From: djs17 <djs17@york.ac.uk>
  To  : Marina Mapelli <Marina.Mapelli@EMBL-Heidelberg.de>
  Date: Tue, 13 Apr 1999 12:45:18 +0100

Re: ultracentrifuge data analysis

Hia Marina,

I know that most people have replied with quite technical answers to
your question, but in my brief experience with Data fitting the greatest
source of error in calculating a molecular weight is a good estimate of
the optical baseline. If you overspeed the rotor to clear the meniscus
you'll be able to determine this directly. Also, if when fitting your
data, constrain the molecular weight the molecular to a monomer, and
float the baseline. If the baseline for all of your data sets comes out
the same (but non-zero), then it is most probably the source of error in
your estimate.
Also, don't assume that because you have the same buffer in the
reference channel that the baseline is zero. Often there is a small but
finite baseline for most biological buffers (especially if DTT is
present).
A baseline error of 0.1 OD's (not as uncommon as you'd think) will give
you an underestimate of Mwt of 24% in something of Mwt 120KDa at 10KRPM
(this is data generated with the data stimulater in the Beckman software
version 4.0, and fitted using WINNONLIN). Your residuals of the fit will
also be skewed.


Secondly, another trivial reason is that the sample is not at
equilibrium, however I guess you probably checked this!

If any, or all of the above applies then you'll get enough error in your
data fitting procedure to give quite an erroneous answer. I know this
from bitter experience.  

Hope this helps and good luck!

Dave Scott.




Marina Mapelli wrote:
> 
> Dear collegues,
> I'm currently working on the characterization of a single-stranded dna
> binding protein. I'm trying to understand the
> behaviour in solution of  different mutats using analytical
> ultracentrifugation. My equilibrium experiments result in a monodisperse
> preparations but the mono fitting leads to  molecular
> weights too small comapared with the ones predicted by the sequence and
> observed in denaturing gel runs (85-90 KDa instead of 120 KDa).
> My experiments take place in a buffer containing 20% glycerol and 300mM
> NaBr. I measured the density of the medium, but I'm wondering whether and
> how the partial volumes of the protein can be affected by the buffer. Or
> whether there's another explanations for this behaviour.
> 
> I also noticed that sometime I got a good fit with a biexponential fitting
> but with Mw that are one too small and one too big. Does this means that
>  protein aggregation is taking place so that the bimodal fit
> is not valid (I know that protein-protein intercations exist)?
> 
> many thanks in advance
> best regards
> marina
> 
> *******************************************
>  Marina Mapelli
>  EMBL Heidelberg
>  Meyerhofstrasse n.1
>  D69012 Heidelberg
>  Germany
> 
>  e-mail:        mapelli@EMBL-Heidelberg.de
>  tel:           +49 6221 387567
>  fax:           +49 6221 387306
> 
> *******************************************

-- 

*********************************
* Dr. David J. Scott		*
* Dept Biology			*
* University of York		*
* YORK				*
* YO10 5DD			*
*				*
* United Kingdom		*
*				*
* phone:  +44 1904 432868	*
* fax:    +44 1904 432860	*
* Email:  djs17@york.ac.uk	*
*********************************