[RASMB] analysis of heterodimerization

[Peter Schuck]

Satinder,

For the analysis of the heterogeneous interaction, I think you should 
switch to an analysis method that can keep track of two different 
components.  As far as I know, WinNONLIN currently is only applicable for a 
single solute component.   It is perhaps tempting to consider in your case 
B as practically the same as A regarding the sedimentation properties 
(because of the similar molar mass, and equimolar mixtures), but I think 
this doesn't really work out in this case with regard to mass conservation 
and mass action law in sedimentation equilibrium.  Also, there are more 
complicated issues, like 'what is the binding site of B on AA',  'are there 
AB complexes, or just AAB complexes, or both, or even AABB complexes' and 
if the latter is true, 'is there any cooperativity between the sites'?

We've developed sedimentation equilibrium models in the software SEDPHAT 
for this purpose, it does have models for mixed interactions of the kind 
you describe.  You can plug in your binding constant for self-association 
of A and determine the binding constants of AB, AAB and/or AABB, and 
estimate cooperativity (if two sites are present).  You can find it at
http://www.analyticalultracentrifugation.com/sedphat/sedphat.htm

In order to facilitate the determination of binding constants, it is 
frequently possible (and highly useful) to apply mass conservation 
constraints.  John Philo has shown how powerful that can be even if these 
constraints are relaxed in the final fit.  In particular, it can also be 
very useful to constrain the relative molar loading concentrations in 
between the different cells if you loaded your sample in dilution series, 
and constrain this value to 1 (if you're sure that it was precisely 
equimolar).  We have developed and implemented this approach of 
cross-correlating calculated loading concentrations of different cells in 
SEDPHAT, you can find a description in Analytical Biochemistry 326 (2004) 
234-256.  Another very useful technique in this context (introduced by 
Roark) is to collect data at different rotor speeds, since that can help to 
improve the mass conservation estimation without a need to predefine the 
bottom of the cell (discussed in the same reference).

Best,
Peter


At 11:18 PM 3/28/2005, you wrote:
>----------------------------------------------------------------------------------
>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 have a question about analyzing a heterodimeric interaction. We
>have collected data using 2 techniques, SEC/LS (size exclusion
>chromatography coupled to light scattering) and sedimentation
>equilibrium.
>
>We have 2 proteins, A & B.
>The molecular weight of A is 33.3 kDa, and the molecular weight of B
>is 31.8 kDa. Another important property is that they both have
>approximately the same extinction coefficient at 280 nm, i.e., they
>have about the same numebr of tryptophans & tyrosines.
>
>When A is analyzed by SEC/LS, the peak (dimer) that comes off the
>column and is analyzed by light scattering is a mixture of monomer
>and dimer. WHen B is subjected to the same analysis, it is only
>monomer. However, when A and B are mixed in an equimolar ratio, the
>"B" peak disappears and shifts into the dimer peak of A.
>Our interpretation was that A and B are heterodimerizing.
>
>
>To acquire some quantitative data on the interaction, we went to
>sedimentation equilibrium and did the following analysis using
>6-sector cells at 280 nm at 4C:
>
>1. Protein A alone at 3 different concentrations (0.2, 0.4, 0.8
>mg/ml) and 3 speeds (13k, 18k, 25k rpm). The data were analyzed
>using WinNONLIN, several models were tried, and the best model was
>a monomer-dimer equilibrium with an SRV of 4.9x10-3 and very nice,
>i.e, random, residual plots. The Kd was calculated to be 22.4 uM.
>
>2. Protein B alone at the same concentrations and speeds. THe data
>were analyzed again with WinNONLIN and could best be fit to a
>single species monomer. I tried other models, like single species
>monomer, mon-dim, etc., but the residuals were all VERY skewed. The
>SRV with the single species monomer was 4.2x10-3 and the residual
>plots looked very nice. So, I am pretty sure this is the best
>model.
>
>3. Proteins A & B in an equimolar ratio using a TOTAL protein
>concentration of 0.2, 0.4, 0.8 mg/ml and the same 3 speeds. We
>analyzed the data using WinNONLIN using the same type of analysis
>described above and came up with a Kd of 30.3 uM.
>
>Now here are the questions:
>1.In the mixture, both homoassociation (A+A --> AA) and
>heteroassociation (A+B --> AB) are taking place, so what does the
>calculated Kd of 30.3 uM represent? I assume no BB homoassociation
>is taking place because it doesn't occur alone (see point 2 above).
>We know what the Kd of AA homoassociation is. Can we calculate a
>heterdimer Kd from these data??
>
>2. Does the Kd of 30.3 uM represent anything at all or is it just
>garbage because I don't think WinNONLIN in equipped to handle
>heteroassociations.
>
>3. What would be the best way of calculating the Kd of the
>heterodimer in light of the data we already have, i.e., can we use
>the data we already have (SEC/LS & AUC) or do we need to do more
>experiments? If we can use the data we already have, how would you
>calculate the heterodimer Kd? If we need to do more experiments,
>should they be equilibrium or velocity? If we do SE, what program
>should I use? If we use SV, what program should I use and how do
>you calculate a Kd from velocity data?
>
>Thank you all in advance for your help.
>
>Sincerely yours,
>Satinder K. Singh
>_______________________________________________
>RASMB mailing list
>RASMB at rasmb-email.bbri.org
>http://rasmb-email.bbri.org/mailman/listinfo/rasmb