gene silencing, and so RIP and methylationtogether seem to represent a belt-and-bracesstrategy to ensure that transposons remaininactive. However, Galagan and colleagues’intriguing finding that some regions of thetext are methylated even though they are notRIPed raises the possibility that methylationalso has another function in Neurospora.

Beadle and Tatum2 said that, “From thestandpoint of physiological genetics thedevelopment and functioning of an organ-ism consist essentially of an integrated system of chemical reactions controlled insome manner by genes”. With the Neurosporagenome in hand, researchers can now movetowards a specification of the chemical reactions that link genes and proteins inprocesses such as metabolism, biologicalclocks and development15. We may also beable to borrow from computer science todescribe the reaction network in terms of a‘biological circuit’ and discover more abouthow this fungus functions. New genomicsapproaches can measure the responses of the circuit (through RNA and protein pro-filing) and the links between its components(through protein–protein and protein– DNA

interaction mapping). The genetic text,together with such approaches, should bringa further revelatory consilience of biochem-istry and genetics16. ■

Jonathan Arnold is in the Department of Geneticsand Nelson Hilton is in the Department of English,University of Georgia, Athens, Georgia 30602-7223,USA.e-mails: arnold@arches.uga.edunhilton@arches.uga.edu1. Davis, R. H. Neurospora: Contributions of a Model Organism

(Oxford Univ. Press, 2000).

2. Beadle, G. W. & Tatum, E. L. Proc. Natl Acad. Sci. USA 27,499–506 (1941).

3. Raju, N. B. Eur. J. Cell Biol. 23, 208–223 (1980).

4. Galagan, J. E. et al. Nature 422, 859–868 (2003).

5. Selker, E. U. et al. Nature 422, 893–897 (2003).

6. Kelkar, H. S. et al. Genetics 157, 979–990 (2001).

7. Kraemer, E., Wang, J., Guo, J., Hopkins, S. & Arnold, J.

Bioinformatics 17, 901–912 (2001).

8. Adams, M. D. et al. Science 287, 2185–2195 (2000).

9. Venter, J. C. et al. Science 291, 1304–1351 (2001).

10. International Human Genome Sequencing Consortium Nature

409, 860–921 (2001).

11.Goffeau, A. et al. Science 274, 546–567 (1996).

12.Dunlap, J. C. Cell 96, 271–290 (1999).

13.Gustin, M. C., Albertyn, J., Alexander, M. & Davenport, K.

Microbiol. Mol. Biol. Rev. 62, 1264–1300 (1998).

14.Selker, E. U. Adv. Genet. 46, 439–450 (2002).

15. Ideker, T. et al. Science 292, 929–934 (2001).

16.Battogtokh, D., Asch, D. K., Case, M. E., Arnold, J. & Schuttler,

H.-B. Proc. Natl Acad. Sci. USA 99, 16904–16909 (2002).

Individual electrons, trapped by electricand magnetic fields, can be used for pre-cise, fundamental measurements. D’Urso,

Odom and Gabrielse1 now report in PhysicalReview Letters that they have succeeded incontrolling the residual random motion of a trapped electron using feedback signals.This degree of control could be the basis ofexperiments to measure elementary con-stants of nature with even greater precisionthan before, and to accurately test the quantum theory that governs electrons andradiation — quantum electrodynamics.

The ‘calming’ of isolated quantum sys-tems, by reducing their kinetic energy, isbehind much of the progress in the physics of light and atoms in the past decades. Laser cooling has produced much narrowerspectral lines of emitted radiation from freeatoms2 and bound ions3–5, and hence moreprecise atomic data6. Evaporative cooling ofatom clouds in magnetic or optical traps —similar to the way in which coffee cools in a cup — made possible the phase transitionto a new state of matter, a Bose–Einstein condensate7,8. Experimenters can force particles such as antiprotons to condenseinto tight bunches inside the storage ring of

an accelerator by detecting their randomexcursions and feeding that informationback in the form of a changing electrode voltage to coax them back into line. Thisreduction of random motion is stochasticcooling9 — a decisive factor in the discoveryof the W and Z bosons, the fundamental particles that mediate nature’s weak force.

Feedback, a controlling operation deter-mined by the result of a preceding measure-ment, may be imposed on any system. In earlier experiments, the residual motion of a single ion inside an electrodynamic trap(consisting of a static and an alternating electric field) has been reduced using a kindof feedback10,11. Here, sequences of almost-resonant light pulses absorbed quanta ofvibrational energy from the ion, damping itsfluctuations. Now D’Urso et al.1 demon-strate that it is possible to cool an individualquantum particle, an electron, by straight-forward, continuous electronic feedback.

In their experiment, an electron is cap-tured and trapped in an electromagneticcage created from a homogeneous magneticfield and an inhomogeneous electric fieldthat is zero at the centre of the apparatus (Fig. 1). This technique was pioneered some

25 years ago in the Nobel-prizewinning workof Hans Dehmelt12. The magnetic field curlsthe path of the trapped electron into a spiral-like trajectory. A positively charged torus andtwo negatively charged caps, aligned on thetrap axis, prevent the electron escaping. Theupper cap also serves as a sensitive probe forthe tiny voltage fluctuations caused by thehovering electron. Through a feedback loop,those voltage changes are amplified and fedback to the lower cap, creating corrective tugson the electron’s motion.

The random voltage that is induced bythe thermally fluctuating electron motionalong the axis of the trap has been measuredin a frequency-selective way. The result is aspectrum of noise power that has a notch atthe frequency value that matches the reso-nant frequency of this axial electronic oscil-lation — the electron can be represented by aserial circuit that acts, on resonance, as anelectric short between the caps. The width ofthe notch represents how much the fluctuat-ing electron motion is damped, characteriz-ing the dissipation of vibrational electronenergy into the electric circuit. Increasingthe signal gain in the feedback loop causesthis damping to decline, and with it shrinksthe width of the spectral notch (Fig. 1).

The temperature of an ensemble of parti-cles quantifies the fluctuations of this system.When this indicator is applied to a single elec-tron, the particle’s motional excursions aremeasured again and again, and the stochasticensemble of observations can be easily trans-formed into recorded values of motionalenergy; these snapshots form, in the equili-brium state, a thermal (or Boltzmann) distribution, with higher energy values beingexponentially less likely than lower ones. Atricky way of monitoring such a distribution— called thermometry — has been sug-gested by Dehmelt13,14. There are, apart from the electron’s axial motion, two more,independent modes of vibration: a fastazimuthal motion in which the electronorbits the trap axis, called ‘cyclotron motion’;and a slow pulsation of the orbital radius,called ‘magnetron motion’. These motionsmay be excited in differently quantized portions of vibrational energy, and a temper-ature measurement amounts to finding outhow often and for how long the first excitedstate (and the second, third, and so on) of thequantized motion is randomly occupied bythe electron — the mean numbers of occupa-tion forming a Boltzmann distribution whenthe electron is in thermal equilibrium.

The job of measuring this probability ofexcitation in a non-invasive and precise waywould seem to require a complex strategy —and the patience of indefatigable Sisyphus.The elegant (although not quite straight-forward) solution is to make frequencymeasurements, which are famously pre-cise and convenient, using two coupledoscillators known to be shifting each other’s

news and views

822 NATURE | VOL 422 | 24 APRIL 2003 | www.nature.com/nature

Quantum electronics

The electron is coolPeter E. Toschek

Temperature is an awkward concept for a single particle, but the energy ofthe particle’s ‘quivering’ is a useful substitute. Feedback control of thismotion can be used to cool a single electron to very low temperature.

© 2003 Nature Publishing Group

frequency of oscillation in step with theirown excitation, as two connected pendula do in the familiar classroom experiment.Here, it suffices to couple the cyclotron and the axial motion through the minuteinhomogeneous magnetic field of a tinynickel wire.

To measure the thermal distribution ofthe electron’s energy of axial vibrationrequires a two-step ‘pump–probe’ opera-tion. Because any axial energy value is corre-lated with a shifted cyclotron frequency, theprobability of an axial excitation — its‘strength’ — is mapped onto the particularstrength of cyclotron excitation when theelectron is irradiated with microwaves of afrequency correspondingly shifted off reso-nance. The quanta of cyclotron excitation aretiny compared with light quanta, but muchlarger than those of axial excitation: the elec-tron absorbs at most one of them when ran-domly jumping to its first cyclotron-excitedquantum state. The strength of excitationmanifests itself in the fraction of successfulquantum jumps to this state when a series ofmicrowave ‘pump’ pulses is applied.

Then, to detect whether the electron hasreached the first excited state after a pumppulse, the inverse correlation, namely ofcyclotron energy and axial frequency, comesinto play. A strong ‘probe’ pulse, at the axialfrequency on resonance, does not affect thecyclotron-excited electron — except whenthe probe frequency is tuned off resonanceby a tiny shift, just 12 Hz in D’Urso and col-leagues’ experiment1. Thus, the mean chanceof axial excitation is detected, via cyclotronexcitation, by recording the response of theelectron, seen by the electronic circuit, to the

probe pulse at the shifted axial frequency. When the microwave pump radiation is

varied in steps around the cyclotron reso-nance, the resulting spectrum of measure-ments shows how likely it is that the electronhas acquired one quantum of cyclotronmotion. This spectrum mimics a Boltzmanndistribution, whose width directly displaysthe ‘temperature’ of the electron’s axialvibration. The measurements1 show that,when the strength of the feedback isincreased, the width of the distributionshrinks down — the lowest temperaturereached in their experiment was 850 mK.

Singling out and cooling electrons gives a means of measuring, with very great

accuracy, fundamental quantities in quan-tum electrodynamics, such as the ratio of theelectron’s spin to its magnetic moment. In aheroic effort by theorists, this number hasbeen calculated to ten decimal places, and sofar, no significant discrepancy has beenfound between it and the value measured inexperimental tests. But it is quite possiblethat even higher precision may reveal a deviation — or even that the value for theelectron’s antiparticle twin, the positron, isdifferent. This is crucial in the matter–antimatter relationship that is at the veryheart of our understanding of matter. D’Urso and colleagues’ feedback coolingmay well contribute to dramatically re-ducing errors in these measurements, furthering our never-ending struggle forhigher accuracy of observation and refinedmodelling of nature. ■

Peter E. Toschek is at the Institut für Laser-Physik,Universität Hamburg, Jungius-Straße 9, D-20355Hamburg, Germany.e-mail: toschek@physnet.uni-hamburg.de1. D’Urso, B., Odom, B. & Gabrielse, G. Phys. Rev. Lett. 90, 043001

(2003).

2. Hänsch, T. W. & Schawlow, A. L. Opt. Commun. 13, 68–69

(1975).

3. Wineland, D. J., Drullinger, R. E. & Walls, F. L. Phys. Rev. Lett.

40, 1639–1642 (1978).

4. Neuhauser, W., Hohenstatt, M., Toschek, P. E. & Dehmelt, H. G.

Phys. Rev. Lett. 41, 233–236 (1978).

5. Neuhauser, W., Hohenstatt, M., Toschek, P. E. & Dehmelt, H. G.

Appl. Phys. 17, 123–129 (1978).

6. Wineland, D. J. et al. in Proc. 4th Symp. Freq. Standards

Metrology (ed. DeMarchi, A.) 71–77 (Springer, Berlin, 1989).

7. Anderson, M. H., Ensher, J. R., Matthews, M. R., Wieman, C. E.

& Cornell, E. A. Science 269, 198–201 (1995).

8. Davis, K. B. et al. Phys. Rev. Lett. 75, 3969–3973 (1995).

9. van der Meer, S. Rev. Mod. Phys. 57, 689–697 (1985).

10.Eschner, J., Appasamy, B. & Toschek, P. E. Phys. Rev. Lett. 74,

2435–2438 (1995).

11.Appasamy, B., Stalgies, Y. & Toschek, P. E. Phys. Rev. Lett. 80,

2805–2808 (1998).

12.van Dyck, R. S. Jr, Schwinberg, P. B. & Dehmelt, H. G.

Phys. Rev. Lett. 38, 310–313 (1977).

13.van Dyck, R. S. Jr, Ekstrom, P. & Dehmelt, H. Nature 262, 776

(1976).

14.Dehmelt, H. Am. J. Phys. 58, 17–27 (1990).

The respective merits of embryonic versus adult stem cells for treating diseases have been widely discussed,

in both the popular press and the scientificliterature. In particular, reports that thedevelopmental potential of adult stem cellsmight be greater than previously supposedhave aroused strong interest and controversy.The papers by Wang and colleagues1 andVassilopoulos and co-workers2 on pages 897and 901 of this issue mark a further turn in

the debate. These authors confirm that stem cells derived from adult bone marrow can repair damaged liver tissue in mice —but not by converting directly into liver cells, as might be expected if the stem cellscould change their developmental ‘destiny’.Instead, cell fusion with the host liver isresponsible for bringing down the barriersbetween bone marrow and liver.

Many tissues and organs in adult mam-mals contain reserves of stem cells to ensure

news and views

NATURE | VOL 422 | 24 APRIL 2003 | www.nature.com/nature 823

Stem cells

Fusion brings down barriersAlexander Medvinsky and Austin Smith

It remains uncertain how tissue-specific stem cells could generate themature cell types of another tissue. In one instance, where bone-marrow-derived stem cells repair damaged liver in mice, cell fusion is the answer.

Figure 1 Cooling down. In the device designed by D’Urso et al.1, a single electron is trapped betweentwo caps and a torus by the combined effect of an electric and a magnetic field. The upper cap detectsthe random motion of the electron, and a signal sent through the feedback circuit triggers a voltagechange on the lower cap to counteract it. The effective cooling of the electron is revealed in the powerspectrum of noise in the circuit (inset; derived from ref. 1): a notch in the spectrum shrinks as thefeedback increases; the frequency at which the notch appears is the resonant frequency of the electronbetween the caps.

Feedback amplifierSignal detection

Feedback loop

Negativelycharged cap

Negativelycharged cap

Electric-field lines

Trappedelectron

Magneticfield

Incr

easi

ng fe

edb

ack

Frequency

Pow

er

Positivelycharged torus

© 2003 Nature Publishing Group