Protein Crystallization Hits

This week there are two fascinating stories in Nature Methods each giving us a glimpse of what structural biology might look like in a decade or so. Both papers describe a technical tour de force, shooting jets of micro crystals into the beam of a X-ray free electron laser and collecting X-ray diffraction images.

The first report utilizes recombinant protein (TbCatB) crystals that are grown in Sf9 insect cells. Yes, that's right: protein crystals grown in vivo, no crystallization setups necessary here. 

Koopmann, R., Cupelli, K., Redecke, L., Nass, K., DePonte, D., White, T., Stellato, F., Rehders, D., Liang, M., Andreasson, J., Aquila, A., Bajt, S., Barthelmess, M., Barty, A., Bogan, M., Bostedt, C., Boutet, S., Bozek, J., Caleman, C., Coppola, N., Davidsson, J., Doak, R., Ekeberg, T., Epp, S., Erk, B., Fleckenstein, H., Foucar, L., Graafsma, H., Gumprecht, L., Hajdu, J., Hampton, C., Hartmann, A., Hartmann, R., Hauser, G., Hirsemann, H., Holl, P., Hunter, M., Kassemeyer, S., Kirian, R., Lomb, L., Maia, F., Kimmel, N., Martin, A., Messerschmidt, M., Reich, C., Rolles, D., Rudek, B., Rudenko, A., Schlichting, I., Schulz, J., Seibert, M., Shoeman, R., Sierra, R., Soltau, H., Stern, S., Strüder, L., Timneanu, N., Ullrich, J., Wang, X., Weidenspointner, G., Weierstall, U., Williams, G., Wunderer, C., Fromme, P., Spence, J., Stehle, T., Chapman, H., Betzel, C., & Duszenko, M. (2012). In vivo protein crystallization opens new routes in structural biology Nature Methods, 9 (3), 259-262 DOI: 10.1038/nmeth.1859 

The second paper describes a similar experiment, carried out with small crystals of the Blastochloris viridis photosynthetic reaction center grown within lipidic phases. The resulting images actually resemble conventional X-ray diffraction images with proper Bragg spots, good enough to build a somewhat meager 8.2 Å resolution electron density map.

Johansson LC, Arnlund D, White TA, Katona G, Deponte DP, Weierstall U, Doak RB, Shoeman RL, Lomb L, Malmerberg E, Davidsson J, Nass K, Liang M, Andreasson J, Aquila A, Bajt S, Barthelmess M, Barty A, Bogan MJ, Bostedt C, Bozek JD, Caleman C, Coffee R, Coppola N, Ekeberg T, Epp SW, Erk B, Fleckenstein H, Foucar L, Graafsma H, Gumprecht L, Hajdu J, Hampton CY, Hartmann R, Hartmann A, Hauser G, Hirsemann H, Holl P, Hunter MS, Kassemeyer S, Kimmel N, Kirian RA, Maia FR, Marchesini S, Martin AV, Reich C, Rolles D, Rudek B, Rudenko A, Schlichting I, Schulz J, Seibert MM, Sierra RG, Soltau H, Starodub D, Stellato F, Stern S, Strüder L, Timneanu N, Ullrich J, Wahlgren WY, Wang X, Weidenspointner G, Wunderer C, Fromme P, Chapman HN, Spence JC, & Neutze R (2012). Lipidic phase membrane protein serial femtosecond crystallography. Nature methods PMID: 22286383

Granted, all of this is currently in the proof-of-concept stage - no actual high resolution structure determined yet - but this is how new exciting breakthrough technologies often start out.  I'm wondering how long it will take for these technologies to mature to a state where they produce useful resolution structures and when they will become applicable to 'the rest of us'. Ten years, mid of the century maybe?

No protein crystallization setups, no crystal harvest, no cryo. X-FEL kills the crystallization champ.

This might change our game quite a bit.

Cheers,

Peter

For all those that are interested in simplifying membrane protein crystallization trials, you may want to check out this paper on the topic of 'simplifying LCP-based crystallization':

Wallace E, Dranow D, Laible PD, Christensen J, & Nollert P (2011). Monoolein lipid phases as incorporation and enrichment materials for membrane protein crystallization. PloS one, 6 (8) PMID: 21909395

Abstract: he crystallization of membrane proteins in amphiphile-rich materials such as lipidic cubic phases is an established methodology in many structural biology laboratories. The standard procedure employed with this methodology requires the generation of a highly viscous lipidic material by mixing lipid, for instance monoolein, with a solution of the detergent solubilized membrane protein. This preparation is often carried out with specialized mixing tools that allow handling of the highly viscous materials while minimizing dead volume to save precious membrane protein sample. The processes that occur during the initial mixing of the lipid with the membrane protein are not well understood. Here we show that the formation of the lipidic phases and the incorporation of the membrane protein into such materials can be separated experimentally. Specifically, we have investigated the effect of different initial monoolein-based lipid phase states on the crystallization behavior of the colored photosynthetic reaction center from Rhodobacter sphaeroides. We find that the detergent solubilized photosynthetic reaction center spontaneously inserts into and concentrates in the lipid matrix without any mixing, and that the initial lipid material phase state is irrelevant for productive crystallization. A substantial in-situ enrichment of the membrane protein to concentration levels that are otherwise unobtainable occurs in a thin layer on the surface of the lipidic material. These results have important practical applications and hence we suggest a simplified protocol for membrane protein crystallization within amphiphile rich materials, eliminating any specialized mixing tools to prepare crystallization experiments within lipidic cubic phases. Furthermore, by virtue of sampling a membrane protein concentration gradient within a single crystallization experiment, this crystallization technique is more robust and increases the efficiency of identifying productive crystallization parameters. Finally, we provide a model that explains the incorporation of the membrane protein from solution into the lipid phase via a portal lamellar phase.

This figure explains how this new PLI (post LCP formation incorporation [I know...]) method works:

Figure: Dispense lipid first, then add (purple protein solution), incubate a few hours, then add precipitation reagents.

The simpler, the better,

Peter

For the optimization of membrane protein crystal growth, it can make a huge difference if the 'proper' lipids are present in the crystallization experiment.

How do you add lipids to a crystallization trial? Lipids don't readily dissolve in water. There are several ways to get these amphipathic molecules to participate in the crystal formation process. When dealing with a standard vapor diffusion crystallization optimization, where the membrane protein detergent complex is combined with a precipitant solution, the lipid can be added in detergent micellar form. Better yet use less detergent and prepare a lipid film inside a glass container that is then solubilized by the detergent that is present in the solublized membrane protein sample. 

This makes it very simple to change the lipid composition in every single experiment.  A systematic approach to study the effect of lipids at comparably high concentrations has recently been described as HiLiDe:

Gourdon, P., Andersen, J., Hein, K., Bublitz, M., Pedersen, B., Liu, X., Yatime, L., Nyblom, M., Nielsen, T., Olesen, C., Møller, J., Nissen, P., & Morth, J. (2011). HiLiDe—Systematic Approach to Membrane Protein Crystallization in Lipid and Detergent Crystal Growth & Design, 11 (6), 2098-2106 DOI: 10.1021/cg101360d

The described re-lipidation is reminiscent of reconstitution of membrane proteins into bilayer membranes, while avoiding the second detergent removal step. I talked to the lead author Pontus earlier this year at the Keystone Conference and he explained to me that he thinks the lipid/detergent mixtures at high concentrations may form a generic crystallization environment for membrane proteins, somewhat reminiscent of lipidic cubic phases or sponge phases. The results (crystallization of  rKv1.2-beta2, E.coli Complex I and T.thermophilus Complex I and similar, previously reported crystallizations with high lipid composition such as pea LHC-II, bovine rhodopsin, bovine Cyt bc1, SERCA1a, pig Na/K-ATPase, Na/K-ATPase) speak for themselves.  

On the other hand, if the crystallization is carried out with the use of lipids that can spontaneously form a range of lipidic materials such as sponge or lipidic cubic phases, additional lipids can be added to the matrix lipid for crystal growth optimization. Let's say for instance, the matrix lipid is monoolein (good choice, by the way). This lipid has a melting temperature of 37C and in its liquid form can dissolve other membrane components, such as Cholesterol. How can this be done practically? To test three different Cholesterol concentrations, one can prepare a 20% Cholesterol in Monoolein mix by melting (i.e. 80 mg) Monoolein to 40C and dissolving the dry Cholesterol  (20 mg) in it, obtaining a 20% (w/w) mixture. Combining  this mixture with neat liquid Monoolein at a 50/50 ratio, one would obtain a 10% Cholesterol content (etc.).  Fortunately, Monoolein and lipid mixtures with Monoolein can be supercooled (i.e. remain liquid at room temperature for many minutes), allowing simple manipulation with pipettors or syringes prior to mixing with the membrane protein to form a lipidic cubic phase (or other lipidic materials).

In case you're still not convinced about the utility of amphiphilic compounds in membrane protein crystallization, a good case is made here:

NOLLERT, P. (2005). Membrane protein crystallization in amphiphile phases: practical and theoretical considerations Progress in Biophysics and Molecular Biology, 88 (3), 339-357 DOI: 10.1016/j.pbiomolbio.2004.07.006

;)

Regards,

Peter

Once you've got a hit with your membrane protein crystallization trial, your life may get really exciting. After confirming that the hit is actually a protein crystal and obtaining even the weakest X-ray diffraction patterns, you're getting into the optimization game. One of the obvious parameter to optimize is the crystallization cocktail. How can that be done without wasting your precious protein sample on unproductive crystallization conditions?  

Designing and making grid-screens around a particular hit condition is complicated due to the presence of detergent in the crystallization experiment. An established approach is to be close to the detergent phase separation boundary while avoiding the 'heartland' of phase separation. How can this be done practically? You could check out the various papers on this topic and spend a lot of time searching and possibly finding hints on formulations that are relevant to your hit. You could also search the phase boundary conditions in a pre-screen type of experiment (a la Song & Gouaux: Membrane protein crystallization: application of sparse-matrices to the alpha-hemolysin heptamer Methods in Enzymology(1997) (60-73)). The quickest way though to obtain guidance on optimization screen design is through mining of existing detergent phase boundary data. Fortunately Mary Koszelak-Rosenblum et al. from HWI have made such data mining a quick and simple experience:

Koszelak-Rosenblum M, Krol A, Mozumdar N, Wunsch K, Ferin A, Cook E, Veatch CK, Nagel R, Luft JR, Detitta GT, & Malkowski MG (2009). Determination and application of empirically derived detergent phase boundaries to effectively crystallize membrane proteins. Protein science : a publication of the Protein Society, 18 (9), 1828-39 PMID: 19554626

The data comes in an Excel spreadsheet called SLICKSPOT.  This tool is available for download here.

The input parameters for this nifty tool are type and concentrations of:

• Detergent (data is available for C10M, C12M, b-HG, b-OG, b-NG, CHAPS, LDAO, C12E8, C8E4, C8E5, FC-12),
• crowding agent PEG (data is available for: PEG400, PEG1000, PEG2000, PEG2000MME, PEG3350, PEG 4000, PEG5000ME, PEG6000, PEG8000, PEG20000)
• Salt (data is available for CaCl2, KCl, LiCl, Li2So4, MgCl2, Na2C3H2O4, NaCl, NaH2PO4, (NH4)2SO4, NH4H2PO4, (NH4)2HPO4)

So, let's say the membrane protein crystal hit was produced at room temperature with a formulation that contained 1% Octylglucoside,  Ammonium sulfate and PEG4K. For these conditions Slickspot produces the following output as a chart:

Example of a Slickspot output. How to read this customized chart: Salt and PEG concentrations below the blue curve form a single phase, those above are phase separated. Sticking to conditions around the blue line is preferred.

So, what's to do from here?

Design your customized formulation screen with Ammonium sulfate and PEG4000 conditions  near to the blue line.

Pretty slick, hm?

Peter

PS: I just realized that Slickspot fails to update the legend when additional conditions are queried; a cosmetics-only bug I hope.

Earlier this month I received an email from John Wiley & Sons, confirming that 'The supplement file is now available at http://onlinelibrary.wiley.com/doi/10.1110/ps.073263108/suppinfo'. More than half a year ago I had pointed out that the 'supplementary material' to a paper was not available online. Now it is. Took a while, but  THANKS A LOT for following through, John Wiley & Sons!

When you go to this page, you can download Simon Newstead's alpha-MP-database.xls.

Although somewhat outdated, this spreadsheet contains a concise summary of crystallization conditions from 121 alpha helical membrane protein structures that are listed in the PDB. The extensive analysis of crystallization conditions is described in the accompanying paper:

Newstead S, Ferrandon S, & Iwata S (2008).

Rationalizing alpha-helical membrane protein crystallization.

Protein science, 17 (3), 466-72

PMID: 18218713

While this accounting exercise is not the sexiest of all science, it is of great interest to all of us membrane protein crystallizers . The database is a handy tool that can be consulted when only small quantities of membrane protein sample is available and one is forced to focus on 'what's worked in the past'.  And of course, this data summary is incredibly for the design of new (although highly biased) crystallization matrices. This is exactly what Simon has done: based on his own analysis, a new sparse matrix crystallization screen, called MemGold  is described.

Great work!

Peter