As the gateway to the nucleus, the nuclear pore complex manages hundreds of intricate cargo-handling operations every second. It decides which molecules enter the 120-nm long channel leading to the nucleus. The NPC ferries pairs of proteins as small as 50,000 daltons from cytoplasm to nucleus and shuttles ribosomal subunits as massive as 2 megadaltons going the opposite way.

To handle the array of molecular interactions on which these transactions depend, the NPC comprises 700 to as many as 1,000 polypeptides, known as nucleoporins or nups, drawn from about 30 different protein varieties. The locations and working relationships of those individual proteins have eluded researchers for more than two decades. But creative uses of cryoelectron tomography, X-ray crystallography, and computer modeling are taking biologists a collective step closer to the answer. Says Susan Wente, chair of cell and developmental biology at Vanderbilt University, "I think within two or three...


At the Max Planck Institute of Biochemistry in Martinsried, Germany, institute director Wolfgang Baumeister has headed a series of projects pioneering electron tomography to gain an increasingly detailed bird's-eye view of the NPC.

As in computed axial tomography (CAT) scans used on medical patients, the technique compiles a series of images by bombarding a sample with low doses of electrons. But, while hospital CAT scans rotate the detector around the stationary patient to make a comprehensive structural picture, in cellular-scale tomographic studies the detector stays still while the sample is tilted back and forth around an axis in minute increments.

Last November the group took as many as 120 images each of more than 300 NPCs found in the cellular slime mold, Dictyostelium discoideum.1 The samples were flash-frozen to capture structures and processes as close to nature as possible.

Taking images of the NPCs still in place in the spherical nuclei was key to improving the emerging picture. Rotating a flattened sample of a round, three-dimensional object above a fixed, narrow detector causes distortions and can leave blank spaces in the data. (Think of the ways in which world maps render the shapes and relationships of continents.) By leaving the nuclei attached to their native spherical surface as they rotate, "you have something more like a little missing cone of data instead of a large missing wedge," Baumeister explains.

The group's latest work shows that the NPC's central channel averages a length of about 120 nm, not the 200 that had been previously thought. By sharpening resolution to about 8 nm from its previous 12, the group also captured the first clear images of the NPC's cytoplasmic filaments – long, trailing arms that seem to guide cargo into the channel.

The group was also able to organize image data to reveal "major rearrangements within the pore complex" as material passes through it – adding evidence to the conjecture that the pore must flex like an esophagus to allow large molecules to reach the nucleus but screen out smaller, harmful ones.


Although the institute's work broke new ground, it didn't address the long-standing debate over the nature of the mass that turns up consistently in images of the NPC's inner channel. No one knows whether this central plug – which some call the transporter – is merely cargo in transit or a structural part of the NPC itself.

In the channel, a disordered collection of protein subunits known as "F-G repeats" is thought to move cargo by a series of hand-to-hand transfers. Those who believe they see a transporter mechanism claim that this sphincter-like structure is a necessary touchstone that the repeats use to organize themselves and manage their actions. Those who think that the central plug is baggage passing through argue that no such organizational platform is necessary.

Terry Allen, a structural cell biologist at Cancer Research UK's Paterson Institute in Manchester, England, is squarely in the transporter's corner. "Our lab is sure because we always see the same internal structure in the same region of every nuclear pore complex we've looked at over the last 15 years," he says. "People say that they can look through the cytoplasmic end of the pore complex and see clearly all the way to the nuclear basket," a framework just inside the nucleus. But Allen's group has been able to biochemically strip away the basket, he says, "and we still see the same structure in the channel," indicating that people who think they see the basket may be looking at something else. He attributes his group's discerning vision to an especially powerful scanning electron microscope combined with specimen preparation techniques that can resolve images of individual pores to a few nanometers instead of averaging data from several.

Champions of the central-plug-as-cargo theory don't see the same structure in the same place in every image and point to Baumeister's recent study as evidence: In it, plugs are seen in various locations as far apart as 40 nm, about a third of the channel's length.

Jan Ellenberg, a research group leader in cell biology and biophysics at the European Molecular Biology Laboratory in Heidelberg, Germany, notes that a good deal of biochemical evidence supports the theory that the F-G repeats form a hydrophobic mesh inside the channel and that no other organizing mechanism is necessary to explain the repeats' behavior. "A central density in the pore complex would rather be expected to result from large cargo molecules caught in transit," he asserts. Ellenberg's own recent work2 indicates that the pore dynamically sheds some nucleoporins, although at a much lower rate than cargo travels through it – a finding that some,3 although not Ellenberg himself, use to shore up the argument that the plug is cargo.

To advance the debate, Baumeister's group is taking cryoelectron images of gold-tagged proteins as they travel through the pores. Baumeister says he hopes to publish results before the year ends.


<p>THE BIG PORE:</p>

© 2004 AAAS

A and B are stereoviews of the Dictyostelium nuclear pore structure from the cytoplasmic side and the nuclear side, respectively. C shows a cutaway view of the pore with dimensions indicated and the central plug/transporter removed. (From M. Beck et al., Science, 306:1387–90, 2004.)

If Baumeister is taking a bird's-eye view of the NPC, a group led by Nobel laureate Günter Blobel at Rocke-feller University is taking the worm's-eye perspective. The researchers are parsing the structures of the pore complex's individual proteins and recently charted the structure of nup133, a member of the pore subcomplex nup160-107 that has been shown to be key in forming the NPC's structural framework.4

Blobel and his researchers discovered that this particular nucleoporin unexpectedly included a β propeller – a distinctive, rigid fold, made of β strands, suitable for contributing stability to the pore complex's structure. The group also offered evidence that three other nups have β propellers, bringing the known total of such polypeptides to 11 – more than a third of the NPC's proteins. Blobel says he is happy with the result, which reveals much more about the modularity of the structure than expected. "That makes the complex more amenable to structural analysis. Otherwise, you would say, 'Who wants to study a structure that is over 100 million daltons in size?' It sounds like going to Jupiter or something."

Although some proteins' functions can be loosely associated with their shapes, it's too soon to divine nup133's specific responsibilities, according to Thomas Schwartz, part of Blobel's team who is now at MIT.

Michael Rout, a colleague of Blobel at Rockefeller University, may speed that understanding. His group recently provided compelling evidence about the origins of the pore – less akin to membrane channel proteins than to membrane-bending vesicle transport proteins.5 But he's currently combining perspectives in a computer model that may place each individual nup in its proper place. His group works with Andrej Sali, computational biochemist at the University of California at San Francisco, to build a complete, working digital model of the entire NPC. "To a first approximation, we think we've got it," Rout says.

The group began by using strings of spheres to represent each protein molecule in the NPC, then programmed in all of the structural information, interprotein relationships, and behavioral constraints they knew. "We obtained data on 70 or 80 subcomplexes, most of them not previously described," Rout says. "If one turned out to be a dimer, such as nupX connected to nupY, that becomes a constraint – a sort of coiled spring that draws the two together." When they run their model, hundreds of such springs organize the spheres. As the group adds new known structures and behaviors, the simulations converge on a single model: "a doughnut within the nuclear membrane" that closely resembles the known shape of a nuclear pore and is revealing previously unseen details about the complex's organization, Rout says. His group is drafting the results for publication.

"Each of these recent [structural] projects is an incremental step," says Vanderbilt's Wente. "But, taken together, the results are quite significant."

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!