It is unfortunate to see the continued publication of likely inappropriate fluorescence quenching methodology in PLoS One. As reported by Grossweiner in 2000 (http://dx.doi.org/10.1016...(00)00113-5), the double log equation used to determine the "number of binding sites" (equation 10) essentially always returns a value of 1, even if it is known there are 2 binding sites.

As discussed later by myself and Lorenzo Stella (http://dx.doi.org/10.1016...) equation 10 is in essence a linearized version of the Hill equation, with "n" the Hill coefficient. The latter is a measure for the cooperativity of binding, not of the number of binding sites.

In my paper with Lorenzo Stella we also discussed various other potential pitfalls in the methodology. For example, equation 10 also assumes a non-fluorescent complex, and the concentration of free ligand should be used in the equation. Based on the authors' contention that their ligand binds in domain IIIA, one should expect the complex to be fluorescent, since this is far away from the two Trp residues. Another issue with the Trp residues in BSA is that preferential quenching of one of the residues (Trp213) may cause the other (Trp134) to be more dominant. Since the latter is more exposed to the solvent, the observed red shift may not be indicative of a structural change, but simply the preferential quenching of the more buried Trp residue.

I also wonder to what extent the inner-filter effect may have contributed to the findings. The authors do not report they corrected for this, but since no UV spectra are shown, I cannot determine whether this may have played a role.

While none of this proves the data and their interpretation are wrong, the authors should clarify how these issues affect their data interpretation. I strongly recommend to repeat the analysis with the appropriate corrections and preferably also another method (e.g. ITC, CE, equilibrium dialysis, etc).