Trouble fitting curves from fragment screening

Hi everyone,

Before I post my question, I'd like to thank the author of these pages. They have been extremely helpful for my SPR experiments and have greatly contributed to my understanding of this method. Please keep up the great work! 

I'm currently having trouble fitting a 1:1 binding model to my data. Basically, I'm doing a fragment screen of my target protein and any hits I find I do follow-up dose response assays with 8 different concentrations in duplicate to get a value for the binding (300 uM down to 4.50 uM + 0 uM). I then plan to go ahead and do some co-crystallisation with my target protein if the binding is in the low uM range (1-10 uM). I get a good response for many fragments but when it comes to fitting the data with the SPR software, it looks awful and gives me nonsensical results for KD and other parameters. I have attached some photos of two fragments following evaluation. One that has a strong binding response and another that has a slightly weaker one. Both give values that I don't trust. I can post more photos but there appears to be a limit of three per post.

I have also tried to see if I can get a better fit by tweaking the parameters, but the software doesn't appear to have many options and the end result is basically the same.

My current setup is as follows:

Immobilisation method: target level set to 10,000. Protein typically immobilised at 10,000 to 11,000 RU. Works consistently each time with no issues.
Dose-response screen method:   type = high performance
                                                        Contact time = 100 s
                                                        Dissociation time = 300 s
                                                        Flow-rate = 30 ul/min
                                                        Extra wash after injection = 50% DMSO
                                                        Solvent corrections are carried out every 16 cycles and I have 8 different concentrations.

Data processing: I use Biacore S200 Evaluation software. Biacore S200 Control software is used to make the methods.

I appreciate that fragments don't tend to bind well, but I think I should be getting more reasonable values than what I'm getting. I was thinking that perhaps I should try and reduce the flow rate or the amount of immobilised ligand? 

Any suggestions would be much appreciated. If more information is needed, I'm happy to provide it.

Many thanks,

Koko

Hi,
and thanks for your kind words.

It looks like your response is high enough to lower the ligand immobilisation.
As you studied the SPRpages it is recommended to use low immobilisation level to do kinetics. I suggest that now you know which fragments do bind you optimize the system to do kinetics.
The fitting is indication that the interaction is not 1:1 and you can't resolve that by changing the flow rate ( www.sprpages.nl/data-fitting/troubleshooting/high-buffer-jump ). You have to lower the ligand density and make sure that the ligand is only one form (and the fragment too).

Kind regards
Arnoud

Hi Arnoud,

No worries, you're very welcome.

Thanks for the quick reply and for the information.

I will have a go with a lower immobilisation level (5000 RU) and analyte concentration series, which should hopefully give me some better curves and more accurate fragment binding data. As you say, I now have a better idea of what fragments bind well and which do not, so I can narrow it down to those.

Best wishes,

Koko

Hi KoKo,

I would even go lower with the immobilisation ( www.sprpages.nl/immobilization/strategy ). Try to have an Rmax < 100 or even better < 50 RU for the best kinetic results. Your Biacore s200 is very capable of measuring low responses.

Arnoud

Hi Arnoud,

Thanks again for the tip (and for the link), I will definitely give that a go.

Best wishes,

Koko