getting fine-grain data

It would seem there is a continuum from single binding site binding, to a small number of binding sites, to a large number, to fully non-specific. Is it the case that for a given fragment, you could see all of these behaviors at different concentrations? I'm wondering if it is important to get a lot more data points at more fine-grain steps in concentration. In general, what is it about the curves that distinguishes super-stoichiometic binding (multi-binding sites, ok sometimes) from non-specific binding (bad).

I think you are missing the point of binding interaction. Most interactions are real specific. The interaction sites are well defined and there is no continuum of binding sites on larger molecules.
When you use higher concentration of one of the interactants you push the interaction to saturation, but when you go higher there is no more interaction because one of the components is limiting.
When you even higher you get problems: precipitation by high local concentrations, non-specific binding, etcetera. So stay on the physiological side. If our compound is working in de nM range, don't inject milli molars.

I recommend reading the surveys of Rebecca Rich and David Myszka ( www.sprpages.nl/reviews/surveys.html ) to have a lot of illustration about what i mean.

Kind regards

Arnoud

Hello Arnoud,
Thank you for the reply, and I completely agree with your statements below, yet how do they apply to fragment-based discover? Fragments are often run at high concentrations where the interactions are pushed to saturation. Even at lower concentrations, multiple binding sites can exist for a fragment. A concern I have is that these are viewed as promiscuous binders, and discarded, yet the fragment could be binding in the site of interest as well as other sites. The paper by Giannetti in 2008 described promiscuous binders (J. Med. Chem. 2008, 51, 574–580) and there are examples of progressing promiscuous binders (J. Med. Chem. 2009, 52, 4794–4809 and Bioorganic & Medicinal Chemistry Letters 20 (2010) 586–590). Two other papers by Peter Schuck (Biophysical Journal 84(6) 4062–4077 and Biophysical Journal 92(5) 1742–1758) touch on the topic of understanding the superposition of signals from binding to sites spanning a range of rate and equilibrium constants. I just do not want to be throwing away useful data. Thanks.

In general, promiscuous binders behave normally at low concentrations but exhibit nonstochiometric behaviour at higher concentrations. At high concentrations, the ligand seems not to saturate and binding responses exceed the maximal (Rmax) expected values. Adding detergent or salt can diminish this nonstochiometric behaviour by shielding the charges, which are responsible for the aggregation.

I think the best way is to establish what the Rmax of your system is by injection of a known well-behaved compound. Or to calculate the theoretical Rmax, although you don't know the real active ligand concentration.
Then set some boundaries on what you discard.
Giannetti (1) distinguishes several types of promiscuous binders:
• SS: superstoichiometric binders. When the analyte is binding to a ligand with a stoichiometry of greater than 5:1. These compounds are eliminated from the hit list.
• NS: nonstoichiometric binders. When binding to a ligand stoichiometry greater than 1:1 but less than 5:1. These compounds get a low priority. Although they can be followed up in other assays if there is sufficient interest in other properties of the compound.
• CD: concentration-dependent aggregators. When binding to a ligand is not saturating at higher concentrations. These compounds are eliminated from further consideration and are not good candidates for structure-based design because they will likely aggregate and nonspecifically associate with protein at the high compound concentrations required for X-ray crystallography.
• NB: nonbinders. When there is no interaction with the ligand of interest. The compounds are eliminated from the hit list.

Depending on what you looking for with your compounds, I would like one with properties I can predict and understand. Like 1:1 stoichiometry, ligand saturation and proper kinetics.

Arnoud

Refs:
1. Giannetti, A. M., Koch, B. D. and Browner, M. F.; Surface Plasmon Resonance Based Assay for the Detection and Characterization of Promiscuous Inhibitors. Journal of Medicinal Chemistry (51): 574-580; 2008.