Skip to main content
Premium Trial:

Request an Annual Quote

Lawrence DeLucas on Crystallizing Proteins, in Space and on Earth


At A Glance

Name: Lawrence DeLucas

Age: 52

Title: Director, Center for Biophysical Sciences and Engineering, University of Alabama at Birmingham, since 1994

Prior Experience: BS, MS in chemistry; BS in physiological optics, UAB

Doctor of optometry, UAB, 1981

PhD in biochemistry, UAB, 1982

NASA chief scientist for International Space Station, 1994-95

Conducted protein crystallization experiments on a 14-day mission on space shuttle Columbia in 1992.


So you just had a teleconference with NASA? Are they planning any new protein crystallization experiments?

I got off that [call] about 20 minutes ago. Yes, [there will be new experiments but] I can’t tell you when they are going to be, because we just don’t know when the shuttle is going to start flying again.

What is the advantage of growing protein crystals at reduced gravity?

When a crystal grows on Earth, lighter molecules tend to float up because of gravity, and heavier ones sink. Because of this buoyancy-induced convection, [there is] more motion, the crystal grows faster, and as a result, when a protein comes in to get attached to the crystal, it doesn’t [always] come in the perfect position. If [the crystal] grows too fast, protein is getting trapped in misalignments. In microgravity, there is no convection because there are no lighter and heavier molecules, basically. The only thing that allows that protein to move is random diffusion. That process is much slower than with buoyancy-induced convection on Earth, and that’s what we think is one of the reasons why when we do get better crystals, they are better. We don’t always get them better, because there are many factors that can affect the quality of the crystals.

Why is it important to find better ways to crystallize proteins?

Here is a very telling statistic: There are nine national centers in the US for high-throughput structural proteomics. If you look on their websites, they have been able to crystallize about a third or so of the soluble proteins they have expressed. How many of those crystals are good enough to determine the structure? It’s no more than 10 percent. I can’t think of a better justification for looking for any means that we can [use to] improve the quality of crystals, on Earth and in space.

Have people solved the structure of a protein that was crystallized in space, where the crystal on Earth was not good enough?

There are many examples where you cannot see all the features in a structure on Earth. [One example is] a complex of human insulin with a drug bound to it. On Earth, it looked like there was just one molecule in between a hexamer of insulin. The crystallographers involved in this wanted to have more detail, so we flew the protein in space and grew crystals there, and those crystals diffracted about 0.4 -1 angstrom further in resolution, and when they recalculated the complex structure, they saw that in an area where they thought there was solvent, there was actually another one of the drug molecules. It’s unlikely they ever would have realized that two drugs could fit in that pocket, had it not been for the improved quality from the space crystals.

Is there any interest from pharmaceutical companies to sponsor space crystallization?

There have been many pharmaceutical companies and also smaller biotech companies that have put proteins on the shuttle or on the space station through our program. They pay for everything related to their experiment, [and] we do expect certain data from those companies; nothing proprietary — we just want to see a comparison of the space data with the best dataset that they have ever collected on Earth by any method.

How about academic researchers, like yourself?

It’s the typical thing, you get a grant from, in this case, NASA, and that grant is to investigate maybe some new way of growing crystals. In the past, NASA funded the fundamental understanding of the physics of protein crystallization. There was a review of NASA’s program about two years ago. The reviewers said, ‘What is the contribution that this program has made to structural biology?’ [For example] a molecular complex, because those are quite difficult to get good crystals on the ground, [or] a protein with a drug bound in the active site, or membrane proteins. Since that review, NASA has said they will start funding people because the biology is exciting.

Do you think space might help crystallize membrane proteins?

I will just be interested to see what happens; I don’t know. I am always worried about membrane proteins benefiting from space, because I think the reason they generally don’t crystallize to high resolution is, they often have lipid that’s randomly associated on the surface of the membrane protein, and I don’t see how space is going to get rid of that randomness. But you know, every time I say something in science that probably won’t work, that’s the one that works. So I think it’s probably worthwhile to pursue some of these experiments.

Tell me what else interests you in your research.

The truth is, probably 90 percent of my career has been spent developing technology for applications right here on Earth. Our nanocrystallization system is one example. It’s a high-throughput robot that we developed with a spin-off company from NASA called Analyza that can set up protein crystallization experiments in droplets that are 15 nanoliters in volume. It basically allows you to save the amount of protein you have to use. Syrrx said [it has] a patent on it, and they put in a lawsuit blocking our company from doing that [in the United States]. But as a researcher at a university, I am allowed to use this technology.

Another thing you have to do if you are setting these experiments up in a very high-throughput mode is, you have to rapidly observe the drops. Our own spinoff company, called Diversified Scientific, developed a program that is commercially available that will observe what’s in a drop, how many crystals there are, how big they are, and even score the quality of the crystal.

We also developed a proprietary neural net, and we are using that in a limited screen, say, 300 conditions, to teach the neural net what factors are important in producing a crystal. Then based on that, the computer learns and can say, of all the possible combinations that you can set up, let’s say 40 million, what are the top 50 that are most likely to give good crystals? Then it’ll tell you those conditions, so that you just go set up 50 and hopefully get an improvement or really nice crystals.

How does that work?

It’s sort of like a good chess player who can think six moves ahead, where the guy that can only think ahead four or five ends up getting beaten. That’s what a neural net is good at, it’s taking data that produces a result and looking for relationships between the factors in that data that produce certain results, and then [it] uses that to predict how to get a specific result. We have developed a neural net to do that for crystallization. So far, the results that we see have been very intriguing. We have taken lactoglobulin, a commercial protein, and we screened 360 conditions. It has to be a balanced screen, a balanced subset of that 30 million [or so conditions] across all the pHs, all the ionic strengths, and so forth. We ran the whole screen on the protein, so we knew the results, and we told the neural net the results only for first 300. And then we asked the question, in the last 60, tell us where the crystals form. There were two places, and it chose those two places exactly. A neural net will not work unless in the training set, there is a place where you get crystals. You have to know what factors produce a crystal to be able to predict that. In one training set, there was one condition that gave a crystal at pH 4.5. The place where it predicted you would get a crystal [and we did] was at pH 8.5. I don’t know a crystallographer that would have made that connection. It was really astounding that it was able to do something like that. We are now doing a huge study, so that statistically, we can see how often this can be helpful, and how good it will be at predicting out of the 30 million conditions that you still haven’t screened, what would be the best one to give crystals. Again, it saves you lots of protein. A paper about this is going to come out soon in Structural Biology.

Overall, how can the bottlenecks in structural genomics be overcome?

Clearly, the biggest bottleneck that I see is the crystallization, getting high quality crystals. We know that controlling the kinetics of growth can really help improve a crystal. So I think there is going to be some attention paid to the physics of crystal growth, just what NASA was supporting in the past. I hope that some of the things we discovered through the NASA program will start being utilized here on Earth.

The last big bottleneck is going to be membrane proteins. That one does involve two things: Getting enough of it solubilized on the membrane, and crystallizing it. Crystallizing membrane proteins is done the same way [as other proteins], but you have added variables. You usually add detergent, [and] the size of the detergent, [as well as] the randomness of how it packs around the membrane protein all play critical roles in whether or not you get good crystals.

Probably one of the most exciting ways to help get a crystal, and maybe even improve the quality of that crystal, is to do site-directed mutagenesis. I think as we learn more from those examples where it has been done, that informatics data may help us predict what to change, more rapidly get the right changes to help form a crystal.

The Scan

Octopus Brain Complexity Linked to MicroRNA Expansions

Investigators saw microRNA gene expansions coinciding with complex brains when they analyzed certain cephalopod transcriptomes, as they report in Science Advances.

Study Tracks Outcomes in Children Born to Zika Virus-Infected Mothers

By following pregnancy outcomes for women with RT-PCR-confirmed Zika virus infections, researchers saw in Lancet Regional Health congenital abnormalities in roughly one-third of live-born children.

Team Presents Benchmark Study of RNA Classification Tools

With more than 135 transcriptomic datasets, researchers tested two dozen coding and non-coding RNA classification tools, establishing a set of potentially misclassified transcripts, as they report in Nucleic Acids Research.

Breast Cancer Risk Related to Pathogenic BRCA1 Mutation May Be Modified by Repeats

Several variable number tandem repeats appear to impact breast cancer risk and age at diagnosis in almost 350 individuals carrying a risky Ashkenazi Jewish BRCA1 founder mutation.