Easier Way to Convert Proteins into Crystals

Roland Piquepaille writes "As you might know, proteins need to be transformed into 3-D crystals before their atomic structures and their properties can be analyzed. And production of high quality crystals from proteins has been a difficult task until now. But scientists in the U.K. have successfully used a porous medium, or 'nucleant,' a material that encourages protein molecules to crystallize. Their first step towards 'holy grail' of crystallography could help speed up the development of new medicines and treatments."

Comments

This is Big

[eldavojohn]

Ok, so I don't know a ton about nuclear medicine, I know just enough to be dangerous. Protein crystallization allows us to see it's structure whereby we better understand its function.

The reason this bit of news is so big is that it will (hopefully) allow researchers a way to quickly look at the structures of proteins in such as (in the second link) infectious diseases transmitted by prions, or protein particles. Prions seem to be pure protein; they contain neither DNA nor RNA.

If we can understand the shape and formation of proteins, we can understand how viruses and cells work because proteins are the building blocks. Viruses are obviously first on the chopping block as they are the smallest and infect millions of people world wide (AIDS, influenza, the common cold, etc.).

Re:This is Big

[ruckerz2k]

Though the parent is correct, this technology greatly reduces the time and effort involved in 'crystallizing' proteins. Most common approach is to use the hanging drop method, where a drop of the sample is suspended over a highly concentrated solution. The sample concentrates due to the negative osmotic pressure and the protein 'crystallizes'. The crystallization process can be hastened by using a 'nucleant', usually a small crystal of the sample that you have previously crystallized. Also, the exact identity and composition of the concentrated solution is varied in order to find the right crystallization conditions. This is a very tedious process (imagine setting up 96 different concentrated solutions, each differing in about 1% concentration of the solutes) and time intensive.

The discovery of a 'universal' nucleant (close to the one suggested by the authors of this study) and the development of a matrix to encourage crystallization would greatly speed the screening process, and ultimately, crystallization of proteins.

Re:This is Big

[SIGFPE]

Talk about Karma whoring! A bunch of sentences culled from a variety of sources from someone who really doesn't know what they are talking about. Now that's dangerous.

Viruses are obviously first on the chopping block...

Non "obviously" at all. There are countless medical applications for X-ray crystallography. Any time you want to study the structure of a protein it comes in handy. Many diseases are attacked by researchers from the point of view of receptor binding - the binding together of proteins to other compounds like a lock and key. Such receptors act like switches activating or controlling biological processes. These are ubiquitous in nature and understanding the shape of these 'locks' and 'keys' can be useful in trying to understand the mechanisms of all kinds of diseases whether or not they are caused by pathogens.

Re:An Even Easier Way to Convert Proteins to Cryst

[napalmfires]

http://en.wikipedia.org/wiki/Urine Urine doesn't have protein in it...

More importantly...

[cryptochrome]

By speeding up the (currently extremely tedious) process of crystallization and hopefully making inroads into the ~70% of all protein which currently can't be crystallized, this will rapidly improve our understanding of the structures of whole classes of proteins.

Understanding protein structure..

[jonasmit]

is critical to translating the information obtained in, for example, the Human Genome Project. The DNA gives us the blueprint but the protein does the work. Currently, there is no way to predict protein structure from DNA. Therefore, you must see the structure to understand how the protein works. Also it is important to note that in Protein-Protein interactions. Protein-Protein interactions are important in normal cell singalling events as well as in how virii infect cells (like HIV1 binding to gp120/gp41).

Re:Understanding protein structure..

[Asam]

There are ways to obtain structure of proteins from DNA, just that they're not so good if a similar structure doesn't exist in the Protein Data Bank (http://pdb.org/ already, its called homology modelling. You have the protein sequence of an unknown structure then look for similiar sequences in the 20,000 known protein structures. That's why the more structures we add to the PDB makes it easier to find structures of unknown proteins.

Nobel Prize material

[smellsofbikes]

To find the three-d structure by x-ray crystallography, you have to crystallize the protein. Actually doing so, with different proteins, is an astoundingly difficult task, so much so that something like five Nobel prizes have been given for research into crystallization and x-ray crystallography development, and another ten or so Nobels given for determination of three-d structure of various proteins were, in essence, awarded for getting the protein to crystallize.

Side story: there was a famous German chemist named Emil Fischer, who originally determined the structures of a bunch of sugars. That was, again, largely a crystallization problem. He had, as Germans did in the 1890's, an enormous beard, and was playing with chemicals all day long, which tended to condense in his beard. It was said that if you could not get something to crystallize out of solution, no matter what you did, you asked Fischer to come to your lab and fluff his beard over your beaker, and the seed crystals falling from it were of such variety that one was almost guaranteed to be correct for your particular situation and get it to crystallize. So this isn't exactly NEW technology.

This should impact future graduate students

[radiashun]

Hopefully this will encourage more individuals to pursue advanced degrees in protein crystallography. I was recently at a talk where a soon-to-be PhD was discussing her crystallography work. She said that many people choose to pursue other areas in biochemistry/structural biology because protein crystallography is very unpredictable. Some proteins will crystallize in months while others can take YEARS! Waiting years before you can really dive into your PhD research is very discouraging.

Re:W00t!

[Red Flayer]

Did you check the submitter link? rel=nofollow. Maybe if you'd read the commments by Taco in last weeks /.metaarticle, you'd see why.

Re:This is Not so Big

[sam_handelman]

This is an improvement on a known technique. The abstract is as over-reaching as the press release (the linked article).

  I'm not a crystallographer, but I work in a lab group that has many crystalographers in it.

  It's been known for some time that you can use a variety of materials - including things with porous surfaces, which is what is used here - to assist the process of crystallization. Crystalization is difficult and, frankly, rather unscientific - you take the protein you want to crystallize, and you try different techniques and tricks (of which porous nucleants are an example) until you can get it to work.

  So, okay, it would be a "holy grail" if you could find one technique that would let you crystallize most things without going through all that trouble.

  However, based on only seven examples (Subscribers only, I'm afraid.), you absolutely cannot conclude that this is a universal nucleant - based on the similarity among the seven examples, I'd be very surprised if it were; even if it were a universal nucleant, nucleation does not always guarantee usable crystals.

  Those caveats aside, it does look like a useful advance.

Re:Ever hear of NMR structure determination?

[Red Flayer]

NMR diesn't rquire crystallization, but it does require transfer of the protein to a non-native solution (which may affect tert structure). Not that crystallization doesn't do this also...

Plus, NMR results or more vague than X-Ray crystallography, and can only be used with small proteins, whereas crystallography works for even very large proteins (provided you can get them crystallized).

Re:This is Big but not that big

[Asam]

Sure being able to find suitable crystallization conditions for proteins is a bottleneck at the moment and this will aid in the process and add to the 20,000 plus know Protein structures, however, this is limited to certain types of proteins. A lot of really interesting proteins however are more flexible, and its this flexibility thats the key to understanding a lot of protein function. Structures derived from crystals don't give so much information about protein dynamics and molten globule like states of proteins, in fact when you crystallize proteins they tend to be locked into one conformation. Everyone in the field should keep this in mind always. A protein structure derived from a crystal structure is just a framework, its not a final representation of final function. More clues about this kind of information can be derived from NMR (Nuclear Magnetic Resonance), however the use of NMR is restricted to relatively small proteins. This is going to help, but its not huge, the real breakthrough will come when we can model any given protein sequence on a computer and have an acurate predicated 3 diminsional structure together with it molecular motions and dynamics in real time. Glad to see /. covering the field, its huge and its just started, we're at the begining of an exponential curve. Maybe /. should make a seperate section for this, I'd certainly be interested in contributing articles and news to such an effort.

Or don't crystallize at all(!)

[infolib]

It might be possible to determine protein structure with just a single molecule, no need for crystallization at all, e.g. with free electron lasers

The brilliance of x-ray sources are right now undergoing a revolution much faster than Moore's law.

Re:Ever hear of NMR structure determination?

[PDoc]

NMR is great, because it's fast, relatively easy to run, and tolerant of substrate. But it isn't absolutely accurate. Techniques such as nOesy and cosey allow for determination of stereochemistry via two-dimensionaly NMR, showing contact through bonding and through-space interactions. However the data from a nOe experiment can be inconclusive, especially in cyclic systems. Only X-ray data can actually confirm the true structure then.
A friend, also doing a Ph. D in chemistry, has just binned half his thesis because the NMR data lead to one conclusion, only to be contradicted by the X-ray data. And this wasn't a small difference either...
Oh well, back to the lab then...

Re:Roland Piquepaille knows nothing.

[cruachan]

I'm afraid not. Nothing beats having an accurate structure from Px. I spent several years as a postdoc attempting to grow large crystals of a membrane protein (at the time one of the first three or four membrane proteins to be crystallized). As we really were interested in knowing the structure rather than how we got it as light relief from purifying the protein on a near industrial scale and seeding thousands of crystallization trials we tried every other structural analysis method we could get our hands on. Many of these gave us interesting and invaluable info, but in all cases they were of most interest after the crystal structure was solved and the results could be interpreted in the light of that.

With most Protein Chemistry having the Px structure is the Gold Standard, and you can't really say you understand a protein until you have it. Trouble is, that as the article says, getting crystals is hard, slow, and extrodinarily painstaking work

crystallization is tough stuff

[!splut]

IAASB (structural biologist), and while I can't verify their findings, I can back up the premise in the article that generating diffraction-quality protein crystals is one of the two major bottlenecks to X-ray crystallography (the other being purification).

It's pretty easy to understand why. Not only do you need pure protein, but one must find conditions under which that protein forms relatively large, single crystals. The chief variables, aside from the homogeneity of the protein you're starting off with, include temperature, pH, protein concentration, choice of and concentration of precipitant (generally a chemical that drives the protein out of solution), choice of and concentration of additive compounds, in some cases detergents... The researcher must traverse this multidemensional search space by trial and error, with a limited quantity of protein, looking for the optimal conditions. On top of that, the conditions that confer the ideal level of nucleation may not be ideal for crystal growth...

We have developed shortcuts over the past 20 years, or so. Kits are available that allow one to screen through frequently successful crystallization conditions. The number of conditions one can test in one go is gradually increasing, as things miniturize somewhat.

The ease-of-crystallization varies amazingly from one protein to another, and tricks that improve one do not necessarily work for another, but anything that simplifies the process will be greatly appreciated by the field.

Re:crystallization is tough stuff

[BabyStewy]

While I agree that this is a major advance, I think calling this the "holy grail" is going too far. Currently there a several ways of doing large screens for crystallization conditions which utilize robots and nanoliter volumes of protein/condition to screen thousands of precipitant mixtures. The two real major stumbling blocks in crystallography are purification of monodisperse, nearly homogeneous protein (with respect to post-translational modifications as well as identity of the protein species) and the real "holy grail" problem, which is ab initio phase determination.
Since X-rays cannot be lensed, the Fourier transform of the diffraction pattern (which is the Fourier synthesis of the ordered electrons in the crystal) requires knowing not only the intensities, a trivial task, but also the relative phase angle of each reflection. This is a problem with possible solutions on the order of 6^n, where n is the total number of unique reflections, or about 5-10,000 for an average macromolecular structure at 3 angstrom Bragg spacing, allowing for a +/- accuracy of 30 degrees for each phase angle. Current solutions rely on searching reciprocal space with similar known structures (Molecular Replacement) or several ab initio methods that require one or more heavy element derivatives of native crystals. The first approach only works for crystals where a very similar structure is already known and available to the investigator. The second approach requires further screening for heavy metals that bind in ordered sites in the crystal without significant alteration of the native lattice, then usually a trip to a tunable X-ray source at a synchrotron. This second approach can burn through an astounding number of crystals and investigator time. There are shortcuts such as making the protein with Seleno-methionine instead of methionine (selenium has a usable X-ray edge for phasing, unlike sulfur),but this is normally done after initial purification and crystallization have been optimized. For more reading on the phase problem, I'd recommend either Alexander McPherson or Jan Drenth's excellent introductory textbooks on macromolecular crystallography.

Re:Ever hear of NMR structure determination?

[Bowling Moses]

The molecular weight limit of NMR has been increasing quite a bit and now proteins on the order of 100 kDa are possible, although technically challenging. Lewis Kay's group at the University of Toronto has done a solution structure of an 80 kDa protein, for instance.

I don't know a ton about nuclear medicine.

[Anon.Pedant]

Quibble #1: This is not "nuclear medicine", it is "structural biochemistry."
The field of nuclear medicine is concerned with things like radiation therapy and PET scanning.

          http://science.howstuffworks.com/nuclear-medicine. htm
          http://jnm.snmjournals.org/
          http://www.biomedcentral.com/bmcnuclmed/

Quibble #2: Your second link is very outdated. Structures for several prion proteins were determined several years ago, using both X-ray diffraction and NMR methods. Science moves on, but many webpages are never updated.

How hard can it be?

[AtomicBomb]

I observe this phenomenon all the time on top of my uncleaned plates in the kitchen sink :)

Re:An Even Easier Way to Convert Proteins to Cryst

[Mignon]

Or, if you want to convert crystals to protein, buy a girl some diamonds and see what she does to thank you...

Probably gonna start a flamewar, but...

[cr0sh]

If proteins can be crystalized (albeit with some difficulty), in order to study structure and makeup, then is the reverse true? That is, could some proteins be created from a crystal matrix?

I know this isn't a new idea. I don't have references handy to prove it isn't, I just know that I have read arguments about it. This theory is used to explain the origins of life (distinct from the theory of evolution). Basically, you have the whole "early earth molecular soup mix with electrical activity providing the spark-o-life" (Miller-Urey Experiment), forming organic compounds, which are then (in some manner) "processed" by crystal structures forming later (?).

It makes me wonder if it wouldn't be possible to study crystals in a similar manner to see whether they could (in some manner) aid the formation of the organic compounds formed by the Miller-Urey (and other similar) experiments into early proteins or protein-like structures? Does anyone know if such a study has already been undertaken? Or, is this idea nothing more than baseless speculation with no foundation in reality? I am sincerely curious...

Re:This is Not so Big

[poincaraux]

Crystalization is difficult and, frankly, rather unscientific
You're not kidding. My favorite example is the fact that many crystallographers add diet coke to aid in crystallization.

Re:Sounds Like A Lot Of People Here Are Really Sma

[SyvanX]

I've thought about this plenty, but to have some benchmarks for the shapes different proteins make, we need lots of evidence that a folding model matches the true output of the mRNA. So yeah, I look forward to the day when we can truly predict how they fold, but it's going to take a lot of work to determine the ways that different amino acids interact and how the conformational shape changes due to different interactions throughout the process of elongation. Then if that wasn't enough, we'll need to also account for the different molecular chaperones that assist some proteins in finding a functional conformational structure. I'd like to see some of the Bioinformatics guys get into this type of research (if they're not already), I'm sure there's a brilliant Biochemist out there just waiting to team up with a programmer to create a simulation that can accurately fold any sequence of amino acids.