[RASMB] SE things

Holger Strauss Holger.Strauss at nanolytics.de
Mon Jun 2 00:31:17 PDT 2008


Hi Eric, hi all,
just a few additional comments to the nice discussion going on while others 
were asleep.
- what are the Kds for both your SE-models? Is there enough (i.e. > 5%) of 
each oligomer in solution? How do the partial concentrations of each 
oligomer compare over your examined concentration range? Ultrascan has a 
nice feature to show this. This can help to visualize what the statistics 
mean in physical terms and if you stand a realistic chance of describing an 
oligomer at all. Also, how certain are you of your sigma-value (i.e. vbar 
and density)? If you fix the molar mass to the expected value, it might be 
off by a couple of percent and this discrepancy will go into the other 
exponential terms (Kd and stoichiometry). Can you fit it along with your 
Kds for the different models? Or vary it over a +/- 5% range and see what 
happens?
- c(s) at "high" concentrations (2-5 mg/mL, depending on the protein) gives 
artefactual peaks at higher and lower s-values. They have no physical 
meaning but the boundary becomes distorted in ways a single f/f0 can't 
describe; it will tend to increase, though, which is a good indication that 
something "more" is happening. Likewise, f/f0 < 1.2 (even <1) is an 
indication of self-association. (Before everyone rushes to makes statements 
about how difficult it is to interpret f/f0 and solvation and hydration and 
shape and reaction boundaries and what not: There's information in this 
value and it can be used to build hypotheses - with care and not as a 
single parameter).
- If you need to show that your crystal-clear trimer is also present in 
solution, you might want to consider plotting the local molar masses (from 
lnc/r2 of the SE-gradients) vs. concentration and see if the values at the 
highest concentrations increase above the expected value of the dimer. If 
they do, you can at least rule out the 1-2-only model; could still be a 
1-2-4/1-2-3/... Overlaying those plots from different experiments will also 
tell you wether your system is fully reversible or not and hence test the 
applicability of the Law of Mass Action.
- It could also help to put more information into single SE-gradients, i.e. 
increasing the column lenght and scanning with interference optics (still 
using multiple speeds & concentrations). You can cover 3 orders of 
magnitude on the concentration scale in a single gradient this way and the 
fitter will like this. It's also good for the lnc/r2-plots. And it's very 
good to keep the centrifuge spinning for days and days.
I have found with other samples that if I increase my protein concentration 
much more than 1 mM, my molecular mass will continue shifting towards the 
aggregate. With the 1 mM data I have now, the trimer is the main species, 
but this shifting trend appears to be starting in this sample, and so I 
fear that increasing my concentration may not help clear anything up.
You can use dc/dt- or ls-g*(s) and plot the average s-values. They are more 
robust against artifacts than the c(s)-approach and compare very well with 
those from c(s).
Using my crystal structure, how can I calculate a theoretical s value for 
my trimer? If I use the calculator in sedfit, it generally overestimates 
the s value of the monomeric protein which I have observed experimentally. 
This isn't surprising since the protein is rod-shaped and not spherical, 
but it makes interpreting the predicted s values from sedfit rather difficult.
Hydropro by Garcia de la Torre is the program everybody is using to get 
hydrodynamic values from the crystal structure, but there others around 
(Solpro, Beads).
Cheers, Holger

=================================
Nanolytics
Gesellschaft fuer Kolloidanalytik mbH
Dr.  Holger Strauss

Am Muehlenberg 11
D-14476 Potsdam

Tel.:		+49 331 581 8360
Fax:		+49 331 581 8361
e-mail:	Holger.Strauss at nanolytics.de
Internet:	www.nanolytics.de
  




More information about the RASMB mailing list