© 2005 Nature Publishing Group

As they describe on page 1191 of this issue1, agroup of researchers led by Christopher Walshhas identified how chlorine is attached enzy-matically to an intermediate during the forma-tion of a natural product. This is not surprisingin itself — the significance lies in the unreactive

nature of the carbon centre concerned.Many natural products require halogens

(chlorine, bromine or iodine) to be strategicallyplaced onto organic molecules at unreactivecarbon centres. Halogenation is essential to thebiological activity and chemical reactivity of

such products (Figs 1a–c), and often serves togenerate versatile molecular building blocks forsynthetic organic chemists. Ideally, these syn-theses would use alkanes — unreactive carbonchains — as their starting materials. These areusually readily available and relatively cheap asthey are the main components of oil. Unfortu-nately, traditional methods for incorporatinghalogens into alkanes often require environ-mentally unfriendly reagents and suffer frompoor control of specificity (Fig. 2a). In con-trast, natural enzymes are benign and do thesame job with extra-ordinary specificity, butlittle is known of the mechanisms of theseenzymatic halogenations. Now, Walsh and colleagues1 have discovered that halogenationof an unreactive carbon centre can be catal-ysed by a halogenase enzyme, called CmaB,that is �-ketoglutarate dependent and containsnon-haem iron.

Similar catalysts are known to be involved inoxygenation chemistry carried out by thehydroxylase family of enzymes. Hydroxylasesinsert oxygen into a carbon–hydrogen bond,an analogous process to halogenation, andhave received much attention because of theirextraordinary specificity and versatility. Innature, several different types of hydroxylasecatalyse such transformations, depending on the substrate. In general, the more reactivesubstrates require the less reactive enzymes.For example, hydroxylation of highly reactivep-hydroxybenzoic acid is readily accom-plished by a flavin-dependent hydroxylase. Incontrast, hydroxylation of relatively unreactivesubstrates, such as the amino acids proline orlysine, requires significantly stronger �-ketog-lutarate-dependent enzymes containing non-haem iron2. In addition to these two extremecases, a variety of alternative hydroxylases has

NEWS & VIEWS NATURE|Vol 436|25 August 2005

1094

monkeys. It will be of interest to determinewhether lateralization is present in otherspecies (as others suggest8), and is thereforerelated to basic properties of sound processing,or whether it is uniquely human and thusmight be a consequence of the development of language.

Now that we know that there are pitch-sensitive neural units, we have to discover howthey work. Sound undergoes many transfor-mations before it gets to the auditory cortex,resulting from the biophysical properties of thecochlea and the many neuronal junctionsbetween cochlea and cortex. We do not yetknow precisely how periodic, temporal infor-mation available in a stimulus is integratedwith the spectral information (or individualharmonics) that is also extracted by the system.We also do not know much about the inputs tothe neurons described by Bendor and Wang.Do they come in a hierarchical arrangementfrom other simpler cells in the auditory cortex?Or do they also receive inputs from subcorticalstructures such as the thalamus? Perhaps

top-down influences from centres associatedwith complex functions in frontal or parietallobes are also significant. This last point is rele-vant, because one technical advantage of thiswork is that the animals tested were awakerather than anaesthetized, meaning that atten-tional and other cognitive factors could have arole. The animals were not trained or behaving,however, so it is difficult to know the signifi-cance of the stimuli for them. Understandingthe interaction between basic perceptual sys-tems and their modulation by higher-ordermechanisms will require more attention tothese factors. Another interesting question is whether these neuronal properties are somehow hard-wired, or whether they are aconsequence of the animals’ environmentalexperience with periodic sounds, which con-tain harmonically related frequencies.

Ian Whitfield9 noted that the problem ofperception is not to determine that two eventsare different, which is actually fairly trivial, butrather that events that might seem to be different are actually the same. It is the job

of the cortex, he argued, to perform the com-putations needed to extract invariancesdespite the different inputs that the environ-ment may provide. The present study1, andthose that will no doubt follow, will lead to amore profound understanding of this funda-mental problem. ■

Robert J. Zatorre is at the Montreal NeurologicalInstitute, McGill University, 3801 UniversityStreet, Montreal, Quebec H3A 2B4, Canada.e-mail: robert.zatorre@mcgill.ca

1. Bendor, D. & Wang, Q. Nature 436, 1161–1165 (2005).2. von Helmholtz, H. On the Sensations of Tone as a

Physiological Basis of the Theory of Music (1863) (trans. Ellis,A.; Dover Publications, New York, 1954).

3. Schouten, J. Koninklijke Nederl. Akad. Wetenschappen Proc.41, 1086–1093 (1938).

4. Bregman, A. Auditory Scene Analysis (MIT Press,Cambridge, Mass., 1990).

5. Zatorre, R. J. J. Acoust. Soc. Am. 84, 566–572 (1988).6. Patterson, R. D., Uppenkamp, S., Johnsrude, I. S. & Griffiths,

T. D. Neuron 36, 767–776 (2002).7. Zatorre, R. J. & Belin, P. Cerebral Cortex 11, 946–953 (2001).8. Wetzel, W., Ohl, F., Wagner, T. & Scheich, H. Neurosci. Lett.

252, 115–118 (1998).9. Whitfield, I. in Cerebral Cortex (eds Peters, A. & Jones, E.)

329–351 (Plenum, New York, 1985).

BIOLOGICAL CHEMISTRY

Just add chlorineNathan A. Schnarr and Chaitan Khosla

Nature provides lessons about developing ‘green chemistry’ in seeminglyout-of-the-way places. One such lesson comes from an enzymatic step inthe production of a leaf toxin by a bacterium.

Figure 1 | Chlorine in natural-product synthesis. a, Coronatine, a leaf toxin. The chlorinatedintermediate (left) goes through several further reactions before coronatine, which does not itselfinclude chlorine, is produced. b, Barbamide, a molluscicide. c, Syringomycin, an antibiotic.Biosynthesis of all three products involves enzymatic chlorination.

25.8 News & Views1093 McP 19/8/05 6:16 PM Page 1094

Nature Publishing Group© 2005

© 2005 Nature Publishing Group

NATURE|Vol 436|25 August 2005 NEWS & VIEWS

1095

been discovered that act primarily on substrates with intermediate reactivity. Byanalogy with the hydroxylases, Walsh and col-leagues now speculate that natural systemsmay employ a similar, tunable strategy forhalogenation.

In this model, more reactive aromatic sub-strates rely on gentler, FADH2-derived agents— and this is indeed the case for chlorinationof tryptophan during the biosynthesis of theantitumour agent rebeccamycin3. Unreactive carbon centres, in contrast, require a morevigorous agent. In the reactions investigated byWalsh and colleagues, the necessary oxidationof a chloride anion to add chlorine to analkane is evidently carried out by a highly reac-tive oxo-iron species from the halogenaseactive site (Fig. 2b). This species presumablybreaks an unactivated carbon–hydrogen bond,leading to a high-energy radical species thatcan be trapped by chlorine.

Analogous enzymes with an apparentlysimilar mode of action are already emergingfrom other biosynthetic systems. For example,Walsh and colleagues4 have also identified aputative halogenase in the biosynthetic path-way for the antifungal antibiotic syringomycin(Fig. 1c). This enzyme is presumably responsi-ble for introducing chlorine into the naturalproduct. Similarly, two relatives of CmaB(BarB1 and BarB2) are found in the bio-synthetic pathway for barbamide, a potentmolluscicide that contains a medicinally inter-esting trichloromethyl group (Fig. 1b).

Generation of high-energy radical inter-mediates has emerged as the common threadamong many oxo-iron catalysts, whose func-tions range from natural product biosynthesisto post-translational protein modification andeven repair of RNA or DNA. The addition ofhalogenation to this impressive array of activ-ities illustrates yet another exciting way to generate useful chemical intermediates fromcomparatively unreactive precursors. Elucida-tion of the precise mechanism for this trans-formation will undoubtedly pave the way fornovel organometallic halogenation catalystsfor chemical synthesis. ■

Nathan A. Schnarr and Chaitan Khosla are in theDepartments of Chemistry and ChemicalEngineering, Stanford University, Stanford,California 94305-5025, USA.e-mail: khosla@stanford.edu

1. Vaillancourt, F. H., Yeh, E., Vosburg, D. A., O’Connor, S. E. &Walsh, C. T. Nature 436, 1191–1194 (2005).

2. Hausinger, R. P. Crit. Rev. Biochem. Mol. Biol. 39, 21–68 (2004).

3. Yeh, E., Garneau, S. & Walsh, C. T. Proc. Natl Acad. Sci. USA102, 3960–3965 (2005).

4. Vaillancourt, F. H., Yin, J. & Walsh, C. T. Proc. Natl Acad. Sci.USA 102, 10111–10116 (2005).

Figure 2 | Chlorinating unreactive organic compounds. a, Using purely chemical synthesis without an enzyme, alkanes are chlorinated using highly reactive radical species (black dots). This requires several steps, and it is difficult to control which particular carbon is modified and included in the final stereochemical form of the product. b, The reaction is much easier and the products more controllable with a dedicated halogenase, as described byWalsh and colleagues1. A chloride anion is oxidized by the oxo-iron species from the halogenase active site for site-specific chlorination of the substrate.

In the 4.5 billion years of its existence, Earthhas been continually losing helium throughthe degassing of rocks from its interior as theymelt during volcanic processes. That there isany helium left on Earth at all is largely owingto its replenishment in the interior throughradioactive �-decay, principally of the heavyelements thorium and uranium — the �-par-ticle emitted in �-decay is a helium nucleus.But �-decay creates only the heavier heliumisotope, 4He. Any trace of the lighter isotope,3He, on present-day Earth is primordial, dating from the planet’s formation.

Admittedly, Earth does not have much 3He:in interior rocks there is only one atom forevery 100,000 atoms of 4He. But the 3He/4Heratio is even smaller in the atmosphere — typically eight or nine times lower, but occa-sionally up to 40 times lower. This imbalancewould seem to imply that Earth still retainssubstantial amounts of primordial heliumtrapped in its interior. But where? Is this gas largely confined to a single deep reservoirthat has remained undisturbed by volcanicactivity all this time? Or is it more uniformlydispersed, with degassing just less efficientthan we had thought? On page 1107 of thisissue, Class and Goldstein1 argue strongly for the latter.

Volcanoes allow a glimpse into the evolu-tion of processes in Earth’s interior. Isotopicanalysis of radiogenic elements implies thatmaterial from Earth’s mantle (the layerbetween crust and core) that is disgorged asvolcanic lava has generally been melted before,at least partially. Helium, a noble gas, is not

chemically bound in minerals, so shouldescape by degassing whenever melting occurs,first to the surface, and from there into space.And although 4He is replaced by radioactivedecay, 3He is not; the higher the 3He/4He ratio,therefore, the less melting and degassing hastaken place.

The existence of rock with high 3He/4Heratios has led to the notion that there is a reservoir of rock, most reasonably in the deep-est mantle, that has escaped melting anddegassing and retains much or all of its origi-nal helium. This idea is supported by the high-est 3He/4He ratios being found in the lavas ofoceanic island volcanoes, such as those onHawaii and in Iceland (Fig. 1). These volca-noes are thought to be produced by convec-tion plumes that carry hot rock from the deepmantle2,3. Yet if a deep reservoir of rock existsin its primordial state, it must be isolated fromthe convection that affects the rest of the man-tle and drives plate tectonics. Seismic imagingof Earth’s interior has, however, consistentlyfailed to find evidence of any layering in thedeep mantle, and implies instead that thewhole mantle is involved in convection4.

The observed ratios of helium isotopes aretherefore problematic. They seem to requirelayered convection that seismologists cannotdetect and geodynamicists cannot reproducein their models. They seem to require a pri-mordial deep mantle, whereas the isotopicratios of other radiogenic elements indicatethat all of Earth’s interior has been affected byearlier volcanic activity.

Class and Goldstein1 attempt to reconcile

EARTH SCIENCE

Helium not in storeWilliam M. White

The ratio of helium isotopes in some oceanic volcanoes seemed to demanda reservoir of virgin primordial gas in the Earth’s mantle. In fact, that mightnot be necessary — a relief for other geophysical models.

25.8 News & Views1093 McP 19/8/05 6:16 PM Page 1095

Nature Publishing Group© 2005