Questions on fragment binding

Hi All,
I have a question about fragment-binding studies. I am working on a Biotin CAP chip, studying a 27 kDa biotinylated enzyme. My purpose was to confirm the fragment hits (150-300 Da) obtained by a primary screening, and if possible determine their KD and kinetics. I attached here two representative examples (KD and kinetic curves).

My questions are:
- Do the fittings of the two curves look acceptable to you? I asked my local Biacore representative, who said that as long as the SE is of 10% or less, result should be reliable. Is it really the case?
- I do not understand the Rmax proposed by the fitting. Do you have any idea of what is happening? When I try to calculate the expected Rmax, the result is always very far from the fitting.
- The software always mentions that the ka is outside the range of measurement of the machine (Biacore T200). Can I still consider the data, even as approximate values?

Thank you so much for your help, and for maintaining this forum alive! All these pages have been very helpful!

-Marie

Hi Marie,

My first concern is that you with a local Rmax. This will make your fitting much better but when fitting a concentration series on the same surface this is not a good practise.
Second the sensorgrams seems to suffer from high bulk effects which will hamper a clean flitting. A RI jump of 46 RU on a response of 55 RU should tell you to optimize first.

Comparing the KD between the two types of fitting shows that they do not match very well. Therefore I would believe the steady state fitting (be aware the RI values count here as analyte response) more over the kinetic fitting. That said, the red line in the steady state fitting indicates the KD. You can see that the highest analyte concentration is lower than the KD. The calculated Rmax is based on full saturation of the ligand.

Questions
- Always look at the fitting. SE values can be very low even as the fitting is not correct.
- How did you calculate the expected Rmax? Due to the high RI and analyte concentrations that are too low the fitting is less accurate as can be seen in the SE of the Rmax (26%).
- ka fitting is no good – don't believe it.

Kind regards
Arnoud

Hi Arnoud,

thank you for your advice!
I indeed read multiple times that for fragments, kinetic data are often poorly reliable, and people normally prefer to use the steady state KD. This is due to important mass transfer, and re-binding effects, from what I remember.
But do you think that curves can at least be compared (they were obtained with the same batch of immobilized enzyme), to qualitatively sort fragments by their koff?

I have tried to fit the curves with a global Rmax, but the result looked terrible... And the integrated software does not allow to fit kon and koff separately. But I'll try again, maybe using another software.

Thanks again!

Marie

Hi Marie,
I am not surprised about the bad fit with a global Rmax, since most of the signal originates from the bulk effect. In general the 1:1 model is fitted with the ka, kd and Rmax as global parameters ( www.sprpages.nl/data-fitting/models ). This will give the most robust kinetic values. If the fit is failing this is a hint to optimize the experiment.

Dissociation ranking (kd-ranking) can be done when you have enough response during dissociation. Bear in mind that de dissociation curve is also an exponential ( www.sprpages.nl/sensorgram-tutorial/exponential ) and therefore the dissociation should have some curvature to fit.

Kind regards
Arnoud