Fragment-based screening is now well-established as a powerful approach to early drug ("lead") discovery.
Once appropriate crystals are available, the experiment entails:
Not only do those drops need to be identified that have crystals, but also for each drop the exact location must be captured where the ECHO should deposit the compound solution (<100um accuracy). Our initial anecdotal evidence is that shooting directly at the crystal will damage it, but we are evaluating various different targeting protocols.
The program TeXRank (Ng et al, 2014, ActaD), has been adapated to turn this laborious process into a matter of minutes: images with likely crystals are placed top of the inspection queue, and the targeting point is selected and stored with a single button-click.
Images of crystal drops can be recorded at Diamond, but if you wish to use images from your own imagers, the software's drop-picking algorithm needs a baseline image for each well, so we'll ask you to send some images of an empty plate without drops. Once you send that and the most recent images of the plates you're bringing, the image processing itself will happen at Diamond.
Compounds are transferred to the crystals using an ECHO (Labcyte) acoustic liquid handler. The crystal plate is inverted over the plate with the compounds, and the ECHO uses ultrasonic pulses to shoot series of 2.5 nL droplets of solution at the crystal drops, as many as required to make up the required dispense volume. The dispense coordinates are uploaded from the TeXRank output. Dispensing is very fast: ~30sec for 100s of compounds.
Soaking incubation times are dictated by how well the crystals survive at a given solvent concentration; one hour is probably sufficient, but it's often convenient to do longer - whereas some systems can't tolerate more than a few seconds, which then needs special treatment.
Crystals are harvested manually, but the pain of doing this for 100s of crystals has been greatly reduced by the Shifter, a microscope x-y stage that also handles unsealing and resealing as well as automatically tracking samples. This trick is surprisingly effective, and typically >100 crystals/hour can be harvested. (Current record: 200 xtals per hour, sustained.)
We have a well-curated collection of loops, with tools that allow them to be cleaned both rapidly and with minimal damage. This certainly speeds up harvesting, but more importantly, it makes the loop centring algorithm on the beamline very reliable indeed (>97%).
The Uniwand and Unitray are now an official Mitegen product, and we highly recommend it for any lab setting.
Data can be collected in automated, unattended mode: crystal-to-crystal is <2min for strong diffracators (60sec X-ray exposure), so that almost 700 crystals can be collected in 24 hours. Since the upgrade in 2015 to the BART robot, 24hrs worth of crystals can to be loaded at once, and these will run fully unattended.
Data are integrated automatically thanks to the existing processing infrastructure; and for evaluation and analysis of datasets, PanDDA and XChemExplorer were have been developed at the SGC, vastly simplifying the challenge.
Data are integrated automatically thanks to the existing processing infrastructure; and for evaluation and analysis of datasets, PanDDA and XChemExplorer were developed at the SGC, both vastly simplifying the challenge and enormously increasing the sensitivity of detecting hits.
For quality control of the fragment library and periodically checking its state, an LCMS (liquid chromatography / mass spec) system with coupled with ELSD (evaporative light scattering detector) is in place. For each compoun, a set of readouts (mass peaks; integrated ELSD response; UV intensity) provide a fingerprint of the state of each compound, and changes in fingerprint indicate problems with a compound.
Update, 2016: the method remains in development, but we have also found very few errors in our library. Libraries are aliquoted into plates, and during use, plates are periodically refreshed by evaporating and replacing solvent. The most reliable QC is providec by the ECHO, which reports which drops could not be dispensed from.
Diamond Light Source is the UK's national synchrotron science facility, located at the Harwell Science and Innovation Campus in Oxfordshire.
Copyright © 2018 Diamond Light Source
Diamond Light Source Ltd
Harwell Science & Innovation Campus