FW: [RASMB] question about WinNONLIN

[James Cole]

Dear Satinder:
I agree with Peter Schuck that you should do a velocity run to see if the
sample contains a low molecular weight contaminant. The presence of a
contaminant would give rise to some systematic deviations in your fit of the
equilibrium data to a monomer-dimer model. There are some other
possibilities you may also want to check out:

1) You need to look at the confidence intervals in order to tell if the
change in the dissociation constant is significant when you allow sigma to
float. The reduced molecular weight (sigma) and the dissociation constant
are correlated parameters and it may be that the confidence intervals are
very broad when you allow them to float such that the change in the
dissociation constant is not statistically significant.

2) The change in sigma is about 15% when you let it float. Assuming that
this change is statistically significant, it is a bit outside the usual
error range. However, if there's a slight mistake your calculations you
could end with a 15% error. I assume that you are determining the predicted
sigma of 0.95 using a calculated vbar and solvent density. Typically you can
calculate v-bar to about 1-3% accuracy from the amino acid composition and
the solvent density calculations are certainly better than 1%.  Given that a
1% error in v-bar or solvent density translates to a 3% error in sigma, you
would end up with 15% error in sigma if your  v-bar or density calculation
is off a bit more than normal.

3) I noticed that the loading concentrations run up to 1.2 mg/ml. Assuming
that the protein has a typical specific absorbance, you would end up with a
very high absorbance at the cell bottom, particularly at the higher
concentrations. Be sure to truncate the data at high OD to avoid problems
from nonlinearity in the optical system. I don't trust the linearity of the
absorbance optics at 280 nm above 1.8-2 OD, and if the lamp is dirty you
would do worse.

Good luck,
Jim Cole



-----Original Message-----
From: rasmb-admin at server1.bbri.org [mailto:rasmb-admin at server1.bbri.org] On
Behalf Of Satinder Singh
Sent: Thursday, November 04, 2004 12:07 PM
To: rasmb at server1.bbri.org
Subject: [RASMB] question about WinNONLIN


----------------------------------------------------------------------------
------
The older archived RASMB emails can be found at:
http://rasmb-email.bbri.org/rasmb_archives
and current archives at http://rasmb-email.bbri.org/pipermail/rasmb/
Search All the Archives at: http://rasmb-email.bbri.org/rasmb_search.html
----------------------------------------------------------------------------
------

Hello.

I am trying to determine the dissociation constant of a homodimer. The 
monomer molecular weight (by both sequence & molecular weight) is 33,400 
Da. At 15,000 rpm, the sigma is 0.95. I ran the experiment at 15,000, 
20,000, 27,000, and 35,000 rpm with 4 different protein concentrations 
(0.2, 0.4, 0.8, & 1.2 mg/ml) using A280 detection.

Before concluding that I indeed had a monomer-dimer equilibrium, I used 
the "sigma test" to see if there was evidence of association. Indeed, the 
sigma increased as a function of protein concentration (for a single rotor
speed). It also increased as a function of rpm (for a single protein
concentration).

The data are best fit to a monomer-dimer equilibrium, although the 
residuals with this model (sigma fixed) still look a bit skewed. The SRV, 
however, is pretty good at 5.6384x10-3. The question I have is 
this: When I FIX sigma at the monomer molecular weight, I get a 
dissociation constant of 12 uM. However, if I let sigma float, sigma falls 
to 0.8276, which corresponds to a molecular weight of 28,000 Da,  and I 
get a dissociation constant of 2.3 uM.

If the data are fitted correctly and nothing has happened to the protein
sample during the experiment, i.e., aggregation, precipitation, shouldn't
the 2 dissociation constants be relatively close? Could anyone tell me 
what I may be doing wrong?

Thanks in advance for your help.

Satinder




_______________________________________________
RASMB mailing list
RASMB at rasmb-email.bbri.org
http://rasmb-email.bbri.org/mailman/listinfo/rasmb