Combinatorial Optimization - some comments

Imagine the following scenario: you have run several screens with a new protein sample, and you've found poor quality hits with needles or small crystals in the following three conditions:

Looking at these results, it's hard to draw up a plan for optimization.  Is the best precipitant PEG, or a salt?  If PEG, does the molecular weight matter?  Are the bivalent cations Mg2+ and Ca2+ important?  And which pH is best?

Obviously, you could optimize each of these hits individually by varying just the ingredients from that one hit (for example using our XSTEP Optimization software).

However, this may not be the most efficient approach.  What you would really like is a “targeted screen”, using just the ingredients in the three successful hit solutions.  This is because ingredients may ideally need to be “reshuffled”, for example by combining PEG4000 with, say, CaCl2 and imidazole.

Douglas Instruments has introduced a new script for Combinatorial Optimization, which allows e.g. the precipitants to be dealt with separately from the other ingredients so that this reshuffling can take place.

The diagram below illustrates the general approach.  The precipitants P1, P2 etc can be placed in the reservoirs of a plate, arranged in columns (this could be done with a 12-channel pipette); the additives A1, A2 etc can be placed in the PCR tubes on the right.  These additives are automatically added to the drops by the script, on a row-by-row basis as shown.  As an example, in the case above, P1 might be 30% PEG4000, P2 might be 25% PEG4000, P3 28% PEG400 etc, while A1 might be 0.2M MgCl2, A2 MgCl2 plus TRIS, A3 MgCl2 plus HEPES etc.

Over the years I have heard many times how successful such targeted screens can be.  Very often they allow crystallizers to go from a set of disparate, poor quality hits to diffracting crystals in a single plate.  The trouble has been that they were time-consuming to set up, and - until now - separate robots were required to make the targeted screen and to set up the drops.  The new Combinatorial Optimization script is based on a simple idea, but it allows a (systematic) targeted screen to be set up in 15 to 20 minutes.

Note that the same script can be used for several other powerful classes of experiments.  For example, microseeding often gives too many crystals.  Using the script, seed stocks with various dilutions can be added to one or more hit conditions.  (Typically a dilution series would be set up, running from the "neat" seed stock to a 1:100,000 dilution.)  Or, during optimization, gradients of one or more precipitants could be placed in the reservoirs, with up to 12 unique additives in the PCR tubes.  In all cases the power of the script lies in its ability to make unique combinations of ingredients (or concentrations) in each well.

Patrick Shaw Stewart, October 2011.

.