I am new to ITC. I am currently using the technique to assess binding between a protein and a peptide to which it binds. The protein has two binding sites for the peptide, however site two is inhibited by by the c-terminus of the protein itself until activated. From ITC I am getting a good binding curve which can be fitted to a one site, or sequential binding site model. I have read papers with similar situations (one site is autoinhibited) analysed using SPR where they have fitted to a two site model, despite showing that the stoichiometry of the interaction is 1:1 in the autoinhibited state. I guess perhaps because it cannot be guaranteed that site two is 100 % inhibited? I therefore have three questions I am seeking advice on.
1) Should I use the sequential binding site model in this situation?
2) What is the difference between the two site and the sequential binding site model? My data cannot be fitted to the two site model.
3) Unrelated but I am also uncertain - how should I convert the error once I have calculated the Kd from K?
Any advice very much appreciated!
The "two binding sites" model built in the commercial software corresponds to two non-identical and independent (no cooperative) ligand binding sites. The "sequential binding sites" model corresponds to more than two ligand binding sites that might be identical or non-identical, independent or cooperative. Thus, the "two bidning sites" model is more restrictive and the "sequential binding sites" is a general model for any possible scenario with n ligand bidning sites. Thus, a given titration might be fitted with the latter, but not with the former, while the opposite cannot be true.
Two important points. First, extreme care should be taken when interpreting the association constants obtained by using the "sequential binding sites", because they do not correspond to intrinsic site-specific constants; they are the step-wise association constants, which are functions of the intrinsic site-specific constants. The same applies to the binding enthalpies. Second, related to the previous point, the name of the model may be misleading, because ligand binding is not sequential in general, but all posible ligation states coexists in equilibrium; that is, there is no sequential/ordered filling of the binding sites (unless the binding sites show considerably different affinities or large negative cooperativity effects). The name "sequential" derives from the way the partition function or binding polynomial is constructed by enumeration of the different ligation states, but it does not imply any binding mechanism or odering, since it is a chemical equilibrium model.
Kd is the inverse of an association constant K. Therefore: Kd = 1/K, and both constants have the same relative error: err(Kd)/Kd = err(K)/K
Just one quick comment. I said that the "two binding sites" model corresponds to two binding sites. But, in fact, it corresponds to two classes or groups of non-identical and independent binding sites.
Hi there, thank you for your responses. Sorry this is all a little new to me..what is considered to be an 'acceptable' degree of error? I have attached a couple examples of my data
- Badges
- Science topic