From: Borries Demeler/Biophysics <demeler@selway.umt.edu>
  To  : rasmb@bbri.eri.harvard.edu
  Date: Mon, 16 May 1994 12:41:11 -0600 (MDT)

Re: van Holde - Weischet method

Geoff Howlett wrote:

>
>Ok, so I don't mind being the fall guy.  I have a few questions concerning 
>recent messages:

I think we are all the same boat, we just "fall" in different places.

>1.  What is the Van Holde-Weischet method and what advantages does it offer?

Allow me to shed some light on this question as a person that has been using
and programming this method since being a graduate student.

The van Holde - Weischet method is a very versatile method for the analysis
of velocity data of multicomponent systems. The method is accurate for a wide 
range of S values (in our laboratory we successfully measured samples between
~1 S to 3000 S) and can be used to reliably resolve components differing as
little as 2%-5% in S, depending on the quality of the data. Diffusion coeffi-
cients provided by this method are unrealiable and the method is generally not
used for this purpose, although a qualitative comparison between diffusion
coefficients of multiple components in the same sample can be made. 

The method attempts a global fit on a set of scans taken over a period of
time during sedimentation. The value for dC/dr for each scan close to the
meniscus and close to the bottom of the cell should be zero to be included
in the analysis. This means that a baseline and a plateau for each scan
should exist, and that the scan should have cleared the meniscus. Samples
with small molecular weights and high diffusion coefficients can be
analyzed if the data was taken at high speed. 

Ideally, all scans should be of the same material, i.e., no processes that
alter the consistency of the sample (degradation, pelleting, active enzyme
reactions) should take place during the sedimentation in order to obtain
a reliable global fit. That means that the samples should be at chemical
equilibrium. If this is not the case, rather than failing, the method
provides a diagnostic for the occurrence and lets the user distinguish
between pelleting, degradation or aggregation, but it will not provide
accurate S-values. 

If samples are already aggregated, accurate values for the various forms
of aggregation can be obtained from this method. Further, the method
provides a qualitative answer about the level of interaction between
individual components of the sample. It is very easy to distinguish
non-interacting samples (for example: DNA restriction fragments) from
interacting samples (for example: monomer-dimer-tetramer interactions) with
this method. 

The van Holde - Weischet plot is easily transformed into a G(S) plot (also
known as a integral distribution plot) that provides a graphical
representation of non-ideality (concentration dependence of S, if it
exists) allowing extrapolations to S values of zero concentration. I know
of no better way to test for non-ideality than to use the van Holde -
Weischet method. 

The algorithm for the method bases on the fact that sedimentation is a
transport process proportional to the first power of time, while diffusion
is a transport process proportional only to the square root of time. By
extrapolating to infinite time, diffusion effects on the shape of the
boundary can thus be eliminated in a graph of S at various points in each
scan vs. the inverse square root of time of each scan.  

This method has been used in our dept. to analyze well over 1500 samples
and has been found it to give useful answers in almost every case. It has
proven itself particularly well for the analysis of the structure and
stability of large, complicated protein/protein, protein/DNA and
protein/lipid assemblies. In fact, in my opinion it is just about the best
method to analyze a host of different situations, to test for homogeneity
or to analyze paucidisperse systems.

The absolute S values obtained with this method agree very well with other
methods. If you are analyzing heterogeneous mixtures, this method will
provide you with the most reliable answers. There is no limit to the number
of components the method can resolve other than the limitation imposed by
the quality of the data. If you test for homogeneity, the method can
provide you with very convincing answers (important for those of you
analyzing drug purity). 

With all the good things that can be said for the van Holde-Weischet
method, there are some limitations, and some situations when the method
should be used with caution: 1. Samples that diffuse too rapidly to be
measured at the speed the XL-A is designed for, i.e., samples that do not
clear the meniscus. 2. Samples that undergo chemical reactions during the
time of centrifugation. 3. Samples that precipitate (pellet) during the
experiment. In those cases, a reduction in speed often solves the problem.
4. Samples that differ too widely in S to be captured in the same scan over
a period of time. In those cases, the individual components can be
accurately analyzed separately, with a dataset of a low speed run and
another dataset of a high speed run. 

I have programmed an Origin/Windows interfaced van Holde - Weischet
analysis program (commercial - one of the things I do for a living is
programming). If you are interested I can send you a limited functionality
DOS version to try it out. 

Regards, 

-Borries Demeler