eukaryote that has been studied. (Alternative splicing is when segments of a transcript can be joined in different conformations to pro­duce different forms of a given protein or lead to different subcellular localization.) Why this organism splices so wildly and how this influ­ences its biology are questions for the years to come.

Both algal genomes also harbour what evolutionary biologists call conflicting phylo­genetic signals: there are a surprisingly large number of genes of apparent green­algal ori­gin in Guillardia, which actually harbours the remnants of a red alga, and too many red­algal genes in the green­alga­harbouring Bigelowiella (Fig. 1b). This Pandora’s box was first opened when such ‘green signals’ were identified in related heterokontophyte algae, and additional layers of endosymbiotic events were proposed to explain this appar­ent conflict8. However, more recent analyses favour phylogenetic errors and sampling bias as the explanation9, 10. As more red algal genomes are sequenced and a more balanced set of reference genomes becomes available, the conundrum of what early evolutionary track these complex cells took will probably be solved.

Perhaps the most exciting insight revealed by the two genomes is why these cells have not lost the nucleomorph — a question that has puzzled researchers since the discov­ery of these organelles. Eukaryotic nuclear genomes are typically replete with copies of mitochondrial or plastid DNA11. Humans, for example, have about 500 kilobases of mito­chondrial DNA strewn across their chromo­somes12. These insertions arise when one of these organelles lyses and releases its DNA, fragments of which then enter the nucleus, with some being successfully integrated into the nuclear DNA.

But what if there were only one mitochon­drion per cell? Its lysis would mean no mito­chondrion­bearing progeny and the death of that cell lineage (because mitochondria are essential for cellular energy provision). Nota­bly, only one nucleomorph–plastid complex exists per cell in Guillardia and Bigelowiella. So, although Curtis et al. find mitochondrial DNA in both nuclear genomes, indicating that recent gene transfer from the mitochondrial organelles has occurred, they find no recently transferred copies of nucleomorph or plastid DNA. This implies that the presence of a single nucleomorph–plastid complex prevents any further gene transfer in that cell, such that the nucleomorph is evolutionarily frozen. This constellation tells us that it takes two to tango on the route to becoming an enduring secondary organelle. ■

Sven B. Gould is at the Institute for Molecular Evolution, Heinrich Heine University, 40225 Düsseldorf, Germany. e-mail: gould@hhu.de

E X T R A S O L A R P L A N E T S

Astrophysical false positivesThe probability that giant-planet-like signals detected by the Kepler spacecraft are not from planets is higher than expected. The result underscores the importance of making follow-up observations to confirm the nature of the signals.

A N D R E W C O L L I E R C A M E R O N

For the past three years, the mosaic of detectors on board NASA’s Kepler mission have been staring at 150,000 stars

in a region roughly halfway between the bright stars Vega and Deneb. They have been hunt­ing for the recurring dips in light that betray the presence of planets whose orbital planes are nearly edge­on to the line of sight. In that short time, the Kepler team has announced some 2,300 candidate planetary systems. Writing in Astronomy & Astrophysics, Santerne et al.1 report that 35% of giant, close­in candi­date planets found by Kepler are astrophysical false positives — that is, stars in binary systems that eclipse their companions with a change in brightness (depth) and duration similar to that expected for planets. This value is several times larger than expected.

Santerne and colleagues’ team studied 46 giant, close­in Kepler candidates over sev­eral weeks using the SOPHIE spectrograph mounted on the 1.93­metre telescope at the Haute­Provence Observatory in wooded hill country north of Marseille, France. The goal of the study was to determine the masses of the putative planets, by measuring the parent star’s reflex orbital motion — the back­and­forth ‘wobble’ caused by the gravitational pull of the planet — around its common centre of mass with the planet. A planetary­mass deter­mination is an unambiguous means of ruling out astrophysical false positives, as well as determining the bulk density, and hence the likely composition, of a planet.

The target stars were selected carefully. They had to be bright enough to yield useful results in a reasonable allocation of telescope time. The candidate planets to be investigated

were also selected carefully. Their passages in front of the host stars (transits) had to decrease the brightness of the star sufficiently (a deep enough transit) to indicate that they had high mass and would thus yield a detectable reflex motion. Finally, the suspected planets had to have orbital periods shorter than 25 days, to ensure that their orbital motion could be detected unambiguously.

The 35% rate of false positives in the sample of 46 candidate planets came as a surprise. A model developed by Morton and Johnson2 — based on transit depths, orbital periods and a detailed model of the Milky Way’s population of binary stars in the direction of Kepler’s field of view — had predicted false­positive prob­abilities no greater than about 5% for most of the objects in the set of giant, close­in candidate planets observed by Santerne and colleagues. The authors’ high rate of false positives calls into question1,3 the validity of models of the Galactic binary population in Kepler’s direction, and the assumptions on which they are based.

This is important, because the main scientific legacy of data collected by Kepler is a two­dimensional scatter plot of derived planet radius as a function of orbital separa­tion (the distance from host star to planet). This plot contains a wealth of clues concern­ing planet formation and migration processes, and has the potential to reveal the incidence of Earth­sized planets orbiting at distances at which they could harbour liquid water on their surfaces. A detailed understanding of the stellar binary population as a function of eclipse depth and orbital period is essential to our understanding of the rate of false positives in different parts of the radius–separation plot.

Eclipsing binary stars can mimic transiting giant planets in three common ways (Fig. 1).

1. Curtis, B. A. et al. Nature 492, 59–65 (2012).2. Parfrey, L. W., Lahr, D. J. G., Knoll, A. H. & Katz, L. A.

Proc. Natl Acad. Sci. USA 108, 13624–13629 (2011).

3. Cavalier-Smith, T. Trends Plant Sci. 5, 174–182 (2000).

4. Gibbs, S. Can. J. Bot. 56, 2883–2889 (1978).5. Douglas, S. E., Murphy, C. A., Spencer, D. F. & Gray,

M. W. Nature 350, 148–151 (1991).6. Gould, S. B., Fan, E., Hempel, F., Maier, U.-G. &

Klösgen, R. B. J. Biol. Chem. 282, 30295–30302 (2007).

7. Barbrook, A. C., Howe, C. J. & Purton, S. Trends Plant Sci. 11, 101–108 (2006).

8. Moustafa, A. et al. Science 324, 1724–1726 (2009).

9. Stiller, J. W., Huang, J., Ding, Q., Tian, J. & Goodwillie, C. BMC Genomics 10, 484 (2009).

10. Woehle, C., Dagan, T., Martin, W. F. & Gould, S. B. Genome Biol. Evol. 3, 1220–1230 (2011).

11. Hazkani-Covo, E., Zeller, R. M. & Martin, W. PLoS Genet. 6, e1000834 (2010).

12. Ricchetti, M., Tekaia, F. & Dujon, B. PLoS Biol. 2, e273 (2004).

4 8 | N A T U R E | V O L 4 9 2 | 6 D E C E M B E R 2 0 1 2

NEWS & VIEWSRESEARCH

© 2012 Macmillan Publishers Limited. All rights reserved

50 Years AgoNatural History of Infectious Diseases. By Sir MacFarlane Burnet — It is a pleasure to review this, the third edition, of such a well­known work, which now presents an up­to­date account of the ramifications of an important subject … Throughout it adopts a Darwinian attitude, often overstepping the realms of human pathology into all kinds of unexpected avenues … In its ambit it presents a picture of the ravages by bacteria, protozoa and viruses …He discredits the practical value of antityphoid inoculation and would rather attribute the favourable results obtained to military sanitation … The readers will find here important information about such diverse subjects as myxomatosis in rabbits, the common cold, plague, German measles, poliomyelitis, the sweating sickness of the Middle Ages and Q. fever (from personal experience). An epilogue on new diseases and the rather bleak outlook for the future finds the author in a gloomy mood in an appraisal of bacteriological warfare.From Nature 8 December 1962

100 Years AgoMr. E. G. Bryant … asks a question regarding the effect of moonlight in “turning” fish. I have lived many years in South Africa, and have encountered the same belief, that moonlight will hasten the turning bad of fish … It seems curious, at first sight, that moonlight, which has so little effect on meteorological instruments, should have this effect on fish. I have thought it probably due to insects or some low form of life which would be abroad, or be stimulated to action, on moonlit nights and not on dark nights. The action of moonlight in stimulating the rise of sap in trees is widely believed in by practical wood cutters in almost every quarter of the world.From Nature 5 December 1912

First and simplest, brown­dwarf stars and stars with masses less than 10% that of the Sun have very similar radii to those of gas­giant planets such as Jupiter. As a result, planets can be dis­tinguished from such eclipsing binaries only by determining their mass. A second type of impostor is a triple­star system, in which a bright single star dilutes the light of a nearby, dimmer, eclipsing binary pair to the point at which the eclipses seem to be as shallow as those caused by planets. A third possibility is grazing binary systems, in which the stars’ disks overlap by only a tiny amount at each eclipse, so that the reduction in brightness is similarly small.

Some of these options can be easily elimi­nated. Kepler’s high­precision imaging delin­eates the shape of a planetary transit clearly. The light from the host star changes rapidly when the planet (which is generally much smaller than the star) is moving on or off the edge of the star’s disk, giving a characteristic ‘U­shaped’ light variation. Grazing stellar binaries have ‘V­shaped’ profiles. The dura­tion of the transit is also a clue. It gives a direct dynamical measure of the host star’s bulk density, which is closely related to its tempera­ture. Therefore, a measurement of the density that is inconsistent with that inferred from the star’s temperature is a good indicator that

the system could be an impostor.This and other validation tests are vital for

confirming the planetary nature of Kepler candidates, and do not require expensive fol­low­up observations — they are simply based on stellar properties determined from existing ground­based surveys, and on the Kepler data.

For single­planet systems, confirmation requires follow­up measurements of the orbital reflex motion caused by the orbiting planet. But this approach is expensive because it requires a large amount of ground­based tele scope time. The vast majority of Kepler small­planet candidates are simply too faint to be confirmed in this way. Santerne and col­leagues’ result, which is based on such follow­up measurements of giant­planet candidates, comes as a salutary lesson on the importance of checking, wherever possible, that indi­vidual Kepler candidates are not impostors. It also serves as a good example of the pains­taking way in which science is advanced. The problem with Morton and Johnson’s model was traced3 quickly to an inadequate treat­ment of the population of grazing binaries. This has already led Morton to develop an improved validation method3, which repro­duces the rate of false positives in Santerne and colleagues’ sample more reliably, by tak­ing better account of the shape of the Kepler

Planet

Star

Blended stellarbinaries

Grazing stellarbinaries

a

c

Brown dwarf orlow-mass star

b

d

Figure 1 | Mimicking a planetary transit. a, A gas­giant planet blocks a small amount of starlight as it passes in front of its host star. The resulting drop in light is similar to that produced by other systems, as follows: b, an orbiting brown­dwarf or low­mass star, both of which have radii similar to gas­giant planets; c, blended stellar binaries in a triple­star system that have deep eclipses strongly diluted by a bright neighbouring star, mimicking the much shallower transits of a planet; d, grazing binary stars, in which the stars’ disks overlap by only a tiny amount at each eclipse. The latter systems are the most common type of ‘impostor’ in Santerne and colleagues’ sample of candidate planets1.

6 D E C E M B E R 2 0 1 2 | V O L 4 9 2 | N A T U R E | 4 9

NEWS & VIEWS RESEARCH

© 2012 Macmillan Publishers Limited. All rights reserved

B I O C H E M I S T R Y

Another aspect of nature’s ingenuityEyewitnesses are sometimes asked to identify a culprit from a line-up of people associated with a crime scene. An enzyme — iridoid synthase — that catalyses an unusual reaction has been identified by a similar approach. See Letter p.138

planetary­transit profile as well as the transit depth and orbital period.

We should not, however, expect the story to end here. Estimates of the true fraction of planetary systems among Kepler candidates over the full range of transit depths and orbital periods are only as good as our understanding

of the binary­star population. Wherever we have the opportunity to test the predictions of the currently favoured Galactic models, we should do so. ■

Andrew Collier Cameron is at the SUPA School of Physics and Astronomy, University of

J O E C H A P P E L L

Compounds known as iridoids are com­mon in plants1, having vital roles in the interactions between a plant and

its environment — for example, offering pro­tection against herbivore attack or microbial pathogens. Perhaps the most widely encoun­tered iridoid is nepetalactone, which mimics feline sex hormones and was first isolated from the catnip plant. Iridoids also form the basis of a diverse array of drugs obtained from natural sources, but, surprisingly, the mechanism responsible for their biosyn­thesis has remained elusive. On page 138 of this issue, Geu­Flores et al.2 identify iridoid synthase, the enzyme responsible for the formation of iridoids. The discovery opens the door to the synthesis of a wide range of iridoid derivatives and related compounds through manipulation of the naturally occurring biosynthetic machinery*.

Iridoids belong to the monoterpene family of chemicals, which themselves constitute a family within the much larger constellation of terpenes. Terpenes arise from isoprene, a com­mon biosynthetic building block that contains five carbon atoms. Multiple isoprene units link together to yield chemical classes such as the monoterpenes (which contain 10 car­bons), sesquiterpenes (15 carbons), diterpenes (20 carbons) and so on.

As with many other classes of natural product, the impressive structural diversity of the terpene family depends on how the carbon atoms in biosynthetic precursors can be rearranged to form new three­dimensional configurations. The mechanisms involved in the formation of molecular scaffolds and in carbon­atom positioning for terpenes have been dominated by the ‘biogenic isoprene

rule’3 — the idea, which won its discoverer Leopold Ruzicka a Nobel prize, that all ter­penes are assembled from chains of isoprene units in reactions that include cationic inter­mediates called carbo cations. This concept led to the discovery and characterization of many terpene cyclase enzymes4,5 (now more appro­priately termed terpene synthases). These enzymes initiate cyclization reactions, in which

*This article and the paper under discussion2 were published online on 21 November 2012.

St Andrews, St Andrews KY16 9SS, UK. e-mail: acc4@st-andrews.ac.uk

1. Santerne, A. et al. Astron. Astrophys. 545, A76 (2012).2. Morton, T. D. & Johnson, J. A. Astrophys. J. 738, 170

(2011).3. Morton, T. D. Preprint at http://arxiv.org/

abs/1206.1568 (2012).

OPP

OPP

Geranyldiphosphate

(4S)-Limonene

OH

O

O

10-OxogeranialGeraniol

OH

O

cis-trans-Nepetalactol

NADPH orNADH

Oxidationsteps

Rearrangement

Carbocation

Loss of H+ Several steps

Rearrangedcarbocation

Iridoidsynthase

Classic synthesis (monoterpene synthase)

Iridoid synthesis

a

b

H

H

Figure 1 | Mechanisms of monoterpene biosynthesis. Monoterpenes are naturally occurring compounds derived from geranyl diphosphate (GPP); OPP is a diphosphate group. a, Monoterpene biosynthesis usually involves positively charged intermediates known as carbocations. Here, a carbocation forms from GPP in several steps (broken bonds in the carbocation indicate charge delocalization), then rearranges to form a new carbocation, which loses a H+ ion to form (4S)­limonene as the final product. All the steps are catalysed by a single monoterpene synthase enzyme. The wedge­shaped bonds project above the plane of the page. b, The biosynthesis of iridoid monoterpenes, such as cis-trans­nepetalactol, does not involve carbocations. Instead, GPP is first converted to geraniol and then to 10­oxogeranial, the direct precursor of iridoids. Geu­Flores et al.2 have identified the enzyme that converts 10­oxogeranial to cis-trans­nepetalactol (reaction shown in red), a process that requires the naturally occurring reducing agent NADPH, or its analogue NADH. The wavy line in cis-trans­nepetalactol indicates that the compound forms as a mixture of isomers in which the bond points either above or below the plane of the page; dashed bonds project below the plane of the page.

linear precursor molecules are converted into ring­containing products through the gen­eration of reactive carbo cations (Fig. 1a). The carbocation intermediates also drive myriad complex reactions that lead to hundreds of structurally diverse scaffolds within each of the terpene classes.

But something different happens with iri­doids — indeed, these compounds are some­times called irregular monoterpenes because their biosynthesis could not be deduced from conventional carbocation chemistry. Iridoids are derived from geranyl diphosphate, which is first converted to geraniol, then successively oxidized to form 10­oxogeranial, a direct precursor to iridoids (Fig. 1b). The iridoid­forming cyclization reaction had previously been studied only in plant extracts6, and was found to depend on NADPH, a naturally occurring reducing agent.

To identify the iridoid synthase enzyme responsible for the cyclization reaction, Geu­Flores et al. examined data on the full

5 0 | N A T U R E | V O L 4 9 2 | 6 D E C E M B E R 2 0 1 2

NEWS & VIEWSRESEARCH

© 2012 Macmillan Publishers Limited. All rights reserved