Mapping and manipulating temperature-concentration phase diagrams using microfluidics
Searching for the correct conditions under which a specific protein crystallizes can be both an expensive and time consuming process, yet is an essential step if any work with the protein is to be done. Recently there was been efforts made towards creating automated microfluidic methods to search phase space for the necessary conditions. The advantages of such screening methods are its low cost, since only small volumes of the protein are needed, along with precise control of the conditions. This paper presents a method developed by the group that searches a two dimensional parameter space (protein concentration and temperature) to create a phase diagram of protein crystallization.
The group has developed a microfluidic device which they call PhaseChip. The device is composed of two layers, both made of PDMS. The top layer consists of a grid of wells (which are of either nL or pL volume, depending on the chip) in which the solute/water drops are stored. These drops are separated by a thin PDMS membrane from the bottom layer, which has a series of reservoir channels through which solutions of water with different concentrations of salt are flowed. The concentration of the protein in a droplet is controlled by the salt concentration in the reservoir it is attached to. The difference in chemical potential between a droplet and its reservoir causes water to flow to/from the reservoir and the droplet through the PDMS membrane (which is water permeable but not protein or salt permeable), until equilibrium is reached. The group has a series of channel reservoirs, with each channel in contact with one column of wells. Thus all the wells along this column have the same concentration. The salt concentration of the reservoir channels goes from a gradient of high to low. This gradient is created by combining two different flows of high and low salt in different proportions using a tree like structure, where the number of unique concentrations is equal to the number of branches on the last level.
A temperature gradient is set up along the chip as well, in the direction perpendicular to the concentration gradient. This is done by placing two thermoelectric coolers on either side of the grid, and setting them at different temperatures. This gives all the wells in the same row the same temperature, while wells in different rows have different temperatures. This gives us now a grid, where each well has a different temperature and a different concentration. (Actually, there were only 12 distinct reservoir concentrations, and many more columns, so there was redundancy in the concentration. However, in principle there could be as many distinct reservoir concentrations as one would like.) This becomes an effective way to search phase space for the conditions under which the protein crystallizes. By just looking to see which of the wells crystallize and which do not, an effective phase diagram can be found.