The purified binder does not exist in a monomeric state.

[Hai-W-L]

Hello,

Thank you for sharing this practical software. I have encountered some issues in my experiments and would like to ask for your advice.

I designed a series of binders with a length of approximately 250–300 aa using BindCraft, followed by protein expression in BL21 (E. coli) cells and subsequent purification. I found that nearly all binders were expressed at extremely high levels, but none appeared to be in the monomeric form. Size-exclusion chromatography (SEC) profiles revealed that they existed as dimers or higher-order oligomers. I would like to enquire if anyone else has encountered a similar situation.

To date, I have tested around 15 binders via pull-down assays, and no binding activity was detected.

[martinpacesa]

Dimers are normal, higher order oligomers we rarely see. How did you purify, what buffers used?

[Hai-W-L]

Hi, thanks for your response.
Here’s what I’m confused about: When I predicted the complex structure of the binder dimer bound to the target, the binding pose, PAE and other key parameters changed drastically. So I’m worried whether the expressed binder’s dimerization will affect its binding to the target protein — do I need to account for this?
Also, I found that the structures of BindCraft-designed binders are very different when re-predicted with AF3, so I’m also unsure how to pick the right candidates.
My binder purification workflow: Expressed in BL21 cells, lysed in buffer (50 mM Tris, 150 mM NaCl, 1 mM TCEP), incubated with Ni resin, washed to remove contaminants, then eluted with 300 mM imidazole. I then performed sequential purification with a HiTrap Q column and a Superdex 75 10/300 column, and found that most of the final purified protein exists predominantly as a dimer.

[martinpacesa]

The binder dimer prediction is not consistent, we have benchmarked that internally, so we don't really do that. The dimerisation can affect binding but most probably not to a level that it won't bind at all, plenty of our high affinity binders are dimers in solution.

The difference with AF3 -> yes, this we see often, that AF3 places it in a different spot, but so far the AF2 predictions turned out to be the correct ones when probed by mutants/competion/structure.

Purification looks good (if the buffer has decent amount of salt after superdex). One can also increase salt concentration to 250-500 mM.

[Hai-W-L]

Thanks for your response.

Got another BindCraft question.
I’m running BindCraft for 8–13 aa short peptide binder prediction with these settings:

python -u /home/BindCraft/bindcraft.py
--settings /data/*******/.json
--filters /home/BindCraft/settings_filters/peptide_relaxed_filters.json
--advanced /home/BindCraft/settings_advanced/peptide_3stage_multimer_mpnn_flexible.json

No output happens super often.

Then I cropped the target properly as suggested, and got some hotspot-binding peptides. But when I re-predicted with full-length target + these peptides (ColabFold or AF3), a lot of them shifted to other sites. Any screening tips for this?