709

Eukaryotic cells are endowed withexquisite machineries for moving pro-teins across their lipid membranes and

out of the cell. But just as every coin has twosides, microorgan-isms have evolvedequally ingeniousmachineries thatallow them to ship

toxic proteins into target cells. Althoughthese machineries seem to have evolvedlargely independently of one another, theyhave converged onto a common basic prin-ciple: to thread proteins across membranesas linear, unfolded polypeptide chains ratherthan as more unwieldy, tightly folded glob-ular structures.

Even when unfolded, a polypeptide can-not by itself penetrate through a lipid mem-brane. It needs some kind of proteinaceouschannel to provide a passageway. On page777 of this issue, Krantz et al. (1) providenew insight into how such a channel—theanthrax toxin pore—works.

Anthrax is caused by the bacteriumBacillus anthracis and normally spreads tohumans via infected animals or contami-nated animal products. Inhalation of bacte-rial spores is particularly dangerous, with amortality rate of nearly 100% if no treat-ment is given. Fortunately, anthrax is a raredisease, although it is regarded as a threat inthe bioterrorism arena.

Anthrax toxin is secreted by the bac-terium in the form of three distinct proteins(2) (see the figure). The pore-forming pro-tective antigen protein f irst binds to areceptor on the surface of the target cell. Itis activated by a proteolytic cleavage eventand assembles into a heptameric preporecomplex. The lethal factor and edema fac-tor proteins then bind to the prepore com-plex, whereupon the whole assembly istaken up by the target cell and deliveredinto an acidic intracellular compartment.The low-pH environment triggers a confor-mational change in the prepore that leads toformation of the toxin pore proper. The low

pH simultaneously causes a partial unfold-ing of the lethal and edema factor proteins,priming them for transport through thepore. Finally, each factor is threadedthrough the pore in a poorly understoodprocess driven by the electrochemicalpotential across the membrane.

The pore itself has the overall shape of amushroom with its stem penetrating themembrane of the target cell. The currentmodel of the stem is that of a 14-stranded βbarrel, 15 Å in diameter, with a water-filledpore running down its center (3). This is thecorrect diameter to allow the passage of anunfolded polypeptide, but how can the porecatalyze the complete unfolding of thetoxin proteins to promote their transportinto the cell?

To address this question, Krantz et al.mutated amino acids thought to line thepore to cysteine, and then modified theseresidues with a small organic reagent.Modification of two neighboring residuesin the pore—Phe427 and Ser429—blockedion conductance, suggesting that theydefine the most narrow part of the pore.Further studies of Phe427 by electron para-magnetic resonance showed that the sevencopies of this residue—one from each ofthe seven subunits—move close to eachother during the prepore-to-pore transitionand thus presumably form an aromatic ringor “φ-clamp” that constricts the pore.

The authors then made a very surprisingobservation. The pore’s ion conductanceincreases, as expected, when Phe427 ismutated to smaller residues such as Ala orSer. In contrast, the translocation ratethrough the pore of a fragment from the lethalfactor protein decreases to an undetectablelevel when Phe427 is replaced by Ala. TheAla427 mutation consequently inactivates thetoxin. Paradoxically, removal of the φ-clampmakes the pore a less efficient translocator,which strongly suggests that the φ-clamp is

M I C R O B I O L O G Y

Translocation of Anthrax Toxin:

Lord of the RingsGunnar von Heijne

The author is in the Department of Biochemistry andBiophysics, Stockholm University, SE-10691Stockholm, Sweden. E-mail: gunnar@dbb.su.se

PERSPECTIVESC

RED

IT:P

RES

TON

HU

EY/S

CIE

NC

E

Cleavage productEdema factor

Lethal factor

Prepore complex

Edema factor

Protein

unfoldingφ clamp

Lethal factor

Low pH

Anthrax pore

Protective

antigen

Cyto

sol

Translocation

Receptor

How anthrax infects cells. Insets show how hydrophobic segments (blue) in the unfolded toxin pro-tein (edema or letal factor) bind in succession to an aromatic ring, or φ clamp, promoting transloca-tion into the host cell cytoplasm.

Enhanced online atwww.sciencemag.org/cgi/content/full/309/5735/709

www.sciencemag.org SCIENCE VOL 309 29 JULY 2005Published by AAAS

710

not simply constricting the pore but has anactive role in the translocation process.

What, then, might be the function of theφ-clamp? To probe its substrate-bindingproperties, the authors tested the inhibitoryeffects on ion conductance of a panoply ofsmall blocking agents such as tetrabutylam-monium or tetraphenylphosphonium. Theyconcluded that the φ-clamp binds theseagents by nonspecific hydrophobic as wellas aromatic π-π and cation-π interactions.This led them to the idea that the role of theφ-clamp is to grab successive hydrophobicsegments in the lethal and edema factor pro-teins as each chain is pulled through thepore. In essence, they propose a “Brownianratchet” model where transient unfolding ofthe protein exposes hydrophobic segmentsthat bind to the φ-clamp, preventing refold-ing and facilitating the conversion to atranslocation-competent form.

There is a striking analogy between the φ-

clamp structure proposed by Krantz et al.and the so-called hydrophobic gasket foundin the Sec61 translocon that mediates proteintranslocation across the inner membrane ofbacteria and the endoplasmic reticulummembrane of eukaryotes (4). In the Sec61channel, a ring of hydrophobic Ile residues isthought to provide a flexible seal around thetranslocating polypeptide, preventing ionleakage through the membrane. At the sametime, the Ile ring is expected to bind tohydrophobic segments in the polypeptide,perhaps shunting very hydrophobic trans-membrane helices into the surrounding lipidmembrane. Another comparable case is theGroEL/ES chaperonin in which hydrophobicresidues project into a central cavity. Thisconfiguration is thought to unfold entrapped,misfolded proteins by pulling on hydropho-bic residues exposed on their surface (5).And recently, it was proposed that movableloops protruding into the central channel of

the ATP-driven ClpA protease push unfoldedsubstrates toward their destruction (6).

These likely represent only a few exam-ples of how cells manipulate unfolded pro-teins. By exploiting the most basic charac-teristic of the unfolded state—the exposureof hydrophobic residues—cells have micro-engineered sophisticated molecularmachines to push and pull proteins, deliver-ing them to their final destinations.

References1. B.A. Krantz et al., Science 309, 777 (2005).2. R. J. Collier, J.A.T.Young, Annu. Rev. Cell Dev. Biol. 19, 45

(2003).3. B. A. Krantz, A. D. Trivedi, K. Cunningham, K. A. Chris-

tensen, R. J. Collier, J. Mol. Biol. 344, 739 (2004).4. B. van den Berg et al., Nature 427, 36 (2004).5. W. A. Fenton, A. L. Horwich, Q. Rev. Biophys. 36, 229

(2003).6. J. Hinnerwisch,W. A. Fenton, K. J. Furtak, G. W. Farr, A. L.

Horwich, Cell 121, 1029 (2005).

10.1126/science.1116630

Precise knowledge of the frequenciesof emission lines produced by quan-tum state transitions in atoms is

essential for tests of fundamental physicsas well as for the development of betterclocks and measurement standards. Themost precisely known atomic resonancefrequencies in the optical spectral range arecurrently those of transitions in the positiveions of strontium, mercury, and ytterbium(1–3). At first thought, these species maynot appear to be models of simplicity, norare they treated very prominently in classi-cal spectroscopy textbooks. They provide,however, a combination of properties thatmake them well suited for the applicationof laser spectroscopy methods that permitthe highest precision. Unfortunately, only alimited number of ions have these specialproperties. On page 749 of this issue,Schmidt et al. (4) demonstrate a tech-nique that dramatically widens therange of atoms or molecules that areamenable to this kind of precisionspectroscopy. Key ingredients of themethod have been developed in thecontext of a different research field—quantum information processing. Now,these quantum logic tools may f indanother important application in the devel-opment of optical clocks.

The seminal ideas for laser spectroscopyof trapped ions were put forward by Dehmeltmore than 30 years ago (5). The methodswere developed and perfected by severalgroups and became highly successful (2, 3).Precision spectroscopic analysis begins withisolation of a single ion, which is stored in aminiature electric trap. A laser is used toexcite a broad resonance line from theground state, and the emitted photons carryaway portions of the ion’s motional energy,leading to efficient cooling. The resultingreduction in thermal fluctuation improvesthe localization of the ion and eliminates fre-quency shifts produced by the Dopplereffect. The fluorescence light that is scat-tered during laser cooling also permits the

optical detection of the ion. In addition, theion must possess a metastable level that cou-ples to the ground state in a very narrowtransition. This sharp line can then be usedfor precision measurements or can serve as areference for an atomic clock. The excitationof the narrow transition is monitored in thefluorescence light emitted during cooling: Ifthe ion is excited to the metastable level, itdecouples from the cooling laser excitationand fluorescence ceases.

The requirement of having two transi-tions with very different linewidths thatoriginate from a common ground state andcan be excited with lasers at technicallyaccessible wavelengths has limited thesestudies to only a handful of atomic ions.

Wineland et al. proposed a scheme (6)that relaxes the requirements by distribut-ing the tasks to two different ions that aretrapped together. The same group (4) nowreports an experimental demonstration ofthe method. The narrow transition to be

P H Y S I C S

Logical SpectroscopyEkkehard Peik

The author is in the Time and Frequency Department,Physikalisch-Technische Bundesanstalt, 38116Braunschweig, Germany. E-mail: ekkehard.peik@ptb.de

Ion trap

Cooling phase Mapping phase Detection phase

Quantum logic spectroscopy. (Left) The “logic ion” (blue) is laser-cooled and the “spectroscopy ion”(yellow) is cooled sympathetically. (Center) A narrow transition of the spectroscopy ion is probedwith a laser pulse. With additional laser pulses, the internal state of the spectroscopy ion is mappedto the logic ion via the common vibrational motion. (Right) The internal state of the logic ion can beread out via laser-induced fluorescence. CR

EDIT

:PR

ESTO

N H

UEY

/ SC

IEN

CE

29 JULY 2005 VOL 309 SCIENCE www.sciencemag.org

P E R S P E C T I V E S

Published by AAAS