[RASMB] Variability in association constants from sedimentation equilibrium data

[David Hayes]

Hi Ronald,

I did very few sedimentation equilibrium experiments, but I would say that I can relate to your current experience.
Every time I fit sed equilibrium data from the first speed (because the collaborator just could not wait a whole week for a peek at the answer) I was always completely wrong compared to the global fit to all three speeds.  If you are still fitting just one speed and getting strange answers, that would be what I would expect.

The limited concentration range of absorbance optics can also be a killer to getting a good data fit.  Do you know the expected kD?  Are you anywhere near the kD concentration?  Can you use interference optics?  You can also try 3 mm cells for higher concentrations and using A230 (depending on buffers) for low concentration.
I found the 6 sector cells too short in column height for really good sedimentation equilibrium data:  I just used an 8 hole rotor and lots of double sector cells filled with 150 - 200 uL.  And yes, it took longer to come to equilibrium, but the extra data was worth it.  (Also, did you use match or SEDFIT to test for equilibrium?)

Also, at most concentrations, the range of A280 linearity, which is about 1 OD really limits what you are actually fitting.  It could be that most of the dimer signal in your system is above 1 OD and you are not even really fitting it at all.  With a bit of work you can get SEDANAL to simulate any interacting system and also get output of the distribution of species.  With this information you can see how many data points that you are actually including in the fit are significantly incorporating signal from dimer (this calculation depends on assuming a kD, so actually you will get a set of answers from a set of guessed kD values).

Sed equilibrium with absorbance optics works better than that with interference optics because you don't have to float the baseline.  But you say that some sectors are mismatched, so maybe you are floating the baseline.  Bad baselines give you bad fits.

I also found it useful to "screen" each channel by doing a fit to a single species without floating the baseline at all as an estimate of that sector's average sigma.  You should see a trend in the sigma versus concentration for a self associating system.  Mis-filled or noisy channels then stick out without making assumptions.

So in conclusion, my personal rule is 3 speeds, 3 concentrations, global fit, or don't tell anyone anything.  I started using 4-5 speeds and 4-5 concentrations after getting burned by having less data.

Also, have you done sed velocity?  How pure is this stuff?  How much irreversible dimer and larger species is in it?  If the kD is near 0.5 mg/ml, then you could get the kD from a concentration series sed velocity data set.  If the kD is far from 0.5 mg/ml, you won't get any good sed equilibrium or sed velocity data at 280 nm.

David Hayes


________________________________
 From: "Toth IV, Ronald T" <Toth.Ronald at ku.edu>
To: "rasmb at rasmb.org" <rasmb at rasmb.org> 
Sent: Friday, October 4, 2013 12:23 PM
Subject: [RASMB] Variability in association constants from sedimentation equilibrium data
 


 
Hello members!

We have a project where we are trying to extract association constants from sedimentation equilibrium data of a self-associating mAb. We are noticing a lot of variability in the association constants fit to between concentrations and even between replicates of the same concentration. My question is, is this expected? If not, what are some possible reasons it could be happening and what can I do to resolve it?

I’ll give more information in case it is relevant. We are using Beckman 6-sector cells and absorbance optics at 280 nm. Our initial guess as to rotor speed was a little high and gave sigma’s between 4 and 5. In some sectors there was a mismatch between sample and reference, but nothing in the buffer absorbs at 280 nm. In the fit we are using Heteroanalysis with a monomer-nmer model and constraining the molecular weight and stoichiometry.

We had a relatively new user load the cells, could cross contamination between reference and sample combined with a mismatched meniscus cause this?
 
Thanks in advance for your feedback!


- Ronald Toth
 
Post Doctoral Researcher
Macromolecule and Vaccine Stabilization Center
University of Kansas
_______________________________________________
RASMB mailing list
RASMB at list.rasmb.org
http://list.rasmb.org/listinfo.cgi/rasmb-rasmb.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.rasmb.org/pipermail/rasmb-rasmb.org/attachments/20131004/726a11f7/attachment.htm>