Charges Fly in Channel Field

Sometimes a picture can be deceiving--or at least that's what some biophysicists are saying about their first structural glimpse of a voltage-dependent potassium channel. This membrane pore ushers K+ ions out of cells during nerve impulses and muscle contraction. The controversy that's been stirred up since Roderick MacKinnon of Rockefeller University unveiled a structure of one such channel last spring was front and center at last month's 48th annual Biophysical Society meeting in Baltimore.

Human function depends on the rapidly traveling electrical signals that course through the nervous system. These electrical signals are driven by an influx of sodium ions into nerve cells, producing rapidly propagated changes in the balance of charge inside and outside the nerve cells' membranes. Voltage-gated K⁺ channels sense such changes in charge distribution and open their gates, allowing potassium ions to flow out of the nerve cell. In doing so, these channels allow the nerve cell to return to its resting state and ready itself for the next impulse.

The voltages that these channels sense are tiny compared with, say, the battery in a car. But spread across a lipid membrane only about 30­40 Å thick, this tiny voltage creates a huge electric field. Voltage-gated K⁺ channels harness this enormous electric field to open and close their ion-conducting pore.

Just how these channels achieve this feat has been the subject of years of intense research. It must involve the movement of electrically charged groups within the protein across the membrane. These charges were eventually traced to a handful of conserved, positively charged arginine residues within a mostly hydrophobic -helical region. But no one had a clue as to what this so-called voltage sensor might look like.

Then, last spring, after years of struggling to crystallize a voltage-gated K⁺ channel, MacKinnon, postdoc Youxing Jiang, and their coworkers finally succeeded [Nature, 423, 33 (2003)]. The tetrameric voltage-dependent channel's K⁺ conducting pore looked just like that of the voltage-independent K⁺ channel KscA, a structure that MacKinnon's lab had solved a few years earlier. But the rest of the picture was not what anyone--including MacKinnon himself--had imagined.

Everyone had expected that the arginine-studded voltage sensor would be packed snugly among the other non-pore-forming helices, well protected from the nonpolar lipid bilayer. But in MacKinnon's structure, the paddle-shaped arginine-studded voltage sensor occupies a seemingly hostile place at the channel's edge, at the protein-membrane interface.

The shocking structure--and the ensuing flurry of experiments to prove it right or wrong--permeated last month's Biophysical Society meeting, at which MacKinnon, a Howard Hughes Medical Institute investigator and newly minted Nobel Laureate in chemistry, gave the keynote address.

AT A SYMPOSIUM on the subject, Eduardo A. Perozo, an associate professor of molecular physiology and biological physics at the University of Virginia, recalled his own mix of fascination and disbelief at seeing the structure for the first time. "When we first saw the structure," he said, "many of us asked, 'Is it for real?' "

Some biophysicists, suspicious of the tricks MacKinnon's team was forced to use to crystallize the membrane protein, are convinced it's not. They wonder whether the channel that MacKinnon's team crystallized--a heat-stable version from an archaebacterium found in deep-sea thermal vents--bears any resemblance to better studied human voltage-gated K⁺ channels. And they question whether the antibody fragments MacKinnon's team used to stabilize the floppy voltage sensors have distorted them beyond recognition.

To test whether the structure exists as described, Jiang, MacKinnon, and their colleagues also probed to what extent the voltage sensor paddle might move during channel opening. They relied on antibody-free experiments using the membrane-bound channel and reported the results in an accompanying paper [Nature, 423, 42 (2003)]. They found that biotin tethered to various spots on the voltage sensor can be dragged from the inside to the outside of the membrane when a voltage change triggers channel opening. The bulky, tethered biotin molecule can't be dragged through the protein environment. Instead, MacKinnon has argued, the biotin--and the voltage sensor--has to move at the protein-lipid interface.

But a large number of experimental observations in mammalian channels cannot be reconciled with MacKinnon's model, according to neuroscience professor Francisco Bezanilla of the University of California, Los Angeles. He and other speakers at the meeting reported the results of many other experiments designed to sort out the controversy.

Some scientists have focused on whether the voltage sensor really moves 15­20 Å during channel opening, as MacKinnon's paddle model predicts. Both neurobiology professor Ehud Y. Isacoff of UC Berkeley and assistant professor of neuroscience Robert O. Blaustein of Tufts-New England Medical Center reported results of independent experiments examining the exposure of cysteines introduced into the tip of the putative paddlelike voltage sensor to thiol-selective reagents that can't permeate membranes. Both found that, unlike what the paddle model would predict, "this part of the voltage sensor is in a similar environment whether the channel is opened or closed," Blaustein said.

Bezanilla described fluorescence resonance energy-transfer experiments to try to resolve the question of mobility. His lab first doped the membrane with dipicrylamine, a negatively charged hydrophobic acceptor, then conjugated a rhodamine fluorescent donor dye to different sites on the voltage sensor and watched energy transfer from the rhodamine to the dipicrylamine. "We see no large translocation of the voltage sensor," he concluded.

Others have focused on whether the voltage sensor is cradled entirely by protein or at least partially exposed to lipid, as MacKinnon's model suggests. Physiology professor Richard J. Horn of Jefferson Medical College, Philadelphia, described experiments in which an arginine on the voltage sensor was replaced with a positively charged ethylamine-derivatized cysteine. He reported that this titratable amine retains its proton during gating. "It's hard to image how that might happen if that residue moves through the lipid bilayer," Horn said. Instead, he argued, that residue must move through a protected proteinaceous environment.

Bezanilla's lab has shown that replacing one of the arginines in the voltage sensor with histidine produces a steady leak of protons across the bilayer. He argued that this observation excludes MacKinnon's paddle model because protons would not have access to a histidine that's surrounded by lipid. In an effort to reconcile all of the currently available data, Bezanilla and biophysicist Benoit Roux of Cornell University's Weill Medical College have proposed an alternative "transporter-like" model of voltage gating, in which subtle tilting of an -helical voltage sensor forces the channel open.

But other evidence presented suggests that the voltage sensor isn't surrounded by a protein cocoon. Perozo reported preliminary experiments using electron paramagnetic resonance to probe the voltage sensor's environment in the open channel. He concluded that the portion of the voltage sensor that carries the arginines is exposed to the lipid bilayer.

MacKinnon defended his model in his much-anticipated keynote address. He noted that similar arginine-studded, lipid-exposed paddles were observed in the X-ray structure of a seemingly unrelated kind of voltage-sensitive ion channel, solved by crystallographer Douglas C. Rees of California Institute of Technology. And he reported that his own lab has used electron microscopy to probe the structure of single particles of the antibody-bound bacterial channel in detergent at low resolution. "What we see is that the paddle has to be at the protein-lipid interface, not buried inside the protein," he said.

MacKinnon also shot holes in another argument against the paddle model: that the toxin vstx1--a component of spider venom that shuts these channels down by binding to the voltage sensor--can inactivate both open and closed channels. Because the toxin was thought to be a soluble protein, this result has been held up as proof that the voltage sensor sticks close to the extracellular side of the membrane at all times. MacKinnon presented evidence that vstx1 actually partitions into the membrane.

None of these studies has yet resolved the controversy ignited by MacKinnon's structure. "The structure has forced people to reconsider their preconceived notions about how these channels sense voltage," notes H. Robert Guy, a senior scientist at the National Cancer Institute and an architect of the conventional helical screw model. "It's energized the entire field."