t is one thing to talk metaphorically about the molecular pumps that enable cells to move molecules against a concentration gradient, and quite another to discover in detail how those pumps actually work, right down to the valves and filters and pistons. But three currently hot papers do just that. At #9 and #11 are two papers from 2003 Nobel laureate in chemistry Roderick MacKinnon’s group at the Howard Hughes Medical Institute, Rockefeller University, New York, laying bare the inner workings of a bacterial potassium channel (paper #11 is Y. Jiang, et al., Nature, 417[6888]: 523-6, 30 May 2002; 26 citations this period). And at #6 is a similar tour de force from Chikashi Toyoshima and Hiromi Nomura of the University of Tokyo, on the calcium pump that underlies muscle contraction.

In muscle cells, the release of calcium ions into the cytoplasm triggers contraction. The muscle relaxes when the calcium ions are pumped back into storage inside the sarcoplasmic reticulum. The pump is an enzyme that sits across the membrane and in muscle cells forms up to 90% of the membrane protein. In 2000 Toyoshima and colleagues published a detailed structure for the enzyme in the form it takes when bound to calcium ions. Now they have looked at the enzyme’s shape again, this time in the presence of a potent inhibitor that fixes the structure closed, as it were, having given up its calcium ions. The differences, in particular the movement of various parts of the molecule relative to one another, not only showed how the pump works, but also astonished the authors. "The molecular mechanism was far beyond what we could imagine," they wrote.

The molecule has three domains on the cytoplasm side of the membrane. The N domain binds the nucleotide, ATP. The P domain is responsible for phosphorylation. And the A domain is the actuator. When bound to calcium ions, the three domains within the cytoplasm are well separated from one another. ATP fits into the gap between the N and P domains, closing the gap and putting the active site of the P domain in reach of a crucial aspartic acid residue, which it phosphorylates. The new work shows the second stage. The P, N and A domains move together to form a cluster. This rearrangement disrupts the calcium binding sites and the calcium ions move out into the sarcoplasmic reticulum. Water from the sarcoplasmic reticulum enters the pore, converting the structure back to one that can accept another two calcium ions from the cytoplasm.

All of which is hard to explain adequately in a few words, but a look at the animations reveals that the movements are relatively large and exquisitely choreographed.

The Japanese animations flip between the two structural changes they have elucidated. MacKinnon’s group goes one better with a movie of the potassium channel opening and closing. One sees four protein subunits that open and shut rather like the spiral leaves of a camera’s aperture iris.

What is actually happening is rather more complex. The four sub-domains are angled together to form a sort of funnel, or teepee, with its tip at the inside of the membrane. This closed tip is the pore’s gate. In the center of the membrane is a space that accomodates a hydrated potassium ion. And on the outside of the membrane the protein sub-domains come together to form a ring that is a selectivity filter, just the right size to let through potassium ions and no others. MacKinnon’s group studied the potassium channels from two different bacteria. In one, Streptomyces lividans, the isolated pores are closed with a potassium ion inside. In the other, from Methanobacterium thermoautotrophicum, the pore is open, calcium having bound to the gate. That provided hints about how the sub-domains move to pump the ions through the membrane.

The structure of the outside portion, including the filter, is practically identical in both forms. But the poles of the teepee are straight in the closed pore, while in the open pore they are bent in the middle and splayed out. The space, into which the potassium ion fitted so neatly, is opened up, freeing the ion to drift out of the pump. The bend takes place at a glycine amino acid deep in the pore near the selectivity filter, and a look at the structures of several different potassium channels shows that this residue is highly conserved. Indeed, the mechanism is essentially similar in bacteria and mammals. Something—calcium binding, a voltage change—forces open the gating ring, which bends the teepee poles and releases the ion.