Bringing It Together with RNA Kate Thodey, et al. Science 333, 412 (2011); DOI: 10.1126/science.

1209685

This copy is for your personal, non-commercial use only.

If you wish to distribute this article to others, you can order high-quality copies for your colleagues, clients, or customers by clicking here. Permission to republish or repurpose articles or portions of articles can be obtained by following the guidelines here. The following resources related to this article are available online at www.sciencemag.org (this infomation is current as of October 12, 2011 ): Updated information and services, including high-resolution figures, can be found in the online version of this article at: http://www.sciencemag.org/content/333/6041/412.full.html A list of selected additional articles on the Science Web sites related to this article can be found at: http://www.sciencemag.org/content/333/6041/412.full.html#related This article cites 13 articles, 5 of which can be accessed free: http://www.sciencemag.org/content/333/6041/412.full.html#ref-list-1 This article appears in the following subject collections: Cell Biology http://www.sciencemag.org/cgi/collection/cell_biol

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright 2011 by the American Association for the Advancement of Science; all rights reserved. The title Science is a registered trademark of AAAS.

Downloaded from www.sciencemag.org on October 12, 2011

PERSPECTIVES
techniques may waste most of their computational time needlessly wandering around suboptimal terraces. This is an important issue, because missing information, and thus terraces, are likely to be common in the increasingly large supermatrices produced by next-generation sequencing. However, the authors draw on past studies to provide several suggestions for improving inference under these conditions. First, they suggest that all trees on a terrace can be enumerated quickly once one tree on the terrace is found. The collection of trees within a terrace need not be explicitly evaluated once one member has been identied; rather, to nd a better path to optimality, it is important to identify those few terrace trees that are connected to other high-scoring trees that stand outside the terrace. Although precisely how to rigorously implement this strategy is an open question, it clearly has the potential to provide substantial improvements in computational efciency. A second important result is the development of a procedure that helps investigators identify the missing data that is primarily responsible for generating large terraces, thus providing a roadmap to where researchers should focus future sequencing efforts to ll these gaps and achieve the highest possible level of phylogenetic resolution. Such roadmaps are often lacking in studies in which it is impossible to have complete sequence information for both a large number of different taxa and many loci. Missing data will create terraces in data sets governed by two assumptions: First, that the partitions of the supermatrix dened by individual loci are allowed to follow their own model of evolution; and, second, that all loci share a common phylogenetic history. The second assumption is an important limitation of Sanderson et al.s approach, which assumes that concatenation of data from different loci is appropriate. However, certain evolutionary processessuch as horizontal transfer and incomplete lineage sorting (4) can generate discord among the true phylogenetic histories across loci. In these situations, the most appropriate methods of analysis are those that explicitly model the sources of this variation. Whether terraces will appear in these models is an open question. Sanderson et al. have improved our understanding of the landscape of optimality scores for phylogenetic trees. If researchers can effectively translate this improved understanding into innovations in phylogenetic software, we may well realize substantial computational improvements in inferring the tree of life.
1. M. J. Sanderson et al., Science 333, 448 (2011). 2. K. Meusemann et al., Mol. Biol. Evol. 27, 2451 (2010). 3. Y. Bouchenak-Khelladi et al., Mol. Phylogenet. Evol. 47, 488 (2008). 4. L. S. Kubatko, J. H. Degnan, Syst. Biol. 56, 17 (2007). 10.1126/science.1209690

CELL BIOLOGY

Bringing It Together with RNA

Kate Thodey and Christina D. Smolke

Synthetic RNA nanostructures can direct biochemical pathways in cells.

Department of Bioengineering, Stanford University, Stanford, CA 943054201, USA. E-mail: csmolke@stanford.edu

Self-assembly required. Graphic depiction of bacteria expressing one-dimensional RNA bers (bottom) or two-dimensional RNA nanostructures (top) in vivo. RNA nanostructures may be used to spatially organize proteins and metabolic pathways.

412

22 JULY 2011 VOL 333 SCIENCE www.sciencemag.org

Published by AAAS

CREDIT: KRISTA SHAPTON/BLOT MEDIA AND FAISAL ALDAYE

he biochemical reactions that underlie life have evolved to be spatially organized within cells. This organization ranges from the simple colocalization of molecules, such as an enzyme and its substrate, to the compartmentalization of metabolic pathways within organelles. There is growing interest in mimicking this spatial organization to engineer synthetic pathways in microbes for producing drugs, fuels, and commodity chemicals. On page 470 of this issue, Delebecque et al. (1) describe the design and construction of self-assembling RNA scaffolds that spatially organize enzymes in bacterial cells. In doing so, they advance nucleic acid nanotechnology into a cellular environment and supply metabolic engineers with a new tool for structurally optimizing biosynthesis processes in cells. The compartmentalization of metabolic pathways allows efcient channeling of substrates to products over several enzymatic steps by limiting the diffusion of intermediates (2). For example, the eukaryotic mitochondrion concentrates biochemical reactions within an endomembrane-bound

organelle to optimize energy production. This level of organization is relatively lacking in prokaryotic cells. But there are exceptions, particularly in photosynthetic bacteria (3), in which light-harvesting protein complexes are incorporated in thylakoid

membranes and carbon fixation is further organized in protein-bound compartments (carboxysomes) in the cytoplasm (4). Synthetic biologists have limited tools to engineer the spatial organization of enzymes in host cells. Protein scaffolds can be designed to colocalize enzymes through interactions between binding domains on the scaffold and target peptides fused to each enzyme (5). However, the complexity of architectures that can be achieved by this approach is limited. By contrast, DNA can be designed to self-assemble in vitro into many and varied nanostructures. This technology exploits the predictable base-pairing properties of DNA to program the folding of one or several strands into two-dimensional (2D) shapes and arrays and 3D balls, boxes, and polyhedral structures (6). Two methods of assembly are the tile-based approach of constructing building blocks from DNA strands that then further assemble into arrays, or the DNA origami method of using a long single-stranded DNA molecule that can selfassemble alone or together with many shorter fragments. By introducing chemical modications or protein-binding sequences into the DNA strand, targeted proteins have been recruited in vitro to specic locations (7, 8).

Downloaded from www.sciencemag.org on October 12, 2011

References

PERSPECTIVES
However, in vivo applications of this technology have been challenging because of difculties in generating suitable single-stranded DNA molecules within cells and in controlling the self-assembly reactions within the complex cellular environment. Delebecque et al. provide the rst example of nucleic acid nanostructures assembled within a cell and used to spatially organize proteins. The authors designed RNA strands that selectively assemble into tiles and include single-stranded overhangs to facilitate their further self-assembly into nanotubes and nanoarrays (see the gure). These nanostructures assembled within bacteria using RNA transcribed by the host cells, and were visualized by transmission electron microscopy and atomic force microscopy. Two distinct protein-binding RNA sequences called aptamers were incorporated in the tiles to serve as protein docking sites. By fusing aptamer target proteins to each half of a split uorescent protein, the authors detected protein-protein interactions mediated by the RNA scaffold as uorescence in live cells. Delebecque et al. demonstrate an exciting role for these nanostructures in engineered biosynthetic pathways. Bacterial [FeFe]-hydrogenases catalyze the production of hydrogen by the reduction of two protons. To perform this reaction they require electrons supplied by a ferredoxin redox protein. The interaction between a [Fe-Fe]-hydrogenase and its partner ferredoxin has been engineered in bacteria to increase hydrogen production by altering protein-protein interaction domains, direct fusion of the proteins, or colocalization through a protein scaffold (9). Delebecque et al. boosted hydrogen production in bacteria by fusing these two hydrogen synthesis enzymes to proteins that bind to engineered docking sites on the RNA nanostructures. Bacterial strains with RNA-scaffolded enzymes produced up to 48 times more hydrogen than those with unbound recombinant enzymes; the nanoarray conformation facilitated more hydrogen production over the nanotube design. Delebecque et al. reinforce the notion that scaffolding is a promising method for optimizing biosynthetic pathways. However, there remain several common limitations in the application of protein and RNA scaffolds. The fusion of peptide or protein tags to enzymes for scaffolding can decrease the activity of the tagged enzymes (5), although this decrease can be more than recovered by the overall increase in product output achieved by the scaffold. It is tempting to assume that RNA scaffolding acts to increase hydrogen production by facilitating sustained interactions between hydrogenase and ferredoxin. However, this increased production could partly be attributed to other factors such as the slight increase in total protein accumulation observed in cells with scaffolds. RNA scaffolds may also have negative effects on cell health and contribute to cellular stress due to nucleotide depletion or disruption of cellular functions caused by the rapid growth of these nanostructures throughout the cytoplasm. It is also unlikely that the RNA scaffolds in the design of Delebecque et al. can be transferred directly to eukaryote host cells, where methods will be needed to export transcribed RNAs from the nucleus to the cytoplasm before self-assembly into programmed nanostructures occurs. Additional concerns associated with RNA processing and degradation pathways, including RNA interference, may introduce challenges in translating RNA scaffolding strategies to higher organisms. RNA has remarkable properties for designing cellular information processing and control functions (10, 11). Future applications of RNA scaffolds may take advantage of the ease with which aptamers can be generated (12, 13) to incorporate many protein docking sites and thereby spatially organize more complex pathways. The coupling of RNA scaffolding technology with ligand-regulated RNA controllers (RNA molecules that control functions such as gene expression) (14) will allow the assembly and disassembly of scaffolds in response to metabolic ux or environmental signals. Ultimately, the integration of spatial organization and dynamic control strategies will advance the design of more sophisticated biological systems.
1. C. J. Delebecque, A. B. Lindner, P. A. Silver, F. A. Aldaye, Science 333, 470 (2011); 10.1126/science.1206938. 2. P. A. Srere, K. Mosbach, Annu. Rev. Microbiol. 28, 61 (1974). 3. J. F. Stolz, Bacterial Intracellular Membranes (Embryonic Encyclopedia of Life Sciences, Nature Publishing Group, London, 2007). 4. T. O. Yeates, C. A. Kerfeld, S. Heinhorst, G. C. Cannon, J. M. Shively, Nat. Rev. Microbiol. 6, 681 (2008). 5. J. E. Dueber et al., Nat. Biotechnol. 27, 753 (2009). 6. N. C. Seeman, Annu. Rev. Biochem. 79, 65 (2010). 7. Y. Liu, C. Lin, H. Li, H. Yan, Angew. Chem. Int. Ed. Engl. 44, 4333 (2005). 8. H. Yan, S. H. Park, G. Finkelstein, J. H. Reif, T. H. LaBean, Science 301, 1882 (2003). 9. C. M. Agapakis et al., J. Biol. Eng. 4, 3 (2010). 10. S. J. Culler, K. G. Hoff, C. D. Smolke, Science 330, 1251 (2010). 11. M. N. Win, C. D. Smolke, Science 322, 456 (2008). 12. A. D. Ellington, J. W. Szostak, Nature 346, 818 (1990). 13. C. Tuerk, L. Gold, Science 249, 505 (1990). 14. M. N. Win, J. C. Liang, C. D. Smolke, Chem. Biol. 16, 298 (2009). 15. C.D.S. is supported by funds from the NIH, NSF, and Defense Advanced Research Projects Agency. 10.1126/science.1209685

References and Notes

GEOCHEMISTRY

Onset of Plate Tectonics

Martin J. Van Kranendonk Analysis of diamonds from the subcontinental mantle reveals that plate tectonics started 3 billion years ago.

he further back we look into the geological past, the more obscured the view, masked by an increasingly fragmentary geological record. This has resulted in a controversy on the rates and mechanisms of early continental crust formation and whether plate tectonicsthe dominant crust-forming process over at least the past 2.5 billion yearsoperated the same way, or even at all, during early Earth history (1, 2). On page 434 of this issue, Shirey and Richardson (3) shed light on this issue by looking
Geological Survey of Western Australia, East Perth, Western Australia 6004, Australia, and School of Earth and Environment, University of Western Australia, Crawley, Western Australia 6009, Australia. E-mail: martin.vankranendonk@ dmp.wa.gov.au

at it from a new anglefrom the bottom up. They investigated the composition of a suite of minerals occurring as inclusions in diamonds dredged up by young kimberlite volcanoes. These diamonds derive from great depths (125 to 175 km) within the ancient lithospheric mantle keels that underpin the stable continental crustal regions known as cratons. Because the minerals can be precisely dated, they can provide a snapshot of the subcontinental lithospheric mantle (SCLM) composition at the time when the crust was being formed. Shirey and Richardson show that the SCLM composition changed dramatically at about 3 billion years ago (Ga); SCLM older than 3 Ga contains only peridotitic mineral

www.sciencemag.org SCIENCE VOL 333 22 JULY 2011

Published by AAAS

413

Downloaded from www.sciencemag.org on October 12, 2011