Erica Ollmann Saphire
One of the two explicit goals of the Burroughs Wellcome Fund is “to advance fields in the basic biomedical sciences that are undervalued or in need of particular encouragement.” In other words, BWF seeks to help fill the funding gap described by the American Academy of Arts and Sciences, by supporting high-risk, high-reward research.
Case in point: Dr. Erica Ollmann Saphire, Assistant Professor of Immunology at The Scripps Research Institute in La Jolla, California. In 2003, Dr. Ollmann Saphire received a Burroughs Wellcome Career Award in the Biomedical Sciences (see Awardee Profile, 2005). Dr. Ollmann Saphire, who is a structural biologist, has devoted the majority of her laboratory’s research since 2003 to an almost quixotic quest to solve the structure of a unique protein on the surface of the deadly Ebola virus.
With the July 2008 publication of "Structure of the Ebola virus glycoprotein bound to an antibody from a human survivor" in Nature, Dr. Ollmann Saphire and her team have fulfilled the promise inherent in the 2003 Burroughs Wellcome Fund Award. In this instance, there can be no doubt that the Fund’s willingness to support high-risk research has produced a high reward in new knowledge. Solving the structure of this glycoprotein, the only protein on the surface of the Ebola virus, solves the mystery of how this extraordinarily simple biological material can be so extremely virulent and for the first time opens the door to potential treatments or vaccines.
Freelance science writer Ernie Hood spoke at length with Dr. Ollmann Saphire in August 2008. The following excerpts from the interview show that she is a tenacious scientist with a formidable intellect who will continue to shed important new light on Ebola.
The Beauty of Structural Biology
Erica, how did you settle on structural biology as the focus of your research?
I went to college to do biology, and I liked biochemistry. In my biochemistry degree, they make you take a biophysics class. The one I picked was X-ray crystallography, and I remember sitting in my class going, “Ohhhh…now I get it!” If you solve the crystal structure, you know the location of every single atom in the protein and how much it moves from that position. It seems like you know everything there is to know about it. You know how it folds up, you know where the kinks in its armor are…it’s just an overwhelming level of information, and information on an order that I just found immensely satisfying. It’s a beautiful thing – proteins are just utterly beautiful. They’ve just got stunning symmetry, and the solutions they’ve come up with to mediate all of their biological functions are artistically and architecturally fascinating.
How does the structure of these molecules have so much to tell us in terms of understanding their functioning?
Proteins stick together like Legos in the cell. So an enzyme that does something does it by attaching to its target and making some change. So if you see the shape of the business end of the molecule you can understand how it works, or in our proteins, if you look at the surface protein of the virus, you understand. The surface of a virus has all of these different jobs. It has to hide itself from the immune system, it has to attach to a new host cell, it has to somehow figure out how to force itself into that cell. It has all of these different pieces of machinery that do all of those functions that are all tied up together into one beautifully symmetric little package. And if you look at the shape of how it’s done, you can see that.
The piece of machinery that drives the [Ebola] virus into the new host cell is in a tensed, unstable shape, and what it really wants to do is rearrange itself and spring into a much more stable shape when it spears into the human cell. You can see how it’s wrapped around something else, like a spear fishing line ready to uncoil. You can actually see how that happens, and you can see the places where the virus has attached sugars to itself to shield itself from the immune system or make itself more stable as it’s out in the environment looking for a new host.
The Keys to Unlocking Ebola’s Structure
What led you to examine the Ebola virus in particular?
As a structural biologist, you learn on your mother’s knee that one DNA sequence makes one protein sequence which makes one protein structure. It’s a very linear one-to-one-to-one. Well for Ebola virus that’s not really true – that whole “one sequence makes one fold” concept is blown apart. The virus makes two different proteins using the same sequence. They share the sequence, but they fold up into two different folds, and then it’s the structure of each fold that mediates its function. And so the problem is that the virus makes mostly this one protein, which is dumped by the bucketload into an infected person’s sera, and it makes just a little bit of this other protein, which it puts on the surface. All your immune system knows is that there’s a lot of foreign protein around. So it’s mostly making its antibodies against the shed protein and not so much against the protein on the virus surface. The shed protein is sort of harmless, so if you’re directing your immune response against this decoy, it’s not really very helpful, and in terms of figuring out how to make a vaccine or a good antibody directed against the virus, we have to understand what the rare protein is on the virus surface, and we have to understand what its structure is. We know what its sequence is, but the sequence is the same as the secreted protein.
So the virus really had these structural problems where in order for the biology to proceed, we really had to have the images of what we’re working with…And that’s what I do.
The structure of the rare protein ended up being the holy grail when it comes to Ebola, right?
That’s the so-called spike protein, which the virus uses to puncture its way into the cell?
So understanding the structure of the protein could reveal the basic mechanism of the disease itself.
Exactly. And we have learned so much from this one structure. We’ve learned how it covers itself with sugar, which plays a couple of roles. The sugar hides itself from the immune system when it buds out of a human cell. It picks up human sugar, so it becomes a wolf in sheep’s clothing. So we see how it hides itself. We also see how it survives months in bat guano in caves until it finds a new host. All these sugars are hydroscopic; they attract water to keep the virus moist and happy, and we can see the structures that it’s made to keep itself alive. We also see the holes in the sugar, the chinks in the armor, the places where an antibody could bind that aren’t already all covered by sugars, so now we know what to target.
Also, it was known that this protein gets clipped at some point in infection, and then that’s really what triggers the infection. But it wasn’t really known where or how or why or what the effect of that is. And we showed in our structure the likely site where we think it gets clipped, and then that was confirmed later by biochemistry. And we showed that the sequence makes this one little tiny cut at this one little tiny loop that effectively lifts off this whole cloud of sugar or carbohydrate which then reveals or unsheathes the pieces of the virus that attach to the human receptors. And then we saw how the virus has all of those pieces of machinery, some of which are responsible for attaching to a new host, and some of which are responsible for driving the virus into the new host, and we see in the structure how they’re intertwined around each other, and how the attachment machinery actually clamps the fusion machinery in place until the trigger is released and it can spring forward. We can see the before-infection structure and the after-infection structure and make the movies of how this protein metamorphoses during infection. So there’s all kinds of biology that we now see that we understand and we know how to go and attack in the lab based on having had a good image.
It took you and your colleagues five years to successfully accomplish this – why did it take so long to solve the structure of this particular protein?
Protein is very difficult to work with. In order to solve the structure we have to grow crystals. We image things with X-rays, but a single protein molecule scatters the X-rays too weakly to see. So you need hundreds of thousands of them all oriented in an identical array in order to get sufficient scatter that you can measure it. And the way you get that scatter is the crystal. In order to get things to crystallize, they all have to be identical and they all have to be stable and they have to attach to each other stably in a three-dimensional lattice. These glycoproteins don’t want to do that. Their whole job in life is to be unstable, so that they will spring into a more stable structure later. So you have to do a lot of coaxing and engineering to keep the virus in its meta-stable state before it goes springing off into the one you don’t want to look at. The virus is also covered in all of that sugar, and the sugars are all heterogeneous – different sugars attaching every place, and at the same site you have different sugars, and the sugars don’t make stable contact, so you have to do a lot of engineering to get the crystals of this thing. And when you finally do get the crystals, because it’s so changeable and fluid, the crystals don’t diffract very well. So instead of growing and screening 25 crystals to solve your structure, we had to grow 50,000 in order to find one that would work.
With it taking so long and being so technically challenging, did you ever doubt that it could be done at all?
(long pause) I thought that eventually it would be done. I worried that it would take long enough that it would really burn out several post-docs without giving them good results. But I thought eventually we’d be able to figure out how to do it.
So you were never tempted to actually abandon this line of research…
Oh no, never. I was never going to abandon this. The biology couldn’t proceed until this worked. It had to work.
What are the potential therapeutic implications of this discovery? Does this also have implications for possible treatments for the other viral hemorrhagic fever diseases?
It would have implications for the one which is shaped like Ebola virus, which is called Marburg virus. The other viruses are going to have a very different surface structure, and they need their own structures to be determined; we just can’t predict what they’re going to look like.
How this is useful in developing treatments is that, you can vaccinate animals with DNA encoding the glycoprotein, and that seems to give a useful immune response, in a month. What we need is something that gives you some immediate protection. So let’s say the virus has been released, there’s an outbreak, or you’re a lab worker and you’ve stuck yourself with a needle; in any kind of emergency situation we’d like to be able to give the person this purified antibody to slow the virus infection long enough for a vaccine to kick in or for the person’s immune response to kick in, and the only way to confer any kind of immediate immunity is to deliver antibody. So now we know where the sites are where one might deliver antibody, and if we want to come up with a cocktail of a couple of different antibodies, we know which sites those need to be and how they might synergistically help each other. We can also see all the different sites where a drug that blocks viral infection or replication might bind, so now we have the structural templates that we need to try to design molecules to bind into those sites.
The next step would be further research to find antibodies or to design small molecules?
Yes, but I think the most direct and useful application of the structure is in explaining what the next few years of biological experiments ought to be, to give an image to work with. Assisting the laboratory biology understanding of Ebola will be more significant than the designing of drugs.
I’m sure solving the structure of the spike protein is still a major step forward when it comes to Ebola.
Absolutely. It’s the only thing on the virus’ surface. It’s the only piece of virus responsible for finding a host. It’s the only piece of virus that can get into the host. It’s the only piece of virus that you can attack from an antibody perspective in prevention of infection. So it’s really a very important thing. And that’s why it had to be this protein.
Seven genes coding for eight proteins…what makes this simple little virus so incredibly virulent, and why has it been so seemingly invulnerable to therapeutic intervention?
I would say there are three roles. One is it shuts down your immune system. Two, it replicates quickly. It makes so many copies of itself it overwhelms whatever defenses you can come up with against it. And three, characteristics of the viral surface mean that it can infect just about every cell type you have, save lymphocytes. So it’s not just destroying one type of tissue, it’s able to infect and destroy all of them.
And I understand that it’s this very speed and lethality that make it unlikely Ebola will ever break out into a widespread pandemic.
Right. It tends to burn itself out quickly. It has been somewhat contained in that it’s only broken out in Africa, and so the worry is that if it broke out in a very large metropolitan area like New York City, it might spread more quickly. But unlike the common cold, where you can walk around for two weeks carrying the cold virus and dropping it on every doorknob or stair rail you see, with the Ebola virus you’re so very, very sick that you’re not going to spread it to that many people – and that’s a fortunate thing.
I understand that’s one of the limitations of Ebola’s or Marburg’s utility as a bioweapon.
Yes, but if you’re a terrorist, one of your goals is to incite fear or panic. If they dropped the common cold, ten million people would be irritated and inconvenienced. But if we had Ebola in LA…
Still Learning in the Old School
Erica, let’s talk about X-ray crystallography. I know that’s your passion, and it’s a technology that’s been around for nearly a century. How has the technology evolved, and has it improved drastically in recent years, as so many others have?
It’s funny. For Christmas a few years ago, my father gave me a 1945 book on X-ray structure determination, and the fundamental aspects in the experiment for how we determine the structure really haven’t changed that much. What has changed is that the source of the X-rays is much more brilliant, so we can work with weakly scattering crystals that would have been utterly hopeless before. The other thing that’s changed is genetic engineering, so now we can recombinantly make any version of the protein, so we can encourage something to crystallize. Before if I wanted to work on Ebola virus structure, I would have to grow up a bunch of Ebola virus and cut the glycoprotein off. That was how the first influenza surface protein structure was determined; they had to grow up a whole lot of influenza and cut it off. You can’t do that for Ebola. You can’t make 200 gallons of Ebola and cut the surface off and then work with it. So genetic engineering lets us grow crystals of things we couldn’t grow crystals of before.
And then we have much faster computers, so we can solve much bigger structures and do things much more quickly. We have software that can index and categorize several hundred thousand spots in a couple of hours. We don’t have to sit there and do each one by hand over several years.
But the principle behind it is still old school.
You’ve been able to take advantage of some of the high-throughput technologies?
Yes, we use robotics to screen crystallizability, because we have to make so many different versions of a protein to find one that’s going to crystallize, it’s helpful if we can triage different versions of it on a small scale using a robot, instead of burning out a whole army of graduate students.