Well, it’s been another week of not too much actual lab work and quite a lot of computer work. The lab work I did do was incredibly frustrating due to spending time searching around for reagents and then having a water problem with the liquid handler that is used to set up the fine screens. But let’s backtrack a bit …
I got some hits for ACVR1 co-crystallised with three different compounds (crystal pics in the previous blog post). I mounted ten of these crystals (with K62821a) to go on the next synchrotron trip to assess their diffraction quality. There were some crystals with the other compounds, but they were very fragile and tended to fall apart as soon as I touched them. Unfortunately, there wasn’t time to screen them on the synchrotron trip this week, so they will be going back on Saturday (tomorrow) for screening. In the meantime, I designed some fine screens around the conditions that I had obtained hits in.
For ACVR1 with K62821a
- 20 % PEG 8000, 0.1 M CAPSO pH 9.5 (varied the PEG from 10-24 %)
- 10 % PEG 8000, 8 % ethylene glycol, 0.1 M HEPES pH 7.5 (varied the PEG from 4-18 % and the ethylene glycol from 4-10 %)
For ACVR1 with K62981a
- 50 % ethylene glycol, 0.2 M MgCl2, 0.1 M Tris pH 8.5 (varied the ethylene glycol from 30-60 %)
For ACVR1 with K62980a
- 20 % PEG 6000, 10 % ethylene glycol, 0.1 M HEPES pH 7.0, 0.1 M CaCl2 (varied the PEG from 12-26 % and the ethylene glycol from 6-12 %)
Unfortunately, the liquid handler that sets up the new screens ran out of water at some point during my run, so I had to start again, but there wasn’t enough sterile water, so I ended up putting it off to this week, and was finally able to set up the screens yesterday. I’ll be setting up crystallisation plates this afternoon with frozen protein, which I normally don’t use (I would normally put it through the gel filtration again before use). However, frozen protein has been used in the past. I don’t have quite enough to gel filtrate, and am keen to get these plates set up as soon as possible.
Fragment screening data processing
As for the fragment screening, I now have over 100 datasets with good resolution, and my model is just about ready to go for the bulk data processing of those datasets. My model is in space group P 21 21 2, although many of the datasets have been processed in different space groups, such as P 1 21 1 or C 2 2 2. I am hoping my model will work for these sets. I have already done about 70 rounds of refinement on my model, with the last week being spent watching the Ramachandran favoured, clashscore and rotamer outliers go marginally up then marginally back down again, no matter how many times I go into Coot and correct things. The Molprobity stats for a recent refinement are below (Table 1). I have now managed to get the clashscore down, but am still working on the Ramachandrans and rotamers. With any luck, I can feed this model to DIMPLE, the bulk data processing software) today.
Table 1. Current Molprobity statistics for 1.9 Å structure of ACVR1 co-crystallised with LDN-193189
|Clashscore, all atoms:||2.73||99th percentile* (N=771, 1.890Å ± 0.25Å)|
|Clashscore is the number of serious steric overlaps (> 0.4 Å) per 1000 atoms.|
|Poor rotamers||9||0.88%||Goal: <0.3%|
|Favored rotamers||973||95.67%||Goal: >98%|
|Ramachandran outliers||1||0.09%||Goal: <0.05%|
|Ramachandran favored||1150||98.21%||Goal: >98%|
|MolProbity score^||1.06||100th percentile* (N=11926, 1.890Å ± 0.25Å)|
|Cβ deviations >0.25Å||0||0.00%||Goal: 0|
|Bad bonds:||0 / 9555||0.00%||Goal: 0%|
|Bad angles:||1 / 12997||0.01%||Goal: <0.1%|
|Peptide Omegas||Cis Prolines:||0 / 44||0.00%||Expected: ≤1 per chain, or ≤5%|