Protein Crystallizers know this simple rule: for crystallization to occur protein samples should be pure, homogenous and the protein be in a stable and non-aggregated state. These requirements can often be met by applying standard purification and concentration procedures to soluble protein targets. This is one of the reasons for the productivity of high-throughput structure genomics-type efforts. The buffer type, pH and salt are typically standard systems - such as 100 mM NaCl, 50 mM Tris-HCl pH7.5 - that rarely requires optimization for a particular protein target.
These requirements are also valid for membrane proteins, but achieving them often requires a lot more effort. A typical parameter to evaluate for membrane proteins is the detergent type and its concentration. Such a crystallization pre-screen can be done sometimes even before starting with the purification and, the procedure applied is called 'detergent exchange'. It turns out that there's a handful of methods available to carry out such detergent exchanges, for instance, using size exclusion chromatography, simple dilution or ultracentrifugation. The logic goes like this: once the membrane protein aggregates in the newly tested detergent solution, it can be detected as an extra peak, running with the void volume in the sizing chromatogram, as an increase in turbidity, or as a decrease in protein concentration in the supernatant of an ultracentrifuge tube.
Anybody who has concentrated a protein sample using a simple MWCO filter realizes, that filtration provides a good means to separate aggregated from non-aggregated protein. Indeed, GE Healthcare's multi trap purification filter kit utilizes this feature for purification scouting of His-tagged proteins. Essentially, the protein is bound to a resin and eluted by filtration through a MWCO filter. The conditions that yields most protein in the filtrate wins. Michael Wiener's membrane protein group has given this concept an interesting twist. They describe in this paper:
Vergis JM, Purdy MD, & Wiener MC (2010). A high-throughput differential filtration assay to screen and select detergents for membrane proteins. Analytical biochemistry, 407 (1), 1-11 PMID: 20667442
an adaptation of the filtration concept to detergent screening and have devised a differential filtration assay that allows to identify those detergents that render a particular membrane protein target 'well behaved' and hence, more crystallizable. How does this work?
Instead of using the individual filtration devices, two 96-well filter plates of different MWCO (the use of two different MWCO sizes allows to distinguish between stability and size) are used sequentially. 94 different detergents can be used in one go. Some people may consider testing with 94 different detergents overkill, but I'm with Michael in his 'knock em dead' approach. However, using fewer detergents and spin-filters may prove more practicable in many labs. In fact, this useful review on membrane protein crystallization parameters:
Newstead, S., Ferrandon, S., & Iwata, S. (2008). Rationalizing α-helical membrane protein crystallization Protein Science, 17 (3), 466-472 DOI: 10.1110/ps.073263108
can be used to argue that by just testing 8 instead of 94 different detergents one can cover more than 80% of the 'detergent landscape'. The table below shows the ranking of detergents that have provided most membrane protein crystal structures. Note that followers of the Pareto Principle (80% of the bang for 20% of the buck) will do with a set of only 8 detergetns or filter tests.
Table: Which detergents to use for membrane proteins? Shown are detergent types and occurrences in successful membrane crystallization screens. Data from supplemental info to Newstead et al., 2008 paper, kindly provided by Simon Newstead. Note 1: Am I seeing this right that HEGA-10 is not included in the set of 94 used by Wiener?; Note 2: adhering to this table will increase investigator bias.
|
Detergent to solubilize membrane proteins
|
# (131)
|
%
|
|
DM (dodecyl maltoside)
|
14
|
11%
|
|
DDM (decyl maltoside)
|
32
|
24%
|
|
NG (nonyl glucoside)
|
14
|
11%
|
|
OG (octyl glucoside)
|
17
|
13%
|
|
LDAO (lauryl dimethylamineoxide)
|
10
|
8%
|
|
C12E8 (octaethylene glycol monododecyl ether)
|
7
|
5%
|
|
C12E9 (polyoxyethylene(9)dodecyl ether)
|
6
|
5%
|
|
UM (undecylmaltoside)
|
4
|
3%
|
|
NM (nonyl maltoside)
|
3
|
2%
|
|
HEGA-10 (hydroxyethylglucamide)
|
3
|
2%
|
|
|
110
|
84%
|
When I saw Michael presenting his differential filtratio assay at the NIH Roadmap Meeting in March 2009 I was particularly excited by these features:
(i) the low sample amounts required (only 10 ug protein / well to create sufficient signal in the dot-blot / western assay) and,
(ii) the type of data that can be pulled out of this assay.
It has taken me a while to appreciate the utility of the 'size stability quad plot'. The gist is that the membrane proteins can be grouped into cohorts that are formed by the quandrants as defined by stability and size. This sorting of detergents helps to predict those detergent species that should work best for crystallization.
And that's exactly the type of result you're expecting from a crystallization pre-screen.
All the best,
Peter