This document discusses different methods for analyzing protein aggregates of various sizes and types, including size exclusion chromatography (SEC), analytical ultracentrifugation, and light scattering. It notes some limitations of SEC, including that changes in solvent conditions can alter aggregate distributions and larger aggregates may be unresolved. Analytical ultracentrifugation methods like sedimentation velocity and equilibrium are presented as alternatives that are sensitive to all aggregate types and sizes. Sedimentation velocity provides high resolution separation while equilibrium can determine equilibrium constants and stoichiometry of reversible associations. On-line light scattering with SEC is also discussed.
2. Outline
Different types of oligomers and
aggregates
Some problems with SEC
Applications and advantages/drawbacks
of:
1. sedimentation velocity
2. sedimentation equilibrium
3. SEC with on-line classical light scattering
4. dynamic light scattering
Some recommended applications for these
biophysical methods
3. I can’t really cover this broad topic in 20 min
so I’m going to omit some background and
summaries but…
1. See poster P-30-W
2. An expanded presentation can be
downloaded from the APL web site at
http://www.ap-lab.com/further_reading.htm
3. Also see our web site [www.ap-lab.com] for
more method background and applications
4. 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
5. Whether aggregates are
“irreversible” or “reversible” depends
on the context
solvent components
salts, sugars, other excipients
organic modifiers (alcohols,
acetonitrile)
pH
temperature
how long you wait
6. The kinetics of non-covalent association
and dissociation (half-times) can vary
from milliseconds to days
metastable oligomers with lifetimes of
hours to days occur frequently
for an antibody example see J.M.R. Moore et al.
(1999) Biochemistry 38: 13960-13967
it may take hours to days for a protein to
re-equilibrate its association after a
change in concentration, solvent conditions
or temperature
7. Any given protein sample may contain multiple
aggregate forms with widely-varying properties
both covalent and non-covalent
different types of non-covalent:
both rapidly-reversible and irreversible
both rapidly- and slowly-dissociating
even the same size of oligomer may
exist in multiple forms
8. We must be aware that any given analysis
technique may not detect all the aggregate
sizes or types that are present
Separation methods (SEC, sedimentation
velocity) typically will not resolve individual
oligomer species for a system in rapidly-
reversible association equilibrium
for associations to oligomers larger than dimer,
from theory multiple peaks may be seen even for
infinitely-fast kinetics, but those peaks usually do
not represent single oligomers
what is detected may depend on the rates of
association and dissociation compared to the
speed of separation
9. The measurement technique itself may
perturb the distribution of aggregates that
was initially present
dissociation or loss of aggregates can be
caused by:
1. dilution
2.change of solvent conditions
3.adsorption to surfaces (e.g. column resin)
4.physical filtration (e.g. column frit)
5.physical disruption by shear forces
creation of new aggregates can be caused by:
1. change of solvent conditions
2.surface or shear-induced denaturation
3.concentration on surface of column resin (IEX)
← SEC
← SEC
← SEC
← SEC
← SEC
11. Some shortcomings of SEC
1. change in solvent may change the
aggregate distribution
2. limited range of sizes; larger
aggregates often unresolved
3. often does not detect rapidly-reversible
non-covalent association
4. aggregates may be filtered out by column
5. elution position not reliable indicator of
molecular mass
6. limited sensitivity; difficult to resolve and
detect species at or below ~0.1%
12. SEC issue #1: solvent-induced changes
Achieving good resolution and high recovery
often requires extreme solvent conditions
that can alter the distribution of non-
covalent aggregates
high ionic strength
addition of organic modifiers (alcohol,
acetonitrile)
I have seen cases where the SEC elution buffer
completely dissociated the non-covalent aggregates,
and also cases where it induced substantial amounts
of new aggregates.
13. 1. want to add organics or salts to improve
resolution, recovery, peak symmetry
2. want to pre-dilute the samples with the
elution buffer
3. want to validate the method by spiking
with pure aggregate
but you cannot a pure sample of non-covalent
aggregates
end up optimizing and validating for covalent
aggregates
“Good” chromatography is often in direct conflict with
good relevance for measuring non-covalent aggregates!
← this can cause false conclusions!
The things chromatographers are taught to
do to achieve “good” and robust methods
often exacerbate this problem!
14. 1. want to add organics or salts to improve
recovery and resolution
2. want to pre-dilute the samples with the
elution buffer
3. want to validate the method by spiking
with pure aggregate
but you cannot a pure sample of non-covalent
aggregates
end up optimizing and validating for covalent
aggregates
“Good” chromatography is often in direct conflict with
good relevance for measuring non-covalent aggregates!
← this can cause false conclusions!
The things chromatographers are taught to
do to achieve “good” and robust methods
often exacerbate this problem!
15. SEC issue #2: limited size range
Often the largest aggregates are not
resolved, and elute in a single peak or
shoulder at or near the exclusion limit
Therefore we often cannot tell whether the
large aggregate species present in different
formulations or different manufacturing lots
are actually similar in size
Thus SEC alone may not distinguish samples with
significant differences (e.g. immunogenicity)
16. SEC issue #5: incorrect masses
inaccuracy due to dependence on
molecular shape and/or undesirable
interactions with the column matrix
many, many examples of wrong stoichiometry
for native proteins
absolute accuracy may be unimportant;
often all we want is fraction main peak
But is that “dimer” peak really a dimer, or is it a
partially-unfolded monomer (which might be much
more immunogenic)?
18. Both AUC methods are sensitive to
all types of aggregates, but…
Sedimentation equilibrium is primarily a
thermodynamic tool for studying rapidly-
reversible self-association (equilibrium
constants and stoichiometry)
Sedimentation velocity is much more a
separation method and is particularly
useful for characterizing irreversible
(covalent or non-covalent) and relatively
stable reversible non-covalent aggregates
19. Both sedimentation methods are
“first principle” methods
based on fundamental physical laws
theory is well understood
true for dilute solutions; concentrations > 10
mg/ml become complex and difficult
require no standard molecules for
calibration
calibration is based only on fundamental units
of distance, time, and temperature
21. 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
shape.
cell base →
friction
←regionofsolute
depletion
boundary
22. 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)
23. Comparability of a monoclonal antibody; detecting
aggregate peaks at levels below 0.05%
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
0
1
2
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
0.00
0.01
0.02
normalizedc(s)
sedimentation coefficient (Svedbergs)
0.42%
0.10%
0.03%
0.30%
lot 1
lot 2
X 100
0.95%
0.05%
0.07%
24. This interferon-β sample is 13.7% non-covalent
aggregate; by SEC (in 30% acetonitrile + 0.2% TFA) 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)
25. 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 non-covalent aggregates out to ~100-mers
26. We must study a range of concentrations
to check for reversible association; this is
a monomer-dimer-tetramer system
0 1 2 3 4 5 6 7 8 9 10
0
2
4
stock
4-fold dilution
normalizedc(s)
s20,w
(Svedbergs)
27. Strengths of sedimentation velocity
1. high resolution (generally 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
5. little dilution of sample (~25%)
6. strong theory; “first principles” method
28. Weaknesses of sedimentation velocity
1. low throughput; often 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
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.
30. 6.30 6.35 6.40 6.45 6.50 6.55 6.60
0.0
0.2
0.4
0.6
0.8
1.0
cell base
Absorbance
radius (cm)
diffusion
buoyancysedimentation
meniscus
The fundamentals of sedimentation equilibrium
The concentration
distribution depends
only on molecular
weight, independent
of shape!
← smaller sample size to reduce time to reach equilibrium →
31. Size-exclusion chromatography of a TNF homolog
6 8 10 12 14 16 18 20
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0absorbance(arb.units)
elution volume (ml)
ovalbumin
TNF homolog,
elutes exactly as
expected for a
17 kDa monomer
32. Linearized plot of equilibrium data for the TNF homolog
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6
-1.9
-1.8
-1.7
-1.6
-1.5
-1.4
-1.3
-1.2
-1.1
-1.0
-0.9
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
Ln(Absorbanceat230nm)
(r2 - ro
2 ) / 2
theoretical slope
for monomer
theoretical slope for trimer
33. Apparent mass vs. concentration for the soluble
extracellular domain of the atrial natriuretic peptide
receptor (monomer mass 58 kDa)
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22
Concentration (mg/ml)
60000
70000
80000
90000
100000
110000
ApparentMass(Da)
34. Global analysis of all 18 samples gives a good fit to a
monomer-dimer association model with Kd = 520 +/- 20
nM (∆G = -8570 +/- 25 cal/mol)
-1.0-0.9-0.8-0.7-0.6-0.5-0.4-0.3-0.2-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
(R^2 - Ro^2) / 2 (cm^2)
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
Absorbance
35. Strengths/weaknesses of sedimentation equilibrium
Strengths
1. equilibrium method; all
forms of association
are detected
2. wide choice of solvent
conditions
3. strong theory; “first
principles” method
Weaknesses
1. difficult to quantitate
irreversible aggregates
2. low throughput (9-21
samples/day typical)
3. expensive equipment
4. difficult data analysis
5. requires highly trained
personnel
36. How can we validate AUC methods for
non-covalent aggregates given we can’t
spike in pure species?
In part these methods are validated by
over 60 years of experience and tens of
thousands of publications
More significantly, we have a theory, and
we can create test data sets in silico
(including realistic noise levels) and submit
them to the analysis software to evaluate
the ability to quantitate various species
38. Typical setup for size-exclusion chromatography
with on-line light scattering detection
light scattering
detector
absorbance
detector
solvent pump injector
refractive index
detector
size-exclusion
column
39. 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
40. Demonstrating that scattering is independent of elution
position and molecular conformation: the ratio of LS to
RI signals is the same even for an unfolded protein
18 20 22 24 26 28
signal(arbitraryunits)
LS
RI
LS
RI
Native
RNase
Unfolded (reduced &
carboxymethlyated)
RNase
Retention Time (min)
41. An example for an Fc-fusion protein:
the aggregate signals are much stronger in 90°
scattering than in the UV chromatogram
4 5 6 7 8 9 10 11 12
Elution Volume (ml)
0
1
2
3
4
5
6
UVSignal(volts)
90° light scattering signal (scaled) UV (280 nm)
42. “Oligomer hunting”: display the absolute
molecular weight from LS in units of monomers
6.2 6.4 6.6 6.8 7.0 7.2 7.4 7.6 7.8 8.0 8.2 8.4 8.6 8.8 9.0 9.2 9.4 9.6 9.8 10.0
Elution Volume (ml)
0
1
2
3
4
5
6
7
8
9
10
relativeMWfromLS/UV
relative MW (LS/UV) UV (arb units)
43. 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)
44. 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)
45. 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
or
extended
dimer
dimer
monomer
SEC/LS shows an antibody sample contains only monomer
and dimer, but dimer elutes at more than one position
46. 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. strong theoretical background; “first
principles” method
5. high throughput, low cost (less than the
HPLC it is used with), fairly easy
47. 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 region near the exclusion limit
4. good signal/noise may require larger
injection amounts than normally used
49. Often when proteins go bad they
develop “snow” (a.k.a. “white amorphous
material” [WAM] or “floaters”)
• may only appear after many months
• often nucleation-controlled reaction
• often ≤ 0.01% of total protein
50. When this happens our valuable
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’
51. Dynamic light scattering: the basic idea
1. In dynamic scattering we measure the
fluctuations in scattering intensity
2. The time scale of those fluctuations depends
on the diffusion coefficient of the
macromolecule, which in turn depends on its
size
3. Like classical LS, the intensity of scattered
light is proportional to M, so the sensitivity
to very large aggregates is very high
52. Typically the data are transformed into a
distribution of hydrodynamic radius; this
distribution shows 2 well-resolved peaks
2.39 nm, 86.6% of intensity
92.3 nm, 13.4% of intensity
0.012% of mass
99.988% of mass
53. A different lot has 3 aggregate peaks, but
they still represent only ~0.3% by weight
<0.0001%
99.7%
of mass
0.3%
0.008%
54. DLS drawback 1: Poor quantitation of
mass fractions
Usually at best the reproducibility of
mass fractions is only within a factor of 2
There is no universally-accepted standard
algorithm; different methods can give
quite divergent results
55. DLS drawback 2: Low Resolution
Generally, to be resolved as a separate
species, a second component must be > 2-fold
different in Rh and thus > 8-fold in mass
Thus small oligomers are not resolved, and
this is generally a poor method for detecting
them
This limited resolution may simply not be
good enough to tell you what you need to
know
56. Although small aggregates are generally not resolved
as separate species, they do shift the distribution to
higher sizes
4 6 8 10 12
0
2
4
6
8
10
12
14
"good" sample
"bad" sample,
>10% aggregates
%Intensity
Hydrodynamic radius, nm
57. DLS drawback 3: “Blinded by the light”
It can be difficult to detect the main component
in the glare from large aggregates; if you lose
the main peak, you can’t quantitate fractions
8.5% of intensity,
96% of mass
as supplied after centrifugation
58. 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 size 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
59. Despite its limitations, DLS can be
quite useful for:
1. detecting large aggregates at levels below
0.01%
2. tracking down which process steps
generate large aggregates
3. relative ranking of different formulations
or processes (which is better)
4. accelerated stability analysis done in situ
in the DLS instrument
60. Recommended applications/approaches
1. Use all these orthogonal approaches to test
whether your SEC method is missing anything
that is significant
2. Use sedimentation velocity and on-line static
LS to help develop better SEC methods
3. Trace onset of damage during manufacturing
using DLS
4. For products formulated at high protein
concentrations, dilute into PBS and run
sedimentation velocity to detect long-lived
aggregates that may persist in vivo [“dilute
and shoot” protocol]
61. With thanks to the people who made
the proteins
1. Several clients who allow me to show
data for “protein X”
2. Kunio Misono at Cleveland Clinic
Foundation (sANPR)
3. former Amgen colleagues