8/2/2019 Peter F. Bernath- Extracting Potentials from Spectra

1/219 JUNE 2009 VOL 324 SCIENCE www.sciencemag.org26

PERSPECTIVES

For most elements, we know whetherthey can form a diatomic molecule,especially for light atoms that have few

electrons and can be treated readily by theory.But for one such light element, surprises are

still in store. For most of the 20th century,experimental and theoretical studies agreedthat the beryllium dimer (Be2) did not exist.The Be atom has filled electron shells andlike the inert gases such as heliumwasexpected to form at most a weak van der Waalsdimer at very large internuclear distances. Yet,as shown experimentally by Merritt et al. onpage 1548 of this issue (1), Be2 does exist andhas a relatively short bond (2.45 ), relative tothe anticipated van der Waals complex with abond length of about 5 . Its unusually flatpotential curve limits the number of vibra-tional levels and provides the rare opportunity

to study the highest vibrational state of a mol-ecule just at its dissociation limit.

Experiments with beryllium are difficultbecause the metal is refractory (it has a lowvapor pressure even at very high tempera-tures) and because beryllium-containingcompounds are generally extremely toxic.However, Be vapor can be created throughlaser ablation of a Be metal target. Rapid cool-

ing of the vapor during supersonic expansionthrough an orifice into vacuum allows prepa-ration of the dimer. The rovibrational states ofthe dimer are then probed with a double-resonance method: One laser excites the mol-

ecule into an excited electronic state where the

atoms are still bound; stimulated emissiopumping (2) by a second laser returns the moecule back into each of the bound vibrationlevels of the ground state.

The data analysis performed by Merret al

. is noteworthy because it allows a bettconnection to theory than stadard methods. Vibration-rotatienergy levels are usually reduceto spectroscopic constants that aparameters in a power series e pansion that uses the relevaquantum numbers of the states (4). However, an excessively largnumber of expansion terms aneeded, particularly for a potentiwith an unusual shape such Be2, and these fitting parametehave lost their physical meanin

In contrast, a parameterized ptential function (see the figurrequires far fewer fitting paramters and makes a direct connectiowith ab initio quantum chemistr

Merritt et al. adopted a mo powerful analysis method tbypassed traditional constants favor of a parameterized potentienergy function obtained fromdirect fit of the energy levels usinthe vibration-rotation Schrding

An analysis of the spectra of the elusive

beryllium dimer, aided by ab initio

calculations, characterizes the molecule

near its dissociation limit.Extracting Potentials from SpectraPeter F. Bernath

CHEMISTRY

Department of Chemistry, University of York, Heslington,York, YO10 5DD, UK. E-mail: pfb500@york.ac.uk

v = 0

v = 1

v = 2

v = 3

v = 4

v = 5v = 6

v = 7v = 8 v = 9

v = 10

2 4 6 8 10

0

200

400

600

800

1000

r ()

Energy(cm1)

Be2(1g)

+

De

Shallow potentials with deeper implications. The potentialenergy function for Be

2as a function of interatomic distance rwas

determined by Merritt et al. from a fit to the experimental obser-vations. The levels become more congested as the energy nears thedissociation limit D

e. The bound vibrational energy levels and the

square of the vibrational wave functions were calculated by LeRoywith his program LEVEL (6).

tumor treatments (another scenario in whichthe immune system is subject to chronic anti-gen exposure) (6, 10), and its therapeuticpotential is appealing. At the same time, thecapacity of IL-21 to enhance the CD8+ T cellresponse may come at a cost: Treating chronicLCMV-infected CD4+-deficient animals withIL-21 led to severe sickness (4). Hence,prospective therapeutic use of IL-21 will needto be finely tuned.

CD4+ T cell help is also important for theCD8+ T cell response to various acute infec-tions (2). In such responses, lack of CD4+

help (or lack of IL-2 sensitivity) typically hasminimal impact on CD8+ T cell priming, butleads to a failure of memory CD8+ T cells tomount a response to subsequent infection bythe same pathogen (2). The current reportsfind no effect of IL-21 or IL-21R deficiencyon the primary CD8+ T cell responses to vari-ous acute infections (35), and in one study,

memory CD8+ T cell responses to LCMVwere intact (5). Such data strongly suggestthat IL-21 is a key element of CD4+ T cellhelp in CD8+ T cell responses to chronic butnot acute infections. Furthermore, the ele-vated and sustained production of IL-21 (butnot IL-2) in CD4+ T cells from chronicallyinfected animals contrasts with the sustainedcapacity for IL-2 (but not IL-21) synthesis byCD4+ T cells following acute infection (3).Thus, the basis by which CD4+ T cells helpthe CD8+ T cell responses may change de-pending on the nature of the infection. IL-21appears to induce a unique differentiationpathway in activated CD8+ T cells (6, 10, 11),potentially equipping them to better clearvirus during a chronic infection. Whether theIL-21producing CD4+ T cells also functionas follicular helper T cells or TH17 cells dur-ing chronic viral infection is unclear [al-though Frhlich et al. argue against TH17

involvement (5)]. Regardless, these studisuggest a key role for IL-21 in mediatinCD4+ help when its needed most.

References1. H. Shin, E. J. Wherry, Curr. Opin. Immunol. 19, 408

(2007).2. M. A. Williams, M. J. Bevan,Annu. Rev. Immunol. 25,

171 (2007).3. H. Elsaesser, K. Sauer, D. G. Brooks,Science 324, 1569

(2009); published online 7 May 2009 (10.1126/science.1174182).

4. J. S. Yi, M. Du, A. J. Zajac,Science 324, 1572 (2009); pulished online 14 May 2009 (10.1126/science.1175194)

5. A. Frhlich et al.,Science 324, 1576 (2009); publishedonline 28 May 2009 (10.1126/science.1172815).

6. R. Spolski, W. J. Leonard,Annu. Rev. Immunol. 26, 57(2008).

7. J. S. Silver, C. A. Hunter, Immunity29, 7 (2008).8. R. Zeng et al.,J. Exp. Med. 201, 139 (2005).9. M. T. Kasaian et al., Immunity16, 559 (2002).

10. C. S. Hinrichset al., Blood111, 5326 (2008).11. K. A. Casey, M. F. Mescher,J. Immunol. 178, 7640

(2007).

10.1126/science.11764

Published by AAAS

8/2/2019 Peter F. Bernath- Extracting Potentials from Spectra

2/2

equation. LeRoy has been one of the pioneersof this modern quantum mechanical ap-proach, and the authors used his freely avail-able computer code (5, 6).

With only eight electrons, Be2 has attractedand challenged quantum chemists for nearly80 years. Ab initio quantum chemistry solvesthe electronic Schrdinger equation to obtainthe interaction energy for a series of moleculargeometries. Within the Born-Oppenheimerapproximation, which separates fast elec-tronic motion from slower vibration-rotationnuclear motion, these electronic energies areused to construct the potential energy functionthat is then used to solve the vibration-rotationSchrdinger equation. It is now even possibleto deal with the breakdown of the Born-Oppenheimer approximation in both experi-ment (7) and ab initio theory (8).

Quantum chemistry attributes the forma-tion of the BeBe bond and the unusual poten-tial shape to the inclusion of more and more p

character in the valence orbitals as the atomsapproach each other (9). The p orbitals aremore directional than the s orbitals thatdescribe the isolated atoms. The comparisonbetween the experimentally derived potentialof Merritt et al. and modern high-quality abinitio potentials such as those calculated in (9,

10) is reasonable at the moment but definitelynot perfect (11), with noticeable differences inthe internuclear distance and in the dissocia-tion energy.

It is experimentally challenging to mea-sure the entire range of vibrational levels andlocate the last bound level at v = 10, which hasa binding energy of only a few wave numbers.Indeed it is even possible that there is anothervibrational level with v = 11 that is bound by atiny fraction of a wave number. As can be seenfrom the square of the wave function, whichgives the probability distribution, the mole-cule spends nearly all of its time at a largeinternuclear separation of 7 forv = 10. Theapplication of stimulated emission pumpingand the availability of a suitable excited elec-tronic state allowed Merritt et al. to map outthe Be2 levels.

Spectroscopic observations often focus onthe lowest few vibrational levels of a potentialbecause they are easier to observe and calcu-

late, but modern computational and experi-mental methodssuch as interactions seen incold atom trapping (12)can provide infor-mation on the last few levels near dissociation.In some cases [for example, MgH (13)], it isonly the combination of modern potential fit-ting methods with high-quality observations

that enables the last few bound levels to blocated in the forest of stronger lines. The fucharacterization of all bound levels with expeimental precision for small molecules such water is still a distant goal, but the first stepwith, for example, H2 (14), MgH (13), and noBe2, are important landmarks on the way.

References and Notes

1. J. M. Merritt, V. E. Bondybey, M. C. Heaven,Science 321548 (2009).

2. C. E. Hamilton, J. L. Kinsey, R. W. Field,Annu. Rev. PhysChem. 37, 493 (1986).

3. J. L. Dunham, Phys. Rev. 41, 721 (1932).4. P. F. Bernath,Spectra of Atoms and Molecules (Oxford

Univ. Press, Oxford, ed. 2, 2005).5. R. J. LeRoy, J. Y. Seto, Y. Huang, DPotFit; a computer

program for fitting diatomic molecular spectral data topotential energy functions, University of WaterlooChemical Physics Research Report CP-662R (2006).

6. See http://leroy.uwaterloo.ca/programs.html.7. R. J. Le Roy et al.,J. Chem. Phys. 123, 204304 (2005).8. D. W. Schwenke,J. Chem. Phys. 118, 6898 (2003).9. J. M. L. Martin, Chem. Phys. Lett. 303, 399 (1999).

10. R. J. Gdanitz, Chem. Phys. Lett. 312, 578 (1999).11. V. Spirko,J. Mol. Spectrosc. 235, 268 (2006).12. J. M. Hutson, P. Soldn, Int. Rev. Phys. Chem. 25, 497

(2006).13. A. Shayesteh et al.,J. Phys. Chem. A 111, 12495 (200714. I. Dabrowski, Can. J. Phys. 62, 1639 (1984).15. I thank R. J. LeRoy for making the figure and for his co

ments. Financial support was provided by the UKEngineering and Physical Sciences Research Council.

10.1126/science.11757

www.sciencemag.org SCIENCE VOL 324 19 JUNE 2009

PERSPECT

Long-term memory storage requiresthe transcription of specific genes inneurons (1). It also requires that the

proteins encoded by these transcripts local-ize to regions in the neurons that forge thecommunicative neuronal connections, orsynapses (2). Yet, how do gene products gen-erated in the neuronal cell body (soma)know to which of all the neurons synapses(up to 30,000) they have to be targeted? Tworeports, by Wang et al. (3) on page 1536 of

this issue and by Okada et al. (4), explorehow long-lasting memory can be imple-mented at specific synapses.

One solution that has been debated is thetagging (or capture) hypothesis (5, 6). The so-called synaptic tag serves as a molecularmarker that targets synaptic plasticity-relatedproteins only to previously activated synapses.

Generally available proteins would be trans-ported into the activated synapses only afterstimulation (4). Alternatively, the synaptictag could spatially restrict new protein synthe-sis (local translation) at stimulated synapses(7, 8). However, until now, there has been nodirect evidence showing that specific messen-ger RNAs (mRNAs) undergo locally restric-ted translation at stimulated synapses onlyduring long-lasting (transcription-dependent)synaptic plasticity.

To approach this problem, Wang et al. cul-tivated sensory and motor neurons of the seaslugAplysia californica (a model system forstudying synaptic plasticity) (1). This mono-synaptic connection is a central part of thegill-withdrawal reflex inAplysia. Multipleapplications of the neurotransmitters sero-tonin or Phe-Met-Arg-Phe-NH2 (FMRFamide)leads to long-term facilitation or depressionof synaptic transmission, respectively, twoeffects that control memory storage. To mon-itor local translation of mRNA during long-

lasting synaptic plasticity, Wang et al. generatan elegant molecular tool using mRNA thencodes sensorin, a sensory neuronspecifpeptide neurotransmitter. Sensorin mRNlocalizes to distal neuronal processes (nerites) and is abundant at synapses. Its transltion to protein is necessary for serotoniinduced long-term facilitation. The authofused the 5'and 3'untranslated regions sensorin mRNA to the region of mRNencoding dendra2, a fluorescent protein (9

Dendra2 switches its fluorescence irrversibly from green to red after illuminatioby ultraviolet light. The authors first coverted dendra2 to red fluorescence by ultravolet irradiation. Afterward, any green dendradetected represents newly produced protevia local translation (after the soma was coff). Wang et al. show that green dendrabecame visible only at synapses that demostrated long-term facilitation, but not unstimulated synapses, nor when long-terdepression was induced.

Imaging with fluorescence reporters reveals th

molecular nature of long-term memory storag

at stimulated synapses.Bridging the Gap and Staying LocalMartin Korte

NEUROSCIENCE

Zoological Institute, Division of Cellular Neurobiology, TUBraunschweig, D-38106 Germany. E-mail: m.korte@tu-bs.de

Published by AAAS