1) New software methods have improved the resolution and sensitivity of sedimentation velocity analysis, making it useful for comparability testing, formulation development, and quality control. 2) A method developed by Peter Schuck increases resolution of multiple components and detects very minor components below 1%. It provides higher resolution than SEC columns. 3) Another new method allows obtaining conformation and mass information at extremely low concentrations below 1 microgram total protein.

1. 1
NEW SOFTWARE METHODS ENHANCE SEDIMENTATION VELOCITY
ANALYSIS OF PROTEIN AGGREGATION AND CONFORMATION
John S. Philo, Alliance Protein Laboratories, Thousand Oaks, CA www.ap-lab.com
Some data analysis methods developed over the last year or two have significantly
improved the resolution and sensitivity of sedimentation velocity analysis, making it an
even more useful tool in comparability protocols, formulation, process development,
and QA/QC. A method developed by Peter Schuck at N.I.H. greatly enhances the
effective resolution of multiple components and also gives high signal/noise. With this
approach the ability to resolve small oligomers appears to exceed that of the highest-
resolution SEC columns, with the additional advantages of being able to analyze directly
in formulation buffers and without potential loss of aggregates to a column matrix. The
high signal/noise also enables detection of very minor components (well below 1%).
Several examples of applications to comparability testing and formulation will be
presented.
The principal drawback of Peter Schuck’s method is that it discards information about
the diffusion coefficient (and hence the mass) of each component. However, another
new method developed in our laboratory extends the sensitivity of the more standard
dc/dt analysis, allowing both conformation and mass information to be obtained at
extremely low concentrations or for minor components. An example of analysis of
antibody samples at concentrations of 1-2 microgram/ml (<1 microgram total protein)
will be shown.

2. A. The high-resolution c(s) method

3. 3
Peter Schuck at N.I.H. has developed a new software method1
that in effect removes the influence of diffusion and thus greatly
increases the resolution of different species. The resultant “c(s)”
distributions also have very high signal/noise so very minor
components can be detected.
1Schuck, P. (2000). Size-distribution analysis of macromolecules by
sedimentation velocity ultracentrifugation and Lamm equation modeling.
Biophys. J. 78, 1606-1619.
5.9 6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 7.0 7.1 7.2
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
Absorbanceat280nm
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
1.2
c(s),normalized
sedimentation coefficient (Svedbergs)
raw data sedimentation coefficient distribution

4. 4
Examples of the c(s) method
1. When applied to a monoclonal antibody sample, this
method resolves at least 4 minor components in addition
to the main peak at 6.5 S: a fully baseline-resolved dimer
peak at ~9.4 S, small peaks at ~13.7 S and 17.5 S
(probably trimer and tetramer), and a low mass
contaminant at 2 S (probably free light chain).
2. A heavily stressed sample of another antibody shows a
whole series of small oligomers, as well as free heavy and
light chains, for a total of 8 species.
3. Another useful feature is the ability to cover an enormous
range of sizes in a single analysis. For example, in
formulation studies of a ~20 kDa recombinant protein,
marked increases in aggregate formation were seen upon
addition of salt, particularly very large aggregates (16-40
S, ~500-2000 kDa) that were not resolved by SEC.

5. 5
0 2 4 6 8 10 12 14 16 18 20
0.0
0.4
0.8
1.2
1.6
2.0
c(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
Minor components in an antibody sample

6. 6
A highly stressed antibody sample shows a well-resolved series of
oligomers up to hexamer, as well as free chains. Clearly the
resolution and range are significantly better than any SEC method.
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
hexamer,0.4%
pentamer1.4%
tetramer5.3%
trimer14.6%
dimer30.6%
main peak (monomer), 45.5%
freeheavychain,0.8%
freelightchain,1.4%
c(s),normalized(totalarea=1)
sedimentation coefficient (Svedbergs)

7. 7
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
Formulation studies: effects of NaCl on aggregate content

8. B. Extending the sensitivity, range
and accuracy of the dc/dt method

9. 9
Can you see a boundary moving in these velocity data for a
sample at 1.2 µµµµg/ml (0.5 µµµµg total protein)? A new algorithm we
have developed can recover both the sedimentation coefficient
and the solution mass from such data!
Monoclonal antibody sample at 1.2 micrograms/ml; every 4th scan shown
5.9 6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 7.0 7.1
Radius (cm)
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
Absorbance(230nm,1.2cmpath)
00002.RA3 00006.RA3 00010.RA3 00014.RA3
00018.RA3 00022.RA3 00026.RA3 00030.RA3

10. 10
What are we trying to do?
• The dc/dt method developed by Walter Stafford has the
useful property that the width of each peak in the
distribution can be related to the mass of that species.
– the peak width measures the diffusion coefficient, and the ratio of
sedimentation coefficient to diffusion coefficient gives the mass
– this mass/diffusion information is sacrificed in Peter Shuck’s c(s)
method in order to get higher resolution of different species
• Unfortunately with the dc/dt method one can use data from
only a small portion of the entire run. This both limits the
signal/noise and makes it difficult to measure samples
containing a large range of sizes.
• The standard method can also give systematic errors of
~10% for masses.
• We have developed some new mathematical approaches
that remove these limitations, as demonstrated below.

11. 11
Analysis of the antibody data at 1.2 µµµµg/ml (0.5 µµµµg total). With the standard
approach the peak at ~6 S is barely seen. With the new algorithm we get the
sedimentation coefficient with a precision of better than 1% and a mass of 175
± 35 kDa (i.e. it is a monomer).
0 2 4 6 8 10 12
0.000
0.002
0.004
0.006
standard algorithm
new algorithm
g(s*),AU/Svedberg
sedimentation coefficient (Svedbergs)
-0.01
0.00
0.01
0.02

12. 12
In these data for a degraded antibody sample at 1.2 µµµµg/ml, we can now see a shift
to lower sedimentation coefficient, and we can be >99% confident that the width
of the peak is too broad to be consistent with a single monomeric species. (Data at
higher concentrations reveal additional species at ~4, 5, and 9 S.)
0 2 4 6 8 10 12
-0.002
0.000
0.002
0.004
0.006
g(s*)
s* (Svedbergs)

13. 13
Significance of the improved algorithms
1. The improved sensitivity permits examining the
conformation and solution mass of <1 µg of protein,
which may be useful for examining side fractions
collected from analytical columns or other materials
available in limited amounts.
2. The higher accuracy for mass values is very helpful in
assessing heterogeneity. If the apparent mass (derived
from the peak widths) is too low, this implies unresolved
heterogeneity in the peak.
3. With the new methods it is possible to analyze a broader
range of sedimentation coefficients in one experiment.

14. 14
Conclusions
• The resolution and size range of sedimentation velocity
analysis is now significantly better than size-exclusion
chromatography, adding to its other advantages:
– wide range of solvent conditions, usually compatible with
formulation buffers
– no loss of components due to column interactions
– absolute method (does not require standards)
• New methods also extend the sensitivity down below 1
microgram, and allow for more accurate determination of
solution mass
• Low throughput is still an issue, but sedimentation velocity
is nonetheless a powerful tool for comparability studies,
validating SEC or other sizing methods, and/or providing
supplementary data from an orthogonal approach