distinguishes between three different olefins in the achiral starting material, sequentially reacting only the required two, thus yielding the product essentially as a single enantiomer in high yield. This reaction could not have been achieved using previously available catalysts. Moreover, their catalyst performs the reaction in the presence of a tertiary amine (an organic base containing a nitrogen to which three hydrocarbon groups are attached) in the sub-strate, which is typically a troublesome motif in metathesis reactions.

Hoveyda and colleagues’ stereogenic-at-molybdenum complex has the potential to become an all-purpose chiral catalyst for olefin metathesis reactions. More broadly, the authors have discovered a bold new design for chiral catalysts that will inspire the development of future generations of catalysts, not only

for olefin metathesis, but also for many other catalytic reactions. ■

Steven T. Diver is in the Department of Chemistry, University at Buffalo, The State University of New York, 572 Natural Sciences Complex, Amherst, New York 14260-3000, USA.e-mail: diver@buffalo.edu

1. Malcomson, S. J., Meek, S. J., Sattely, E. S., Schrock, R. R. & Hoveyda, A. H. Nature 456, 933–937 (2008).

2. http://nobelprize.org/nobel_prizes/chemistry/laureates/2005/animation.html

3. Seiders, T. J., Ward, D. W. & Grubbs, R. H. Org. Lett. 3, 3225–3228 (2001).

4. Funk, T. W., Berlin, J. M. & Grubbs, R. H. J. Am. Chem. Soc. 128, 1840–1846 (2006).

5. Van Veldhuizen, J. J., Garber, S. B., Kingsbury, J. S. & Hoveyda, A. H. J. Am. Chem. Soc. 124, 4954–4955 (2002).

6. Alexander, J. B., La, D. S., Cefalo, D. R., Hoveyda, A. H. & Schrock, R. R. J. Am. Chem. Soc. 120, 4041–4042 (1998).

7. Fujimura, O. & Grubbs, R. H. J. Org. Chem. 63, 824–832 (1998).

EPIGENETICS

How to lose a tail Mary Ann Osley

Everyone carries some baggage they would like to lose. For the histone protein H3, that baggage is a chunk of its tail, which when clipped off affects the expression of genes with which the histone is associated.

Chromatin — the complex of DNA and histone proteins — is highly dynamic. The histone constituents of chromatin can be reversibly modified, replaced with related proteins, or cleaved by proteolytic enzymes. Two stud-ies1,2, published in Cell and Nature Structural and Molecular Biology, now show that histone modification by regulated proteolytic cleavage is required during cell differentiation in both mammals and yeast to alter gene expression.

A histone molecule is divided into two distinct regions: a carboxy terminus that organ-izes DNA on the surface of each unit of the DNA–histone complex (the nucleosome), and a flexible amino terminus dubbed the N tail, which protrudes from the nucleosome. N tails make contact with DNA and with histones in adjacent nucleosomes, thereby enabling chromatin folding. Histone proteins are also typically decorated with a large number and variety of small chemical groups, such as methyl and acetyl groups, and these modifi-cations have a central role in regulating gene transcription3. Why so many modifications, and how do they affect transcription?

The prevailing view holds that distinct patterns of N-tail modifications, or ‘marks’, provide signals for the recruitment of specific transcription factors to chromatin or stimulate the activity of proteins that either modify the chromatin further or disrupt it3,4. As a con-sequence, transcription is turned on or off, depending on whether activating or repress-ing marks are present. The corollary of this is

that, when the transcriptional status has to be reversed, the N-tail marks must be removed to allow appropriate restructuring of chromatin.

Commonly, marks are actively removed using the opposite reaction to the one that added them; for example, histone deacety-lase enzymes remove acetyl marks, which were added by histone acetyltransferases. But what happens if more than one mark must be removed simultaneously? Most enzymes engaged in histone modification show specificity — that is, they only attack single, or several, marks of the same kind. Using this strategy, each type of modification would have to be removed by a different enzyme. Duncan et al.1 and Santos-Rosa and colleagues2 describe a more dramatic way in which multiple marks are simultaneously removed from the N tail of

N C

2 4 9 14 18 27 36

ARTKQTARKSTGGKAPRKQLATKAARKSAPATGGVKKP H3

23

Figure 1 | Cleavage of the H3 amino-terminal tail. The amino-acid sequence of the mammalian histone H3 N tail is shown along with the major sites of methylation (circles) and acetylation (triangles) that have been mapped to arginine (R) and lysine (K) residues3. Two groups1,2 find that the enzyme cathepsin L in mouse embryos and a serine protease in yeast cleave H3 after alanine 21 (red line). Minor cleavage sites generated by cathepsin L are also indicated (blue lines). The orange and green symbols indicate modifications that were shown to respectively inhibit or activate the proteases when present on synthetic peptides during in vitro cleavage assays.

histone H3 in mouse and yeast cells, respec-tively. This mechanism involves enzymatic cleavage of the first 21 amino-acid residues at the N terminus of H3, thus sweeping away en masse all modifications present in this region (Fig. 1).

Despite the evolutionary conservation of this mechanism, there seems to be a striking differ-ence between mice and yeast in the endopepti-dase enzyme each organism uses for N-tail cleavage. The ‘clipper’ in mouse is cathepsin L, which belongs to a class of protease enzymes called cysteine proteases. Cathepsins are found in the cellular organelles called lysosomes, but are also present in the nucleus5. The as-yet-unidentified yeast enzyme is likely to be a ser-ine protease2. The feature these enzymes share, however, is that they are regulated: cathepsin L is induced during differentiation of embryonic stem cells in mice, whereas the activity of the yeast enzyme is triggered under conditions of nutrient deprivation that lead either to the sta-tionary phase of growth or to sporulation.

The signals that promote the activity of the mouse and yeast proteases are unknown. Both groups1,2 report, however, that the two enzymes are sensitive to the modification state of the target H3 N tail. The presence of certain marks inhibited the activity of the endopeptidases, whereas other marks increased their activity (Fig. 1). So it could be that enzymatic cleav-age of H3 is regulated by an interplay between the endopeptidases and other chromatin-modifying activities.

What are the biological consequences of H3-tail clipping? The role of H3-tail cleav-age during differentiation of embryonic stem cells is unknown, although stem cells contain a signature mark of trimethylation on lysines 4 and 27 (H3K4me3/H3K27me3) in the H3 N tail that is altered upon differentiation6. By removing this and other modifications during differentiation, H3-tail clipping could set the transcriptional state of a particular cell lineage.

At the cellular level, H3-tail clipping could simply clear all repressive marks from chromatin, thereby allowing the binding of transcription-activator complexes to the affected DNA. Indeed, the results of Santos-Rosa and colleagues2 support a role for this

885

NATURE|Vol 456|18/25 December 2008 NEWS & VIEWS

modification in regulating the selective activa-tion of transcription in yeast, as mutation of the H3 cleavage site impaired expression of several genes normally activated during sporulation or entry into the stationary phase. The authors’ work2 also hints at an additional effect of H3-tail clipping in yeast, involving regulation of nucleosome displacement. Cleavage of the H3 tail precedes loss of nucleosomes at several promoter sequences in vivo — an event that exposes promoter DNA and thus enhances the binding of transcription-activator complexes to the promoter during gene activation7. Intrigu-ingly, a trimethylation mark, K4me3, which prevents clipping by the yeast endopeptidase in vitro, is maintained in chromatin at promot-ers during gene activation8. This observation supports a mechanism in which nucleosomes that do not contain K4me3 are marked for H3 cleavage and subsequent displacement.

Apart from the unknown identity of the

serine protease that clips H3 in yeast, and the mysterious way in which clipper activities in mouse and yeast are regulated, many other questions arise from these fascinating observa-tions1,2. The first relates to the unexpected role of lysosomal proteases in chromatin activities. Do other cysteine proteases also have nuclear targets? Second, is N-tail clipping unique to H3, or are other histone proteins also cleaved at their N tails in a regulated manner? There are reports that other histone N tails are pro-teolytically removed, notably during develop-ment in the ciliate Tetrahymena, in which many striking observations relating to chromatin structure have been made9. Third, could the loss of the N tail affect gene expression not just by removing some marks but also by actively preventing formation of others on N tails? And finally, how is the cleaved H3 molecule replaced with an intact H3, and could this provide a mechanism for substituting one set

of tail marks for another? Just when it seemed that we knew how chromatin structure is regulated, previously unknown pathways emerge to keep this field of research vital and interesting. ■

Mary Ann Osley is in the Department of Molecular Genetics and Microbiology, University of New Mexico School of Medicine, Albuquerque, New Mexico 87131–0001, USA. e-mail: mosley@salud.unm.edu

1. Duncan, E. M. et al. Cell 135, 284–294 (2008).2. Santos-Rosa, H. et al. Nature Struct. Mol. Biol. doi:10.1038/

nsmb.1534 (2008).3. Kouzarides, T. Cell 128, 693–705 (2007).4. Jenuwein, T. & Allis, C. D. Science 293, 1074–1080

(2001).5. Goulet, B. et al. Mol. Cell 14, 207–219 (2004).6. Bernstein, B. E. et al. Cell 125, 315–326 (2006).7. Lee, C. K., Shibata, Y., Rao, B., Strahl, B. D. & Lieb, J. D.

Nature Genet. 36, 900–905 (2004).8. Liu, C. L. et al. PLoS Biol. 3, e328 (2005).9. Lin, R., Cook, R. G. & Allis, C. D. Genes Dev. 5, 1601–1610

(1991).

CONDENSED-MATTER PHYSICS

The eternal triangleMark Harris

The frustration that atomic interactions can undergo is not unlike that occurring when human aims are thwarted. An elegant study offers a way of visualizing the hitherto mysterious dynamics of ‘frustrated’ systems.

It is one of the great embarrassments of condensed-matter physics — but also one of its greatest strengths — that we understand so poorly the fundamental interactions inside materials. On the one hand, any elementary textbook will provide simple descriptions of such interactions (ionic, covalent and van der Waals bonds, as well as various magnetic inter-actions). But on the other hand, wonderfully unexpected and qualitatively new behaviour is invariably observed in real, experimental materials when these same interactions are allowed to mix, match and, especially, ‘fight’ each other.

The properties of ‘frustrated’ systems offer a classic example of the end result being greater than the sum of the parts — the geometry of the atomic lattice in such systems prevents the simultaneous minimization of the interaction energies between the neighbouring atoms, pre-cluding the existence of a unique ground state — and explain why they have long been at the forefront of research in the physical sciences. Many commercial materials (such as the ferrite ceramic magnets used in microelectronics) contain strongly frustrated interactions that affect their physical properties. A better under-standing of frustration might allow for more subtle ‘tuning’ of such materials to achieve desirable commercial properties.

One of the long-standing problems in this area is that of the Ising anti-ferromagnet on

the triangular lattice, the archetypal frustrated magnet. This is stunningly simple to concep-tualize but surprisingly elusive in nature, with a relative scarcity of experimental realizations. On page 898 of this issue, Han et al.1 describe experiments on a new kind of system, based on non-magnetic colloidal monolayers, that is shown to approximate the triangular Ising anti-ferromagnet in a number of ways. What’s more, its dynamics may be visualized directly in real time.

Although Han and colleagues’ work involves an entirely non-magnetic soft-matter system, its significance is best appreciated by consid-ering the way magnetic interactions behave on simple two-dimensional atomic lattices (Fig. 1). In the case of the triangular lattice, Gregory Wannier’s pioneering study2 predicted that the lowest-energy state of the system (its ground state) is disordered, and retains finite entropy even at absolute zero temperature. A closely related result3 was obtained by Philip Anderson in 1973. He considered the case of quantum spins on the triangular lattice and predicted that the ground state is a ‘spin-liquid’ state, containing strong dynamic correlations but no static order whatsoever, even down to the lowest temperatures. This tantalizing spin-liquid state has preoccupied physicists ever since, especially because it was highlighted as a possible explanation for high-temperature superconductivity.

Research has snowballed exponentially to consider many other types of geometrical frustration and spin-liquid behaviour, such as spin ice, pyrochlores, kagomé lattices and kagomé ice, to name but a few examples4–8. Recent work has also pointed to the possibil-ity of fabricating some of the classic frustrated models, for instance using nanoscale magnetic islands in thin-film form9. But so far, these techniques have not been exploited to probe the dynamics of the frustrated systems, espe-cially how these systems explore their highly degenerate ground states. One intriguing sug-gestion has been to test soft-matter colloidal systems in which the interactions between the particles can be tuned10, and the results of Han et al.1 demonstrate the enormous potential of these systems.

The experiments were performed on a system of microgel colloidal spheres that self-organize into a buckled triangular lattice. The up- or down-displacements of the spheres are closely analogous to the up- or down-spin states of the Ising anti-ferromagnet. In terms of structure and interactions then, this new sys-tem presents a tantalizing approximation to the triangular Ising anti-ferromagnet. Remarkably, the system is fully dynamic, which means that a number of poorly understood issues associated with frustrated systems can be investigated, in particular how individual particles reorganize themselves upon changes in the local environ-ment, how defects arise and travel around the lattice, and how transitions to frozen, glassy states occur.

Han and colleagues’ system comes with two further novelties. First, the interactions and dynamics of the system can be carefully fine-tuned (something that is difficult, if not impossible, with magnetic materials) through changes in the diameter of the microgel spheres. This clearly points towards the potential for investigating other statistical-mechanical

886

NATURE|Vol 456|18/25 December 2008NEWS & VIEWS