plastron was connected to the laterally expanded carapace. Thus, an alternative interpretation is that the apparent reduction of the carapace in Odontochelys resulted from lack of ossification of some of its dermal components, but that a carapace was indeed present.

This interpretation of Odontochelys leads us to the possibility that its shell morphology is not primitive, but is instead a specialized adap-tation. Reduction of dermal components of the shell in aquatic turtles is common: soft-shelled turtles have a greatly reduced bony shell and have lost the dermal peripheral elements of the carapace. Sea turtles and snapping turtles have greatly reduced ossification of the dermal com-ponents of the carapace, a condition similar to that seen in Odontochelys.

From the geological context of their fossils, Li et al.1 conclude that Odontochelys lived in a shallow marine environment. That, combined

with similarities between its carapace and the reduced shells of modern aquatic turtles, leads us to propose that the absence of most of the dermal carapace in Odontochelys is a secondary loss associated with aquatic habits rather than a primitive condition, as inferred by Li and col-leagues. Given the similarities between its shell morphology and early growth stages in living turtles, a simple truncation of carapace ossi-fication, in which the adults retained juvenile features (paedomorphosis), could have been a developmental mechanism in the evolution of the reduced carapace.

Regardless of the primitive or derived nature of its shell, Odontochelys is in evolutionary terms the most ‘basal’ turtle yet found. Its dis-covery opens a new chapter in the study of the origins and early history of these fascinating reptiles. Both interpretations alter our views of turtle evolution: Odontochelys either represents the primitive ecology for turtles, consistent with the hypothesis that the turtles’ shell evolved in aquatic environments7, or it represents the earliest turtle radiation from terrestrial environments into marine habitats. Either way, these ancient turtles demonstrate yet again the value of new fossil discoveries in changing our understanding of vertebrate history. ■ Robert R. Reisz and Jason J. Head are in the Department of Biology, University of Toronto Mississauga, Mississauga, Ontario L5L 1C6, Canada.e-mail: robert.reisz@utoronto.ca

1. Li, C., Wu, X.-C., Rieppel, O., Wang, L.-T. & Zhao, L.-J. Nature 456, 497–501 (2008).

2. Gaffney, E. S. & Meylan, P. A. in The Phylogeny and Classification of the Tetrapods Vol. 1: Amphibians, Reptiles, Birds (ed. Benton, M. J.) 157–219 (Clarendon, 1988).

3. Lee, M. S. Y. Science 261, 1716–1720 (1993).4. Joyce, W. G. & Gauthier, J. A. Proc. R. Soc. Lond. B 271, 1–5

(2003).5. Scheyer, T. M. & Sander, P. M. Proc. R. Soc. Lond. B 274,

1885–1893 (2007). 6. Hedges, S. B. & Poling, L. L. Science 283, 998–1001 (1999).7. Rieppel, O. & Reisz, R. R. Annu. Rev. Ecol. Syst. 30, 1–22 (1999).8. Gilbert, S. F. et al. Evol. Dev. 3, 47–58 (2001).

and its close relatives from the Late Triassic of Argentina, Germany and North America, it has been suggested that the earliest turtles lived in terrestrial environments4,5.

The discovery and description of Odonto-chelys by Li et al.1 challenges these hypotheses. Odontochelys is not only the oldest recogniza-ble turtle, but its skull also shows that it is more primitive than other turtles because it retains a full complement of marginal teeth, rather than a beak, and also possesses free sacral ribs and a long tail. Its osteology contradicts the view that turtles have pareiasaurian affinities, and, along with molecular data, supports evolution-ary hypotheses that they are closely related to another group, the diapsid reptiles6,7 . Li et al. argue that Odontochelys represents an early stage in the evolution of the turtle shell because the plastron is present and fully developed, but the carapace is apparently absent, with only dorsal ribs and neural (midline) dermal ossi-fications present. The authors infer, therefore, that the plastron evolved before the carapace, reflecting the timing of shell ossification dur-ing embryonic development in living turtles.

Although this evolutionary scenario is plausible, we are particularly excited by an alternative interpretation and its evolutionary consequences. We interpret the condition seen in Odontochelys differently — that a carapace was present, but some of its dermal compo-nents were not ossified. The carapace forms during embryonic development when the dorsal ribs grow laterally into a structure called the carapacial ridge, a thickened ectodermal layer unique to turtles8. The presence of long, expanded ribs, a component of the carapace of all turtles, indicates that the controlling devel-opmental tissue responsible for the formation of the turtle carapace was already present in Odontochelys. The expanded lateral bridge that connects the plastron to the carapace in other turtles is also present, implying that the

Figure 1 | Differences in turtle ecology. There are about 300 extant species of turtle, which vary in weight from about 100 grams to about 860 kilograms, and whose habitat ranges from wet meadows to deserts, and from rivers and lakes to the open ocean. Top, a fully terrestrial turtle (Gopherus agassizii), representing a similar ecology to that of the primitive turtle Proganochelys and its relatives inferred in previous studies4,5. Bottom, an aquatic turtle (Apalone spinifera), representing a potentially similar ecology to that of the fossil species Odontochelys semitestacea described by Li and colleagues1.

ORGANIC CHEMISTRY

Short cuts to complexityAndré B. Charette

The credit crunch is forcing people to tighten their belts, but chemists have long known the benefits of being economical with atoms. The latest synthesis of an anticancer agent shows how effective parsimony can be.

Nature produces an almost infinite number of structurally complex organic compounds that have fascinating — and potentially use-ful — biological properties. This has inspired generations of synthetic chemists to make not only the naturally occurring compounds, but also structurally modified analogues that have tailored properties and functions. Marine organ-isms are a particularly rich source of natural

products, but so far only one such class of compound has entered clinical trials: bryostat-ins, which have anticancer activity in vivo1. Reporting in this issue (page 485), Trost and Dong2 describe a synthetic route to a particular bryostatin — bryostatin 16 — that drastically reduces the longest sequence of consecutive steps from the previous best of 40 down to a much more manageable 26. This might open

R. R

. REI

SZ

451

NATURE|Vol 456|27 November 2008 NEWS & VIEWS

50 &

100

YEA

RS A

GO

could serve as a precursor to produce many other bryostatins. Their strategy relied on coupling three fragments of the molecule at a late stage in the synthesis.

They made the B ring using a spectacular ruthenium-catalysed coupling reaction8 between an alkene and an alkyne (a compound that con-tains a carbon–carbon triple bond). Although both fragments contain an array of potentially reactive chemical groups, the ruthenium catalyst binds selectively to the alkene and the alkyne, so inducing the formation of a new carbon–carbon bond (Fig. 2b). Such chemoselectivity is essential to obtaining efficient synthetic routes because it avoids the use of protecting groups — chemical groups that are attached temporarily to reactive parts of molecules to prevent them from inter-fering in desired reactions. Not only does the ruthenium-catalysed reaction form the B ring with the correct three-dimensional arrange-ment of substituents, it also controls the geom-etry of the potentially troublesome exo-cyclic alkene. Furthermore, the reaction is perfectly atom-economical: all of the atoms found in the starting materials (bar the catalyst) are present in the coupled product.

Trost and Dong2 prepared the C ring in a two-step process involving two different transition-metal catalysts. The first step is a palladium-catalysed coupling9 of two alkynes (Fig. 2c), which not only sets the scene for the formation of the C ring, but also forms the characteristic large ring present in bryostat-ins. This reaction is remarkable, both because it is perfectly atom-economical and because it is the first demonstration that such a large ring structure (containing 22 atoms) can be made using this kind of carbon–carbon bond-forming reaction. Such ‘macrocyclizations’ are notoriously problematic, so Trost and Dong have effectively added a useful entry to the existing roster of possible reactions.

The authors completed the synthesis of the C ring using a gold-catalysed reaction to form a carbon–oxygen bond between an alcohol group (OH) and the alkyne produced in the previous step (Fig. 2d). Gold catalysts activate alkynes selectively10 in the presence of a wide variety of other chemical groups, a use-ful property that is used here to great effect.

the door to a practical process for preparing the compound, and provide ready access to bryostatin analogues for drug discovery.

The bryostatin family of compounds is derived from the bryozoan Bugula neritina3 (Fig. 1), and encompasses more than 20 struc-turally related natural products. Not only do these compounds exhibit anticancer properties, but they have also been shown to improve cog-nition and enhance memory in animals4, mak-ing them interesting leads for drug-discovery efforts targeting Alzheimer’s disease. But the concentration of bryostatins in B. neritina is low, so extraction from bryozoans is not a via-ble means of producing these compounds — at least, not in quantities that would allow a com-plete evaluation of their biological profiles.

Bryostatins have therefore long been synthetic targets of choice for chemists — not only because of the need to find a practical way of making them, but also because the struc-tural complexity of the compounds provides a perfect opportunity to test new synthetic methods. Despite intensive efforts, only three total syntheses of bryostatins have been reported5–7. These achievements were rightly hailed as landmarks in organic chemistry, but they required lengthy sequences of consecu-tive synthetic steps (at least 40), making them impractical for the preparation of more than milligram quantities of material.

So what is it that makes bryostatins such interesting synthetic targets? A rule of thumb in synthetic chemistry is that the more chemical groups are squeezed into a small molecule, the more difficult that molecule will be to prepare. In this respect, the core structure of bryostatin features three rings (designated A, B and C; Fig. 2a) that are densely populated with chemi-cal groups. Furthermore, two of these rings, B and C, contain motifs known as exo-cyclic tri-substituted alkenes (alkenes are distin guished by carbon–carbon double bonds). In the pre-vious syntheses of bryostatins5–7, controlling the geometry of the groups attached to these alkenes was a truly formidable challenge, for which ingenious solutions had to be invented. Equally difficult to prepare, on the basis of the experiences of the previous syntheses, is another alkene group in the molecule, which forms part of the linker between rings B and C.

Multi-step syntheses of complex molecular targets will be efficient and practical only if as many of the required reactions as possible are ‘atom economical’. In an optimal atom-economical process, all of the atoms in the reactants end up in the desired product. This idea is one of twelve accepted criteria used to quan-tify how environmentally friendly chemical reactions are, and has been fully implemented by Trost and Dong2 in their synthesis of bryostatin 16. The authors deliberately chose this molecule as their target because it

Figure 1 | The natural source of bryostatins. Approximately 1 tonne of the marine organism Bugula neritina is needed to isolate 1 gram of bryostatin anticancer agents.

50 YEARS AGOIn an address to the School Broadcasting Council on November 7, Sir Ian Jacob, director-general of the British Broadcasting Corporation, referred particularly to the dependence of Great Britain, and possibly its survival, upon the widest possible diffusion of scientific skills and knowledge. He suggested that the British Broadcasting Corporation might be able to make, in its own way, “a new and massive contribution to the understanding of science in the secondary modern schools, where the need perhaps is greatest and where the shortage of good teachers is likely to be most acute”. From Nature 29 November 1958.

100 YEARS AGOI have just acquired for the Canterbury Museum the skeleton of a huge blue whale (Balaenoptera sibbaldii). The whale was cast on to the beach at Okarito, on the west coast of the South Island of New Zealand, early this year, and measured 87 feet in length. My statement that the Okarito whale is among the largest known has been freely challenged in the local Press … I have naturally sought information as to the length of skeletons of great whales preserved in museums, but have been unable to obtain satisfactory data. I shall be pleased, therefore, if directors of museums possessing the skeletons of large whales will kindly communicate with me direct, or, as the matter is one of general interest, through the medium of Nature.ALSO:Students of the occult will welcome the elaborate paper by Dr. W. L. Hildburgh in the current issue of the Journal of the Royal Anthropological Institute on Sinhalese magic. He illustrates with copious detail the equipment of the magician, devil-dancer, and astrologist, describes their methods, and provides an ample supply of curious charms, amulets, and horoscopes. He does not enter upon the question of the origin of this system of magic.From Nature 26 November 1908.

SOU

TH

EAST

ERN

REG

ION

AL

TAX

ON

OM

IC C

ENT

ER/

SOU

TH

CA

ROLI

NA

DEP

T N

AT.

RES

.

452

NATURE|Vol 456|27 November 2008NEWS & VIEWS

The authors’ reaction could, in principle, yield two products, either the desired C ring that contains six atoms, or a smaller ring made up of five atoms. Indeed, when the authors tried a palladium catalyst for the reaction, a mixture of the two rings formed. But the gold catalyst provided the six-membered ring alone.

Trost and Dong’s synthesis2 of bryostatin 16, by far the shortest to date, is a remark-able achievement. Nevertheless, more atom-economical and chemoselective methods are still needed to generate complex molecular targets with minimal effort. In particular, the

development of chemoselective reactions is pivotal to reducing the need for temporary pro-tecting groups, because every protecting group used in a synthesis generally adds two steps to the route — one to put the group on, and another to take it off. By drastically reducing the total number of steps previously required to make bryostatins, Trost and Dong’s work clearly demonstrates the huge difference that chemoselective, atom-economical strategies can make. ■

André B. Charette is in the Department of Chemistry, University of Montreal,

Montreal, Quebec H3C 3J7, Canada.e-mail: andre.charette@umontreal.ca

1. Banerjee, S. et al. J. Nat. Prod. 71, 492–496 (2008).2. Trost, B. M. & Dong, G. Nature 456, 485–488 (2008).3. Pettit, G. R. J. Nat. Prod. 59, 812–821 (1996).4. Hongpaisan, J. & Alkon, D. L. Proc. Natl Acad. Sci. USA 104,

19571–19576 (2007).5. Kageyama, M. et al. J. Am. Chem. Soc. 112, 7407–7408 (1990).6. Evans, D. A. et al. J. Am. Chem. Soc. 121, 7540–7552 (1999).7. Ohmori, K. et al. Angew. Chem. Int. Edn 39, 2290–2294

(2000).8. Trost, B. M. et al. Org. Lett. 7, 4761–4764 (2005).9. Trost, B. M. & Frontier, A. J. J. Am. Chem. Soc. 122,

11727–11728 (2000).10. Gorin, D. J. & Toste, F. D. Nature 446, 395–403 (2007).

OMeO2C

CO2Me

MeMe

Me OH

O

O

OH

O

Me MeHO

Bryostatin 16

AB

a b

O

R1

R3

R2

OH+Ruthenium

catalyst

O

R2

R3

c

O

R1

B

Palladiumcatalyst

CO2Me

+

CO2Me

R4R4

H

OH

R5HO R5

Goldcatalyst

O

CO2Me

R4 R5

C

d

O

OMe

MeMe

H H

OC

MOLECULAR BIOLOGY

The Bloom’s complex mousetrap Robert M. Brosh Jr

Genomic instability often underlies cancer. Analyses of proteins implicated in a cancer-predisposing condition called Bloom’s syndrome illustrate the intricacies of protein interactions that ensure genomic stability.

Bloom’s syndrome, which is characterized by severe growth retardation, immunodeficiency, anaemia, reduced fertility and predisposition to cancer1, is caused by mutations in the gene BLM. At the cellular level, the hallmark of this genetic disorder is a high rate of sister-chro-matid exchange — the swapping of homolo-gous stretches of DNA between a chromosome and its identical copy generated during DNA replication2. Understanding how mutations in BLM lead to this chromosomal abnormal-ity has been of considerable interest to both scientists and clinicians. So the latest clue to solving the mystery of Bloom’s syndrome, which Xu et al.3 and Singh et al.4 report in Genes & Development, is a welcome advance.

The product of BLM, the enzyme BLM helicase, functions as part of a protein com-plex of the same name, which is involved in both suppression of sister-chromatid exchange

and maintenance of genomic stability. Besides this helicase, which unwinds complementary DNA double strands, this complex was thought to contain three other protein components: a topo isomerase (Topo 3α) enzyme, which unknots the two DNA strands by introduc-ing transient nicks; RPA, which binds tightly to single-stranded DNA; and RMI1, which binds directly to BLM and Topo 3α, and perhaps less tightly to other factors.

Through its combined nicking and unwind-ing activities, the BLM complex catalyses the splitting (resolution) of a double-cross-shaped DNA structure called a double Holliday junction that arises from reciprocal exchanges of single strands of DNA between homolo-gous double-stranded sequences during the pro cess of homologous recombination (Fig. 1, overleaf). Thus, the BLM complex prevents the formation of hybrid recombinant DNA

molecules called crossovers that could lead to a phenomenon known as loss of heterozygosity, which significantly contributes to cancer.

Xu et al.3 and Singh et al.4 independently iden-tify a fifth component of the BLM complex: a small protein designated RMI2. Strikingly, both teams find that RMI2 is required to maintain the stability of the BLM complex, and that its defi-ciency in vertebrate cells results in chromosomal instability. RMI2 and BLM seem to function in the same pathway to suppress sister-chromatid exchange3. Moreover, RMI2-depleted cells are sensitive to DNA damage that stalls the pro-cess of replication, and this protein is essential for efficient targeting of BLM to chromatin (complexes of DNA and histone proteins) and to nuclear foci during replicational stress4.

How does RMI2 stabilize and orchestrate the activity of the BLM complex? The answer is not simple and awaits further investigation. For starters, the current studies3,4 suggest that RMI2 functions by interacting directly with RMI1, which, in turn, binds to BLM and Topo 3α. It is possible, however, that — in addition to the members of the BLM complex — RMI2 also binds less tightly to other protein complexes in the cell that survey and repair the genome, or to peculiar DNA configurations that arise at stalled or converging replication forks (struc-tures formed by the separation of two comple-mentary DNA strands during replication).

Figure 2 | Catalytic steps in the synthesis of bryostatin 16. a, Bryostatin 16 has been prepared in a remarkably short synthesis by Trost and Dong2. The molecule contains several rings, including three designated A, B and C. The large ring highlighted in purple was constructed in the palladium-catalysed reaction shown in c. The structure also has two exo-cyclic trisubstituted alkenes (circled in red) and a disubstituted trans-alkene (circled in blue) that are difficult to prepare. The authors used three

pivotal transition-metal-catalysed steps in their synthesis. b, A ruthenium-catalysed reaction was used to construct the B ring. c, d, The C ring was made in two steps: a palladium-catalysed reaction (c) followed by a gold-catalysed reaction (d). All three reactions are perfectly atom economical — all of the atoms in the starting materials are present in the products. Me represents a methyl group; R1 to R5 represent fragments of bryostatin (or of synthetic precursors to bryostatin).

453

NATURE|Vol 456|27 November 2008 NEWS & VIEWS