1.
Analyzing Aggregates by Sedimentation
Velocity and Light Scattering
John Philo
Director of Biophysical Chemistry
© copyright 2006, Alliance Protein Laboratories, Inc.
Images or text may not be reproduced without permission.

2.
Outline
Quickly review some basic facts about
aggregate sizes and types
Basic principles and application examples
for 3 methods
1. sedimentation velocity
2. classical light scattering used with SEC
3. batch-mode dynamic light scattering

3.
The word “aggregate” covers a wide spectrum
of types and sizes of associated states
1. rapidly-reversible non-covalent small
oligomers (dimer, trimer, tetramer…)
2. irreversible non-covalent oligomers
3. covalent oligomers (e.g. disulfides)
4. “large” aggregates (> 10-mer)
could be reversible if non-covalent
5. “very large” aggregates (diameter ~50 nm
to 3 μm)
could be reversible if non-covalent
6. visible particulates
probably irreversible

4.
Aggregates have a spectrum of lifetimes
rates of non-covalent association and dissociation (half-
times) can vary from milliseconds to days
metastable oligomers with dissociation rates of hours to
days occur fairly frequently
for an antibody example see J.M.R. Moore et al. (1999)
Biochemistry 38: 13960-13967
see also Philo, J.S. (2006) AAPS Journal, in press
many common analytical methods will detect only the
longer-lived species
it may take hours to days for a protein to re-equilibrate
its association after a change in concentration, solvent
conditions or temperature

5.
Our analytical challenge
1. Any protein sample may contain aggregates
with a wide range of sizes, types, and lifetimes
2. Any one analysis method may not detect all
the aggregate sizes or types that are present
3. The measurement itself may perturb the
aggregate distribution that was initially present
dilution may dissociate reversible aggregates
change of solvent conditions may dissociate or increase
aggregates
adsorption or filtration effects may remove aggregates

6.
Sedimentation velocity

7.
The fundamentals of sedimentation velocity
6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 7.0
Radius (cm)
-0.1
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
Absorbance
6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 7.0
Radius (cm)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Absorbance
centrifugal
force
diffusion
← meniscus
The sedimentation coefficient
is determined from the
boundary motion over time. It
depends on both molecular
weight and molecular shape.
cell base →
friction
←regionofsolute
depletion
boundary

8.
High resolution analysis of a highly stressed antibody
sample resolves 6 aggregate peaks plus 2 fragments
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)

9.
The peril: c(s) distributions are also often
misunderstood
1. the effective resolution goes down as the fraction of
minor peaks goes down
2. the resolution you can achieve for a 150 kDa antibody
is much greater than for a 20 kDa cytokine
3. in general it is not possible to uniquely assign a
stoichiometry to each aggregate peak
4. the nature of the noise (variability) is very different
than in chromatography
5. for reversibly associating proteins the peaks probably
do not represent individual molecular species

10.
This interferon-β sample is 13.7% non-covalent aggregate;
by the standard SEC method it would be pure monomer
0 2 4 6 8 10
0
1
2
3
4
5
6
7
0 2 4 6 8 10
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
IFN-β in 5 mM glycine, pH 3, 86.3% main peak
c(s)
sedimentation coefficient (Svedbergs)

11.
0
1
2
3
0.0
0.5
1.0
1.5
0 8 16 24 32 40
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
no salt
c(s)
+50 mM NaCl
c(s)
+150 mM NaCl
c(s)
sedimentation coefficient (Svedbergs)
0 2 4 6 8 10
0.00
0.05
0.10
0.15
20X expanded
0 2 4 6 8 10
0.00
0.05
0.10
0.15
20X expanded
Adding NaCl to interferon-β formulations leads to a broad
distribution of non-covalent aggregates out to ~100-mers

12.
Strengths of sedimentation velocity
1. high resolution (often better than SEC)
2. covers very large range of masses in a single
experiment (much larger than SEC)
3. detects both covalent and non-covalent aggregates
4. generally can be done directly in formulation buffers
Tween and high levels of sugars do cause some
interference
5. little dilution of sample (~25%)
6. absolute method; requires no molecular standards
7. strong theoretical background; “first principles”
method

13.
Weaknesses of sedimentation velocity
1. low throughput (3-7 samples/day)
2. equipment and data analysis not automated like
HPLC; labor intensive
3. expensive equipment (~250-300 K$)
4. requires substantial training
5. never been validated for lot release
Sedimentation velocity can not replace SEC, but it
is an excellent tool to test whether SEC is missing
important features. It can also serve as a “gold
standard” to help improve SEC methods.

14.
Multi-angle classical laser light
scattering used on-line with SEC
(SEC-MALLS)

15.
Typical setup for size-exclusion chromatography
with on-line light scattering detection
light scattering
detector
absorbance
detector
refractive index
detector
size-exclusion
column
injectorpumpsolvent

16.
Getting molecular mass from static light
scattering: the basic idea
the light scattering signal is proportional to the
product c × M
we measure c simultaneously with a UV or RI
detector
then the ratio of the scattering to concentration
signals will be proportional to M
masses obtained this way are absolute, and
independent of conformation and elution position

17.
Demonstrating that scattering is independent of elution position
and molecular conformation: the ratio of LS to RI signals is the
same whether the protein is folded or unfolded
18 20 22 24 26 28
signal(arbitraryunits)
LS
RI
LS
RI
Native
RNase
Unfolded (reduced &
carboxymethlyated)
RNase
Retention Time (min)

18.
An example for an Fc-fusion protein: the aggregate signals are
much stronger in 90° scattering than in the UV chromatogram
scattering intensity RI
elution volume (ml)
5.0 6.0 7.0 8.0 9.0 10.0 11.0
relativescale
0.0
0.2
0.4
0.6
0.8
1.0

19.
“Oligomer hunting”: display the absolute molecular
weight from LS in units of monomers
6 7 8 9 10 11
0
1
2
3
4
5
6
7
8
9
10
massratiorelativetomainpeak
elution volum e (m l)

20.
This antibody sample has traces of dimer and trimer
12 13 14 15 16 17 18
Elution Volume (ml)
0.0
1.0
2.0
3.0
4.0
5.0
relativemassfromLS/UV
relative mass (LS/UV) UV (arb units)

21.
A different lot contains more higher oligomers, and they
are so sticky that even dimer is no longer resolved
11 12 13 14 15 16 17 18
Elution Volume (ml)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
relativemassfromLS/UV
relative mass (LS/UV) UV (arb units)

22.
volume (ml)
5.0 6.0 7.0 8.0 9.0 10.0
relativescale
0.0
0.2
0.4
0.6
0.8
1.0
This highly stressed sample of a VaxGen test antigen
showed high levels of an SEC peak eluting near the
position expected for a dimer
monomer
dimer?
large
aggregates

23.
However SEC-MALLS immediately shows that alleged
aggregate is actually an altered form of monomer!
molar mass vs. volume
volume
5.0 6.0 7.0 8.0 9.0 10.0
molarmass(g/mol)
5
1.0x10
6
1.0x10
7
1.0x10

24.
Strengths of SEC + classical LS
1. absolute molecular mass, independent of
conformation or elution position
2. gives us at least an average mass for the
“aggregate” fraction near the exclusion limit
3. helps tell us whether our chromatography is really
working properly
4. high throughput, low cost (comparable to the HPLC
it is used with), fairly easy
5. absolute method; requires no molecular standards
6. strong theoretical background; “first principles”
method

25.
Weaknesses of SEC + classical LS
1. it inherits all the problems of SEC (change in
aggregate distribution from dilution, change in buffer,
adsorption/filtration, etc.)
2. while it is very sensitive to high MW aggregates,
quantitation of % by weight still relies on the
concentration detector (RI or UV)
3. particles shed from columns may obscure the elution
region near the column’s exclusion limit
4. good signal/noise may require larger injection amounts
than are normally used in standard SEC

26.
Batch-mode dynamic light scattering
(DLS)
also known as quasi-elastic light
scattering (QELS) or photon
correlation spectroscopy (PCS)

27.
One particularly vexing type of aggregation is “snow”
(a.k.a. “white amorphous material” [WAM] or “floaters”)
• may only appear after many months
• often a nucleation-controlled reaction
• often ≤ 0.01% of total protein

28.
⇒
When this happens our valuable therapeutic
protein can only be used for…
Dynamic scattering is one of the few tools that may be
able to detect the precursors that eventually form ‘snow’

29.
Dynamic light scattering: the basic idea
1. In dynamic scattering we measure the
fluctuations in scattering intensity (~100 ns to
30 ms)
2. The time scale of those fluctuations depends
on the diffusion coefficient of the
macromolecule, which in turn depends on its
size
3. As in classical LS, the scattering intensity is
proportional to M, so the sensitivity to very
large aggregates is very high

30.
Typically the data are transformed into a distribution of
hydrodynamic radius; this distribution shows 3 peaks
2.16 nm, 79.0% of intensity
92.3 nm,
13.3% of intensity
6.58 nm,
7.8% intensity
0.015% wt.
99.1% by weight
0.9% wt.

31.
Two key weaknesses of DLS
1. Low resolution
two species are not resolved as separate peaks
unless their radii differ ~2-fold (~8-fold in mass)
consequently DLS is generally not useful to detect
or quantify small oligomers (dimer-octamer)
2. Poor quantitation of weight fractions
Usually at best the reproducibility of weight
fractions is only +/- a factor of 2
There is no universally-accepted standard
algorithm to calculate weight fractions; different
methods can give quite divergent results

32.
Here is an example for a small peptide that
forms visible thread-like particles
7.9% intensity,
99.998% by weight
92.1% intensity,
0.002% by weight
species in the ~20-400
nm size range are often
precursors or nuclei for
formation of visible
particulates

33.
In our hands DLS has been the most effective
tool for detecting precursors of visible particulates
1. Useful for qualitative assessment of different
formulations, ‘good’ vs. ‘bad’ lots
2. Useful to track where in the manufacturing process
damage to the protein is occurring
in one case tracked to specific pump
in another case to viral filtration step
3. Useful to detect contaminant particles that can serve
as nuclei onto which protein aggregates
(heterogeneous nucleation)
silicones
glass particles from vials
vacuum pump oil from lyophilizers

34.
Strengths of DLS
1. high sensitivity to large aggregates that may be
immunogenic and/or precursors to visible
particulates
2. covers an enormous range of sizes in one analysis
(range of mass > 109)
3. done at equilibrium; theoretically senses all forms of
aggregates
4. batch mode
no dilution
no change of solvent conditions
no loss of species to frit or column matrix