221H_1
dimerextended
dimer
SEC with on-line classical light scattering shows an
antibody sample contains only monomer and dimer, but
two conformations of dimer are present

12. 12
Analysis of products from a 5:1 reaction of 5 kDa
PEG with ribonuclease A (13.7 kDa)
28.6 kDa
3 PEG + 1 RNAse
23.9 kDa
2 PEG + 1 RNAse
18.3 kDa
1 PEG + 1
RNAse
excess PEG

13. 13
Dynamic light scattering
• measures Brownian motion of molecules, which is directly
related to their hydrodynamic size (which depends on both
mass and conformation)
• data are usually converted to a distribution of
hydrodynamic radius graph
• excellent for detecting trace amounts (0.01% to <1 p.p.m.)
of very large aggregates
• covers extremely large range of sizes at one time
• works in wide range of solvent conditions
• useful for continuous monitoring of aggregation at elevated
temperatures (accelerated stability)

14. 14
DLS can easily detect large aggregates (~100 nm) at levels of
~0.01% and very large ones (~1 µm) at ~1 ppm
0.1 1 10 100 1000 10000
0
4
8
12
16
13% intensity
1.8% mass
3.5% intensity
<1 ppm mass
main peak
44% of intensity
98.1% of mass
40% intensity
0.06% mass
%intensity
hydrodynamic radius (nm)
0.1 1 10 100 1000 10000
0
5
10
15
20
main peak
72% of intensity
99.95% of mass
aggregates
28% of intensity
0.05% of mass
%intensity
hydrodynamic radius (nm)
data for “good” lot data for “bad” lot

15. 15
What about detecting smaller aggregates with DLS? They are
generally not resolved as separate species, but they may shift the
distribution toward larger sizes (and increase total scattered light)
4 6 8 10 12
0
2
4
6
8
10
12
14
"good" sample
"bad" sample,
>10% aggregates
%Intensity
Hydrodynamic radius, nm