Flu-drug flap

A scientific spat over how drugs affect the structure of an influenza channel could be imperiling the development of new drugs targeting seasonal and pandemic flu viruses such as the H1N1 swine flu.

DeGrado's M2, complexed with amantadine

In the fall of 2007, the scientific advisory board of Influmedix, a Radnor, PA-based biotech company aimed at developing novel anti-flu medications, convened a conference call to discuss an important flu protein, a proton channel called M2. Two different researchers associated with the company had just cracked the proton channel's atomic-scale structure, but the two then-unpublished structural models came to strikingly different conclusions about how two flu drugs targeted this channel.

Influmedix founder William DeGrado of the University of Pennsylvania in Philadelphia, thought that the drugs amantadine and rimantadine prevented the channel from opening directly. Alternatively, scientific advisor James Chou of Harvard Medical School in Boston, Massachusetts, had evidence that the drugs bound at a site outside the pore that modulated the protein by changing its shape. The two were at an impasse.

Chou's M2, complexed with rimantadine

Since most influenza viruses, including the H1N1 pandemic swine flu strain, are now resistant to amantadine, rimantadine, and other drugs that target M2 -- one of only two proven anti-flu drug targets -- the hope is that researchers can use a structure-based approach to find new, critically needed compounds that target the same proton channel and make once-resistant flu strains vulnerable again. But that's contingent on settling the academic altercation, noted Lawrence Pinto of Northwestern University in Evanston, Illinois, who collaborates with DeGrado and advises Influmedix. "There has to be an end to it."

DeGrado and Chou published their structures side-by-side months later, in January 2008, each claiming that their own model explained the drugs' mode of action. The two men haven't spoken since -- although they have exchanged occasional emails -- and Chou resigned from Influmedix's scientific advisory board a few months later.

The dispute is now firmly stuck in the scientific literature as each researcher continues to publish follow-up papers that support his viewpoint. "Both sides have become increasingly rigid," said Christopher Miller, an ion channel researcher at Brandeis University in Waltham, Massachusetts, who reviewed the original two papers and wrote an accompanying commentary.

When Chou started his Harvard lab five years ago, he decided to apply his expertise in nuclear magnetic resonance (NMR) imaging to solve the structure of membrane-gated ion channels, which no one had ever done before using this particular technique. He turned his attention to M2, a small channel that affects how the influenza virus replicates.

Like most people, Chou initially figured that the drugs that targeted this channel worked like corks in a bottle -- stick something in the middle of the opening and nothing can get through. "That's just intuition," he said. "It's common sense." But something didn't sit right about that idea. M2 blockers are tiny drugs, so how could several disparate mutations inside the larger pore all confer drug resistance if the compounds adhere to one particular spot, Chou wondered.

After a four-year effort, Chou and his postdoc Jason Schnell, now at the University of Oxford, UK, eventually concluded that the drugs didn't block the pore directly. Rather, their NMR imaging showed that they bound on the outside of M2 and caused the channel to lock in a closed state. "We unambiguously found that the drug interacts at that binding site," said Chou.

DeGrado's results suggested otherwise. His postdoc Amanda Stouffer, now at the Swiss Federal Institute of Technology (ETH) in Zurich, spent a year crystallizing the transmembrane region of the protein using X-ray crystallography, and concluded that the drugs nestled right into the pocket of the channel's pore. "It's quite clear that we know where the drug binds," said DeGrado.

Both can't be right. DeGrado doesn't doubt Chou's data, but suspects that Chou is observing non-specific drug binding to a non-pharmacologically relevant part of the protein. Chou's NMR structure was created with a "massive amount" of drug compounds, DeGrado said, so the drugs might just be getting stuck on the outside of the channel. The channel's locked position is likely caused by a drug bound inside the pocket that Chou missed, DeGrado reasoned.

Chou stands by his results and counters that what DeGrado saw in the middle of the channel was not the drug at all. At DeGrado's crystal resolution of 3.5 Angstroms, there's no way to say definitively that the pocket wasn't filled with leftover reagents from the crystallization protocol, rather than the drug, Chou argued.

Robert Lamb, a Northwestern University virologist and Influmedix scientific founder, concedes that the resolution would need to be better in both experiments to definitively prove either hypothesis. But he points to mounting evidence arguing in favor of the pore-blocking (DeGrado's) model.

Last year, Lamb and Pinto mutated the sites that Chou had proposed were important for drug binding outside the pore and measured the electrophysiological properties of the channel in cell models and in live viruses. The channel was still sensitive to the drug, arguing against Chou's model. Chou counters that he analyzed the same mutations using a different approach and found that the amino acid changes indeed altered the drug sensitivity.

The only other person actively publishing papers supporting Chou's data is his father, Kuo-Chen Chou, the founder and president of the Gordon Life Science Institute, a non-profit research organization in San Diego, California. In July, Kuo-Chen Chou and his colleagues in China published two papers based on computational analyses that support James Chou's model of the M2 structure.

James Chou stressed that he was not involved in his father's research, and that he doesn't trust conclusions based solely on computational modeling. "I want to make it clear to you that I had nothing to do with those papers," said James Chou.

DeGrado and his colleagues want to wash their hands of the whole controversy. "I'm not saying there might not be another binding site, but the question is, 'Is there a pharmacologically relevant one?' And the answer is clearly no," DeGrado said.


Top image credit: PDB ID: 3C9J/ A.L. Stouffer, R. Acharya, D. Salom, A.S. Levine, L.D. Costanzo, C.S. Soto, V. Tereshko, V. Nanda, S. Stayrook, W.F. DeGrado, Structural basis for the function and inhibition of an influenza virus proton channel, Nature 451: 596-600 (2008)
Bottom image credit: PDB ID: 2RLF/ J.R. Schnell, J.J. Chou, Structure and mechanism of the M2 proton channel of influenza A virus, Nature 451: 591-595 (2008)


Related stories:

Swine flu windfall
[23rd September 2009]

Avian flu promotes Parkinson's?
[10th August 2009]

Can biotech tackle swine flu?
[27th April 2009]