Here's Looking at You

Structure-Guided Protein Design for a Better HIV Antibody

For several years now, the lab of structural biologist Pamela Bjorkman, Max Delbrück Professor of Biology at California Institute of Technology, has been trying to find a new way to stop HIV with antibodies that prevent the virus from infecting a cell.

Normally, the body forms antibodies on its own. But in the case of HIV, which mutates rapidly and has few handholds for antibodies to grab onto, scientists have had trouble uncovering natural antibodies. Only recently have the numbers of natural neutralizing HIV antibodies been significant enough to enable researchers to compare them with one another. Bjorkman's work starts there. She uses structural biology to look at these antibodies and then she uses what she’s learned from structures to design better ones.

In a recent study, published in Science in December 2011, Bjorkman's team used X-ray crystallography to look at how a set of natural antibodies bind an HIV spike protein called gp120. The antibodies selected had shown high potency and breadth and also were between 85 and 96 percent identical in sequence. This similarity allowed the researchers to tease out which unique features of the structures make each antibody work well. "Ron Diskin, the first author, is very good at this," says Bjorkman. "He's just staring at these structures and figuring out what should be where."

Using this method, Bjorkman and her colleagues have designed a new antibody that introduces a single amino acid change to a natural HIV antibody. The changed antibody maintained its breadth and improved its potency an average of 10-fold. "When we compared the original to our mutant, there wasn't a single HIV strain that the mutant was worse against," says Bjorkman. "It was always better."

At worst, the mutant was 2 times better. At best, 2000. Such increased potency means the designer antibody could be used therapeutically at much lower concentrations. "That's an enormous practical difference in terms of whether this is realistic for therapies or prophylaxis," says Bjorkman, who is now working to include this antibody in clinical trials. In addition, she has another unpublished designer antibody with two amino acid changes that is even better than her first.

At CalTech, Bjorkman' work is supported by a facility called the "Molecular Observatory," which hosts crystallization robots and helps support the CalTech beamline at the Stanford Synchrotron Radiation Lightsource. In addition, she relies on SBGrid, something that she didn’t have back when she worked as a graduate student at Harvard Medical School, training with the late Don Wiley. "Nothing was automated then," she laughs. "We grad students spent most of our time writing conversion codes to get files from one program to another."

One data gathering technique Bjorkman is experimenting with (using SBGrid software including ATSAS) is small angle X-ray scattering, in which the protein is in solution rather than in a crystal. "You can't get an atomic resolution 3D structure," she says. "But you can learn things about shape and radius of gyration and behavior in solution."

Story: Elizabeth Dougherty Image: Micheline Pelletier / Gamma

Published April 25, 2012

From Curiosity to Cure

Viral proteins as promising new targets for cancer therapies

Marc Kvansakul decided to become a structural biologist as a young teen after watching a documentary that described proteins as assemblies of Lego-like blocks. Today Kvansakul’s newly formed lab in the department of biochemistry at La Trobe University in Victoria, Australia, is using what he has learned about the sequences and structures of anti-apoptotic viral proteins to start developing new treatments for Burkitt lymphoma, a form of the disease known to be caused by the Epstein-Barr virus.

Not exactly the same as building yellow and red plastic-brick scale models, but, says Kvansakul, “it’s all very good fun.”

Kvansakul began this line of research during post-doctoral work under structural biologist Peter Colman at the Walter & Eliza Hall Institute in Melbourne. First they studied M11L, a Myxoma pox virus protein that helps viruses survive by disabling an infected host cell’s self-destruction program. Kvansakul and Colman wanted to understand how.

The challenge was particularly interesting because M11L, amino acid sequence-wise, looks unlike anything else. But Kvansakul found that, crystal structure-wise, M11L looks a lot like a known mammalian anti-apoptotic protein called Bcl-2. “We uncovered a completely new area in sequence space that forms the same fold,” says Kvansakul, who used SHARP and PHASER to solve the structure, Coot to build the model and CNS to refine it. He published the findings in 2007 in Molecular Cell.

Now, he says, “we’re going fishing” for other viral proteins that, sequence-wise, look a little like M11L to see if they also turn out to be anti-apoptotic. So far, in his own lab at La Trobe, he has found a few hits suggesting there may be a conserved sequence pattern that builds this important fold. “The fact that we found a molecule that forms the same fold but is only 3% identical on a sequence level tells us that there surely must be loads of these proteins in lots of viruses,” says Kvansakul. Only they can’t be found with a computational search. “Once we work out the biology, we can think about how we can build a search algorithm to find other proteins,” he says.

After that 2007 study, Kvansakul and Colman completed a study of BHRF1, a viral protein in Epstein-Barr virus that blocks apoptosis. Kvansakul found that the viral protein blocks multiple pro-apoptotic proteins in host cells, and that in a mouse model of Burkitt lymphoma it also renders chemotherapy agents ineffective, making it a potent oncogene. The results, also obtained using SHELX, PHASER, Coot and REFMAC, appear in 2010 in PLoS Pathology.

Now in his lab at La Trobe, Kvansakul is working with a medicinal chemist to identify small molecules that block BHRF1’s anti-apoptotic function. “Out of me just being interested in structures has turned into, ‘Oh, why don’t we develop new treatments for Burkitt lymphoma?” he says.

While setting up his lab in 2010 and worrying about how he would manage he lab’s computing needs, Kvansakul happened upon mention of SBGrid online. “I thought, this is sent by Heaven!” he says.

Being in Australia, Kvansakul didn’t make it to SBGrid’s June 2011 Symposium personally. Instead, he invited his lab over for a late dinner and then, around 11pm, they all settled into his lounge and watched the live web feed until 7am. “During prime party hours the students were happy to listen to the speakers and take notes,” he says. “It was brilliant.”

As a graduate student and post-doc, Kvansakul used O—“praising and cursing Alwyn Jones in equal amounts”—and CNS and a few other tools. “The more I work, the more tools I learn about,” and happier he is that that he has SBGrid to manage it all.

As an example, at a conference this summer, he heard about a new technique recently implemented in CNS called deformable elastic network refinement (DEN). “I got home and a couple of weeks later, there was an upgrade from SBGrid and we could run it,” he said. “This is exactly why I signed up with these guys. There are probably lots of things out there I haven’t heard of yet, and hopefully I’ll start running into them too.” —Elizabeth Dougherty

Published December 13, 2011

Crystallography for Kids

Finding new ways to help patients with cystic fibrosis

Most people, armed with tartar-control toothpaste and a miniature scrub-brush, do battle with biofilms every morning. Biofilms form when bacteria attach to a surface, like a tooth, and form a colony encased in a protective coating.

SBGrid member Lynne Howell, senior scientist at The Hospital for Sick Children (SickKids) and a biochemistry professor at the University of Toronto, studies biofilms formed by Pseudomona aeruginosa, a bacterium that afflicts Cystic Fibrosis (CF) patients by forming biofilms inside their lungs. She recently uncovered the structure of AlgK, an outer-membrane lipoprotein on P. aeruginosa that helps the bacteria form a biofilm.

For CF patients, the genetic mutation that causes the disease also causes a chloride imbalance in the lungs. The altered lung environment mutates the bacteria and encourages a change from the typical, free-floating form, in which they are susceptible to antibiotics, into colonies encased inside a biofilm that protects them from antibiotics.

These biofilms not only shield the bacteria from drugs, they also help them evade the human immune system. “A lot of the damage to the lungs in cystic fibrosis patients is caused by the immune response that is trying, but failing, to get rid of the bacteria,” says Howell.

Recently Howell's lab, with the help of SBGrid programs Phenix and CCP4, among others, solved the structure of AlgK, a protein required for P. aeruginosa to produce alginate, the polysaccharide that forms the bacterial biofilm inside the lung. Her work, published in the Feb 10 2010 Structure, shows that AlgK helps situate the porin AlgE on the outer-membrane pore, allowing it to export alginate outside of the cell.

Howell's lab has followed up this work by solving the structure of the export porin AlgE, which was published in the July 21 2011 issue of the Proceedings of the National Academy of Sciences. Together the structures of AlgK and AlgE suggest a novel mechanism for polysaccharide export across the outer membrane. By understanding the structures, and chemical and physical properties of the proteins that export the polysaccharides that form the biofilms, she hopes to uncover new ways to battle these lethal biofilm infections.

In parallel work, Howell is investigating an organelle, the type IV pilus, that anchors the bacterium surface in the first place. “If we can understand how this organelle is assembled, we may be able to prevent it from assembling and allowing surface attachment,” says Howell. Without surface attachment, the bacteria lose their virulence.

Regardless, the end game remains the same. “We're trying to improve the lives of kids with this disease,” says Howell.

– Elizabeth Dougherty

Published May 17, 2011

Let Science Be Your Guide

Ning Zheng's drug discovery ideas come from human pathology and plant biology and ...

When Ning Zheng got side-tracked from his studies of protein degradation, he never expected to end up in the plant world. Today, Zheng, associate professor of pharmacology at the University of Washington and an HHMI investigator, runs a triplicate of research agendas, all rooted in Xray Crystallography, and all aiming to find new therapeutic drugs for human diseases.

Zheng started his career solving large protein-protein complexes of ubiquitin ligases and the proteins they bind with to degrade them. Malfunctions in this process of ubiquitination are involved in several diseases including cancers, neurological disorders and viral infections. “A lot of companies are trying to find small molecules to manipulate ubiquitin ligases, but nobody has succeeded,” says Zheng.

In 2005, Zheng got side-tracked. A group in Indiana discovered that a plant hormone called auxin, which regulates plant growth, binds directly with ubiquitin ligase to regulate protein degradation. One year later, Zheng solved the structure of this complex.

By doing so, he discovered a surprising phenomenon. He found that the plant ubiquitin ligase and its target protein have an imperfect interface. Auxin, says Zheng, acts like “molecular glue” by filling in the gaps in this weak binding and allowing the proteins to bind with high affinity.

Zheng is now trying to translate this finding to human pathology, where mutations in human ubiquitin ligases cause low affinity bindings to target proteins. “The idea of using a small molecule, like auxin, to rescue a protein-protein interaction is new in the pharmaceutical industry,” says Zheng.

Once Zheng added the study of small molecules to his repertoire, his research took yet another side road. “All of the sudden science led us to study surface membrane proteins, including hormone receptors and ion channels,” says Zheng. “Most drug targets are on the membrane.”

Zheng, being a membrane protein rookie, had what he calls a stroke of beginners luck when he and his team solved the structure for a voltage-gated sodium channel, a channel others in the field had long been working to solve. “We took a new approach, with crazy ideas,” says Zheng. “SBGrid also helped us. Their software is now available to us and we used it for analyzing these membrane proteins.”

This work, published in the journal Nature, may help drive the discovery of new drugs such as local anesthetics and antiarrhythmics, which directly target this ion channel. Meanwhile, Zheng will let science be his guide.

– Elizabeth Dougherty

Published May 13, 2011