Jack Reynolds wrote:
> My name is Jack Reynolds and I am economics major. I graduate in August and begin an MBA program shortly thereafter. I am exploring a business idea that involves protein crystal growth and x-ray diffraction in the microgravity environment of space.
>
> I hope the group can indulge me a few questions. If this group is not the proper forum for this, please feel free to let me know and I will go elsewhere. Thanks.
>    
Don't worry.  This is definitely not the most off-topic post I have seen 
on this list!
> 1. Have there been some new techniques developed to grow crystals in terrestrial labs that may produce crystals comparable in quality to those grown in space. Despite these advances, is there still a role for protein growth in an actual microg environment?
>    
Microfluidic systems show some promise on this "front" because 
sufficiently small volumes also drastically limit convection.  Myself 
and others are actively working on in-situ diffraction from microfluidic 
trays, and we found that we could get mosaic spreads that were ... well, 
too small to measure.  Definitely less than 0.01 degrees, and probably 
much less, but our x-ray beam divergence dominated the rocking width in 
those experiments.

No, I do not get kick-backs from Fluidigm, but I did play a role in 
developing the product described here:
http://www.fluidigm.com/pdf/topaz/FLDM_TOPAZ_MRKT00114.pdf
   For any diffraction experiment it is really important to to reduce 
the amount of background x-ray scattering (non-crystal solids and 
heavier-than-carbon atoms in the x-ray beam) to less than or equal to 
that of the crystal itself, and this is probably why in-situ results to 
date are not usually as good a results from T-Y Teng's cryo-loop 
mounting technique.  Most trays are far thicker than crystals!  But, the 
new Fluidigm trays have the same x-ray background as a loop mount, and 
so I think this is actually a major step forward.

It is true that cryo-cooling generally ruins the advantages of 
micro-gravity growth, but a more gentle cryo-cooling procedure was 
developed recently (Warkentin and Thorne, J. Appl. Cryst. 2009) that 
demonstrated much better preservation of mosaic spread.  Also, radiation 
damage rates at room temperature may not be nearly as bad as previously 
thought if the dose rate is chosen properly (Southworth-Davies et al. 
Structure 2007 and also the recent RD6 workshop proceedings).  Combine 
this with other microgravity advantages: larger crystal volumes, lower 
Wilson B factors, and smaller diffracted beam spots (Holton and Frankel 
Acta D, 2010), and you could potentially make room-temperature 
diffraction competitive with cryo-cooled methods again.



Another "alternative" technology you seem to have also found is the 
"Bitter solenoid" idea.  I thought this had potential ever since I first 
saw the levitating frog:
http://www.ru.nl/hfml/research/levitation/diamagnetic/
which seems to have also made it on to YouTube.
http://www.youtube.com/watch?v=A1vyB-O5i6E

I suppose here the problem is the cost of building a 96-well 
crystallization "tray" made of these solenoids, and the energy needed to 
run it.  Then again, this may not be too bad when compared to the energy 
required to lift the tray into low-earth orbit.

Crystallographers are notoriously penny-wise and pound-foolish.  We will 
balk at spending more than $5/ea for a 96-well crystallization tray, but 
gladly blow $1e5 or more on a "device" that has only been shown to 
improve diffraction once in a blue moon.  I'm not going to name names, 
but I will say that the latter becomes far far more attractive if the 
one case where the device "worked" produced a Science, Nature or Cell 
paper.  This is not because we are mindless bandwagon-chasers, but 
rather because a "big splash" is seldom made by doing an easy experiment.

As a beamline scientist I can tell you that there is definitely a 
"market" in improving diffraction.  Incredible amounts of time and 
effort are now typically spent trying to get a given crystal system to 
yield spots out beyond the water ring (3.5 A).  Typical investments are 
months to years for a single structure determination.  Particularly with 
large complexes and membrane proteins.  Nevertheless, the general 
approach is to exhaust all cheap options first (such as graduate 
student's time), before resorting to more exotic ideas.


My perception of the main problem with the "space crystals" was that 
there is nothing more infuriating to a crystallographer than seeing what 
looks like huge amounts of money being spent on a protein that you are 
not particularly interested in.  Unfortunately, when you work with NASA 
you are not allowed to fly experiments that have a significant 
probability of "failure", and since the only protein that is pretty much 
guaranteed to crystallize is lysozyme, that's what the "space crystals" 
had to be.  The resulting research has been much maligned because of 
this, but if you read the papers carefully you will find that many of 
these workers managed to learn a great deal about the basic chemistry 
and physics of macromolecular crystal growth that was not known before.  
This is not an easy thing to study, and there are a very small number of 
people who do, despite how incredibly important it is.

I think the current literature is convincing that microgravity can have 
a positive impact on diffraction in some cases, but just like everything 
else in crystallography there are plenty of cases where it won't help.  
The only way to know is to try, and you have to do something to convince 
crystallographers that trying out your new method is worth it.  This 
generally involves demonstrating a "success" with a "hard" problem (the 
kind that gets into a big-name Journal), or gathering statistics from a 
very large number of different proteins.  You can also arrive at a 
fundamental understanding of the process and do your convincing 
scientifically, but all of these routes are expensive.  Such is the 
plight of methods development.

As a "business model", what I would recommend is some kind of "return 
policy" for when the experiment "fails".  As a service provider, you 
will be facing a very high likelihood of failure, and a customer who 
will never be entirely convinced that you "tried hard enough".  So the 
motto: "Improved diffraction, or your money back!" is an attractive 
one.  Or maybe "Improved diffraction, or (most of) your money back!".

> 2. Has the process of growing crystals been automated? Is there any step in the process of growing crystals that has not or cannot be automated?
>    
Yes and no.  The process itself has certainly been automated, not once 
but many times.  But none of these automation systems have a 100% 
success rate.  Not even close.  There are both stochastic and systematic 
components to the success rate of macromolecular crystal growth.  For 
example, many proteins simply don't crystallize, no matter what you do 
(a systematic failure).  Most that do crystallize are EXTREMELY finicky 
(sensitive to both controlled and uncontrolled variables).  For example, 
it is not uncommon for an identical chemical mixture to produce crystals 
in one kind of tray, but not in another.  In fact, even identical 
conditions in the same tray will not always produce crystals every time 
(uncontrolled variables).  This is generally regarded as being due to 
the stochastic (random) nature of nucleation, and seeding can help a 
great deal in such cases.  Seeding is generally done by hand, but at 
least one group has an automation system for it (Newman et al. Acta F 
2008, Newman et al. J. Biomol. Screen. 2009).

Nevertheless, once you ave studied a particular crystallization system 
enough to understand its quirks, it can become very "routine".  That is, 
everything becomes easy once you know how to do it.  My favorite example 
of this is thaumatin.  Crystals of this protein were incredibly fragile 
and difficult to work with for more than ten years until the discovery 
of the tartarate-dependent crystal form (Alex McPherson, personal 
communication).  Now thaumatin crystals are generally regarded as being 
"too easy" and it is now difficult to convince people that things you 
learn by studying thaumatin crystallization are relevant to "hard" 
experiments.


  Possibly the weakest link in crystallization automation is at the end: 
harvesting.  There are a few systems out there for automated harvesting 
of crystals (Viola et al. J. Struc Func. Genom. 2007), but right now 
almost everyone still does this by hand.  The challenges are mainly in 
object recognition.

> 3. Has the process of x-ray diffraction been automated? Is there any step in the process of x-ray diffraction that has not or cannot be automated?
>    
Again, everything has been "automated", but nothing is 100% effective.  
For diffraction, much effort has been spent eliminating stochastic 
failure modes (robots dropping crystals on the floor), but that usually 
comes at the expense of introducing systematic ones (narrow range of 
supported methodologies).

I think the weakest link here is getting the crystal centered in the 
x-ray beam.  A tremendous amount of effort has been expended on this 
problem, but, personally, I don't think it will ever be 100% effective.  
I formed this opinion watching hundreds of highly intelligent human 
beings centering crystals that they had grown, harvested and mounted 
themselves, and there are still plenty of cases where they can't figure 
out where their crystals are in the loop.  Using polarized light or UV 
illumination to light up tryptophan can help, but not every protein 
crystal is birefringent and not every protein contains tryptophan.

There are X-ray based centering approaches, such as "peppering" the loop 
with x-ray shots (available at most beamlines), phase-contrast x-ray 
tomography (Brockhauser et al. J. Appl. Cryst. 2008), or just looking at 
the shadow of the sample on the x-ray detector (something I am working 
on).  All of these are currently time consuming (much more time 
consuming that clicking on a video image of the loop), but all show 
promise of becoming faster.  You can also just illuminate the whole drop 
with a broad x-ray beam, but then the background scattering problems I 
mentioned above must be cubed (xtal volume vs illuminated volume).

Anyway, I certainly cannot summarize the whole field in one email, but 
most of the people in the field are on this BB and I'm sure they will 
now chime in to correct or expand the statements I have made above.

For shooting crystals in space, I would recommend a well-designed 
in-situ growth/diffraction system (with thin walls) and as compact a 
light source as you can find that can still deliver ~100 Gy/s or so.  I 
think that an up-and-coming gallium jet technology (Otendal et al. Rev. 
Sci. Inst. 2008) could potentially have the most photons/kg.

-James Holton
MAD Scientist