[ccp4bb]: SUMMARY:Inhibitors do not show up...

[Zsolt Bocskei <Zsolt.Bocskei@sanofi-synthelabo.com>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

Dear All,

I am grateful to Sherry Mowbray, Rik Wierenga, George Kontopidis, Gerard Bricogne and Bernhard Rupp for their reflections to my question
Please find below a summary.
The question:

Recently I have found no inhibitors bound in an alarmingly large portion of the co-crystal structures that I examined. I would be grateful if people could send me their comments, experiences etc. on this topic. 
This seems to occur with relatively weekly binding, water insoluble inhibitors (let's say in the micromolar range). I dissolve them in DMSO or DMF in quite a high concentration (10-100 mmol) and then I replace 1-10 % of my crystallization mother liquor (or a more concentrated version of that) with this DMSO/DMF solution of the inhibitor and then I soak my "empty"/non-complexed crystals in this liquid. I have to admit that after mixing the inhibitor solutions with the crystallization mother liquor a lot (or most, or maybe even worse: perhaps all) of the inhibitor precipitates, but one can read papers in the literature when adding the powder form of the inhibitor to the crystallization droplet has also produced the required complexes and indeed I have had cases like that, too.
So again, in what portion of their experiments do other people get useless "results" like the ones described above and how do they treat these cases. If this is so frequent then I find can one consider the results of the high throughput crystallographically driven screening approaches reliable at all (many false negatives)? 

Rik Wierenga

If you use cryoprotectants, they migh compete away your inhibitors.
regards,
rik wierenga

My comment

I haven't seen cryoprotectants either in my density maps at the actives sites in the referred cases

George Kontopidis:

There are several reasons for inhibitors do not show up in electron density
maps after soaking.
Some of them listed here.
- Solution binding studies of protein-inhibitor complex was done in
different conditions, usually low buffer low salt concentration and not in
high salt or PEG or DMSO/DMF solution. In Brown, N., et al Nature Cell
Biology vol1 7 p438 1999 the ligand did not show up in electron density
after soaking in high salt concentration but it showed up after soaking to a
nonpolar medium.
- The ligand is not soluble enough to give you 100% occupancy or occupancy
high enough to show up in electron density maps (what is the minimum
occupancy for a ligand to show up in electron density maps see Su-Ying Wu,
et al Angewandte Chem. Int. ed. 40, No. 3, 582-586)
- Finally one of the reagents in your solution might competes with the
inhibitor in the same binding site. This is the case from personal
experience in small molecules bound to cyclophilin A when they dissolve in
DMSO. 

Cheers

George

George Kontopidis, PhD
Cyclacel Limited
James Lindsay Place
Dundee DD1 5JJ
Scotland, U.K.
Tel: :+44 (0) 1382 207 577
Fax:+44 (0) 1382 206067
E-mail: gkontopidis@cyclacel.com <mailto:gkontopidis@cyclacel.com> 
Internet: www.cyclacel.com <http://www.cyclacel.com> 

My comment

I found the Angewandte paper very interesting and somewhat surprising. It seems that visual inspection cannot detect 25% occupancy even though in a favorable case the occupancy of a known model can already be refined. It was also shown that in the case where the Kd value of the given inhibitor is around 20 mM, a ligand concentration of around 20 mM was necessary to produce a density map where a bound molecule could  already be visually detected (due to an occupancy of about 0.4-0.45). This sort of concentration range is clearly inaccesible in my cases, although due to  stronger binding, lower concentrations should suffice for me. In fact, usually I am targeting 1 mM final concentration in the soaking solution, but due to low solubility most of the inhibitors precipitate.

Berhnard Rupp wrote that for him soaking worked alarmingly well.

Sherry Mowbray reminded me that one can also find quite a large number of false positives, where a reported ligand conformation is not substantiated by electron density.

I also had a question from Gerard Bricogne concerning the procedure I use to detect the inhibitors. He was wondering if I used difference maps calculated between structure factors computed from a model and amplitudes from my putative complex or I make use of the measured amplitudes from the native target protein. I had to confess that in the given cases I used the first (lazy) approach, even though I vaguely remember that I  had heard Gerard talking about the importance of using the measured amplitudes for this purpose. 

Thanks again for everybody who contributed.

Zsolt 

Dr. Zsolt Bocskei
Sanofi-Synthelabo Recherche
16 rue d'Ankara
67000 Strasbourg
France
Tel:33(0)388454116
Fax:33(0)388459070