NEWS AND VIEWS Ras activation3 opens up new avenues to explore the biochemical mechanism underlying this critical immunological process. Exciting times are clearly ahead as previously unappreciated phosphatidic acid takes on new and increasingly important roles, and GTPase network diagrams are reassessed and rewired. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests.

1. Schlessinger, J. Cell 103, 211–225 (2000). 2. Zhao, C., Du, G., Frohman, M. A. & Bar-Sagi, D. Nature Cell Biol. 9, 706–712 (2007). 3. Mor, A. et al. Nature Cell Biol. 9, 713–719 (2007). 4. Hancock, J. F. Nature Rev. Mol. Cell. Biol. 7, 456–462 (2006). 5. Jacobson, K., Mouritsen, O. G. & Anderson, R. G. Nature Cell Biol. 9, 7–14 (2007). 6. Kusumi, A., Koyama-Honda, I. & Suzuki, K. Traffic 5, 213–230 (2004). 7. Di Fulvio, M., Lehman, N., Lin, X., Lopez, I. & Gomez-Cambronero, J. Oncogene 25, 3032–3040 (2006). 8. Plowman, S. J., Muncke, C., Parton, R. G. & Hancock,

J. F. Proc. Natl Acad. Sci. USA 102, 15500–15505 (2005). 9. Rizzo, M. A. et al. J. Biol. Chem. 274, 1131–1139 (1999). 10. Rizzo, M. A., Shome, K., Watkins, S. C. & Romero, G. J. Biol. Chem. 275, 23911–23918 (2000). 11. Mor, A. & Philips, M. R. Annu. Rev. Immunol. 24, 771–800 (2006). 12. Lockyer, P. J., Kupzig, S. & Cullen, P. J. Curr. Biol. 11, 981–986 (2001). 13. Welsh, C. J., Yeh, G. C. & Phang, J. M. Biochem. Biophys. Res. Commun. 202, 211–217 (1994). 14. Daniels, M. A. et al. Nature 444, 724–729 (2006).

Coding RNAs: separating the grain from the chaff Amin Ghabrial Small open-reading frames are difficult to detect both computationally and by mutagenesis. An mRNA previously thought to be non-coding has now been found to produce four tiny peptides that function non-cell autonomously to organize epithelial actin during Drosophila development.

Two previous studies of mRNA-like non-coding transcripts in Drosophila identified the locus, pncr001:3R or MRE29, as a polyA-containing RNA lacking open-reading frames of more than 100 amino acids1,2. On page 660 of this issue, Kondo et al. reveal that this mRNA codes not just for one, but four (albeit tiny) proteins3. Furthermore, they found that mutant embryos deficient for this gene lacked the ventral denticles and dorsal hairs that decorate the epidermis of wild-type embryos. Because mutant embryos had a smooth appearance similar to rice that has been milled to remove the husk and bran, the gene was renamed polished rice (pri). The pri story is of particular interest because of the novel structure of the gene — it encodes a polycistronic message that produces four peptides less than 100 amino acids in length — and because the small Pri peptides seem to function non-cell autonomously to promote changes in epithelial-cell morphology. It has become clear that eukaryotic genomes contain large numbers of transcribed sequences with no apparent coding potential4. These transcripts include thousands of polyA-containing RNAs that have only small open-reading frames (sORFs) and are often annotated as non-coding Amin Ghabrial is in the Department of Biochemistry and Howard Hughes Medical Institute, Stanford University School of Medicine, 300 Pasteur Drive, Stanford, CA 94305, USA. e-mail: [email protected]

RNAs (ncRNAs). Computational approaches based on an analysis of codon bias and crossspecies conservation suggest that approximately 5% of annotated genes in the genomes of yeasts, plants, flies, nematodes, mice and humans contain sORFs that may, in fact, be translated into exceptionally small proteins5. One example is that of the mille-pattes gene from the flour beetle Tribolium. Savard et al. reported that mille-pattes seems to encode four small peptides (10, 12, 15 and 23 amino acids) in a single polycistronic mRNA6. Fortuitously, the pri gene characterized by Kondo et al. is the Drosophila homologue of mille-patte, and they find that it has a similar gene structure, encoding four or five small peptides (three of 11 amino acids, one of 32 amino acids and a potential fifth sORF of 49 amino acids)3. The first four predicted sORFs of pri and the first three predicted sORFs of mille-pattes share a septomeric motif (LDPTGXY) present in one or two copies in each peptide. Savard et al. infer translation of four millepattes peptides from the conservation of the putative protein-coding sequences between Tribolium and Drosophila. Kondo et al. go one step further, showing that the Drosophila message can indeed be translated. They tagged each of the five putative peptides with GFP, and tested expression of the five pri–GFP fusion proteins in transformed S2 cells. GFP expression was detected for all but the sORF5–GFP

NATURE CELL BIOLOGY VOLUME 9 | NUMBER 6 | JUNE 2007

fusion, and western blot analysis confirmed that the fusion proteins were of the predicted size. Interestingly, the AUG of mille-pattes sORF4 (homologous to pri sORF5) overlaps with the stop codon of the preceding peptide and lacks a Kozak consensus sequence, suggesting that it may not be translated, or that translation may occur through an unusual read-through mechanism. Whether mille-pattes mRNA is normally translated has not been directly tested, although an embryonic requirement for mille-pattes transcript has been established by RNA interference (RNAi). Kondo et al. demonstrate that the four Pri peptides, although redundant with each other, are required for Drosophila development — expression of any one of the peptides rescues animals deleted for the endogenous locus, but a frame-shift mutation eliminates the rescuing activity. Polycistronic RNAs are rare in eukaryotes, where they comprise two general classes of RNA: polycistronic pre-mRNAs that are processed into mature monocistronic mRNAs and mature polycistronic mRNAs for which translation of downstream proteins depends on an internal ribosome entry site (IRES) or some sort of ribosome re-initiation mechanism7. Until now, all examples of this second class of message were dicistronic, coding for two, often functionally-related, proteins. The mille-pattes and pri genes define a novel subset of RNAs within this second class — unique because of 617

NEWS AND VIEWS Direct model: a secreted signal

a

Row:

1

1

2

4

5

6

7

3

4

5

6

7

6

7

Indirect model: assembly of aECM

c

Row:

3

Direct model: a membrane-permeable actin regulator

b

Row:

2

1

2

3

4

5

Figure 1 Direct and indirect models for Pri function. In each panel, the seven rows of denticlesecreting cells are shown. Kondo et al. demonstrated rescue of denticle formation in pri-mutant animals expressing a pri transgene in two (indicated in green) out of seven cells. (a) Model of direct action through a secreted signal. Pri (white ovals) produced in Pri-expressing cells might rescue the morphogenesis of neighbouring cells by activating a Pri receptor (red). (b) A second direct model of a membrane-permeable actin regulator. Here, Pri spreads from the Pri-expressing cells to neighbouring cells by crossing multiple cell membranes. In each cell, Pri interacts with regulators (red) of the actin cytoskeleton. (c) Indirect effects through assembly of the aECM. Pri may function within the Priexpressing cells to generate a functional aECM (a Pri-dependent aECM component is indicated in red). The morphogenesis of neighbouring cells would be rescued because the aECM provides the necessary cue or substrate for cell-shape change.

the number of peptides coded for, the extremely small size of those peptides, and the presence of a conserved core motif in each of the peptides. Savard et al. have proposed the name “polycistronic peptide coding RNA” (ppcRNA) to describe genes of this type, and have identified another candidate ppcRNA in Drosophila that codes for two closely related small peptides. So will ppcRNAs remain spectacular but rare, or will they define a large class of protein coding transcripts hidden in our genomes? The cell biology of pri is at least as compelling as its unusual gene structure. Embryos lacking pri function show defects in the cuticle structures of multiple epithelial tissues. In particular, ventral epidermal cells in mutant embryos lack denticles (hooked, tooth-like structures that extend from the apical surface 618

of the cells) and the tracheal epithelia fail to make taenidia (rings of cuticle that line tube lumens). Both denticles and taenidia are patterned by the actin cytoskeleton: the denticle shape reflects that of the actin-based projections present at their core, whereas taenidial pattern forms on a template of actin bundles and depends on the formin protein DAAM8,9. In pri embryos, the actin patches that normally assemble before the denticles and taenidia appear fail to accumulate. Intriguingly, the process of tracheal-tube expansion9, in which the diameter of the lumen increases dramatically over the course of approximately 3 h, is also mis-regulated. In pri mutant embryos, tracheal tubes expand unevenly, with some sections becoming grossly distended and cyst-like, whereas other sections

do not expand at all. Actin-cytoskeleton dynamics are also implicated in tube expansion, which may be suggestive of a common basis for Pri function in these processes11. To determine whether Pri functions cellautonomously during denticle formation, Kondo et al. expressed a pri transgene specifically in two out of the seven rows of cells that normally make denticles in each parasegment. Surprisingly, this proved sufficient to rescue denticle formation in all seven rows. The implication of this result is that the Pri peptides themselves, or a downstream target of their activity, can spread beyond the cells where Pri is expressed. There is yet another wrinkle to the pri story: despite sequence conservation at the aminoacid level between Pri and Mille-pattes and the presence of the shared septameric motifs, RNAiknockdown of mille-pattes results in a strikingly different embryonic phenotype. Tribolium embryos with knocked down mille-pattes show a dramatic change in body plan: segments of the embryo that would normally become abdomen are either lost or are transformed into supernumerary thoracic segments. Thus, whether the functions of the Pri and Mille-pattes peptides in Drosophila and Tribolium can be accounted for by a common mechanism of action remains to be determined. How might the Pri peptides function in a non-cell autonomous manner to regulate the morphogenesis of neighbouring epithelial cells? The dearth of data directly addressing the mechanism of Pri function leaves this question open to speculation. That said, most models fit into one of two general categories, defined by whether the non-cell autonomous effect of Pri is mediated directly by Pri activity, or indirectly, by a downsteam target of Pri activity (Fig. 1). In the direct models, Pri peptides exit the cell in which they are synthesized and act on neighbouring cells; in the indirect models, Pri peptides act within the cell where they are made, and one or more targets of Pri activity mediate the non-cell autonomous effects of Pri. In one direct model, the tiny peptides could represent a novel class of signalling molecules, perhaps related to neuropeptides or epitheliopeptides, but in which the signalling molecule is made as a tiny peptide rather than processed from a larger precursor protein12. Kondo et al. favour a more radical direct model in which the Pri peptides may be membrane permeable, spreading from cell to cell and functioning in each to regulate actin dynamics.

NATURE CELL BIOLOGY VOLUME 9 | NUMBER 6 | JUNE 2007

NEWS AND VIEWS Although no indirect models have been proposed by the authors, one possibility worth considering is that the Pri peptides may mediate the production of the apical extracellular matrix (aECM). As aECM components are secreted, it would then be possible for a few Pri-expressing cells to compensate for neighbouring cells deficient in some aspect of aECM formation. Both the tracheal epithelium and the epidermal epithelium secrete an aECM composed of proteins and chitin, and the aECM seems to be important for organizing epithelia13. In fact, cells lacking the Mummy protein —important for chitin synthesis,

protein glycosylation and GPI-anchor formation — not only block proper aECM formation but also cause defects in tube expansion, as well as denticle and taenidia formation14,15. Future studies addressing the mechanism of Pri and Mille-pattes action, and how they enable communication between neighbouring cells, will hopefully shed light on this intriguing family of polycistronic coding RNAs. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. 1. Tupy, J. L. et al. Proc. Natl Acad. Sci. USA 102, 5495– 5500 (2005). 2. Inagaki, S. et al. Genes Cells 10, 1163–1173 (2005).

3. Kondo, T. et al. Nature Cell Biology 9, 660–665 (2007). 5. Kastenmayer, J. P. et al. Genome Res. 16, 365–373 (2006). 6. Savard, J., Marques-Souza, H., Aranda, M. & Tautz, D. Cell 126, 559–569 (2006). 7. Blumenthal, T. Brief Funct. Genomic Proteomic 3, 199–211 (2004). 8. Dickinson, W. J & Thatcher, J. W. Cell Motil. Cytoskeleton 38, 9–21 (1997). 9. Matusek, T. et al. Development 133, 957–966 (2006). 10. Beitel, G. J. & Krasnow, M. A. Development 127, 3271–3282 (2000). 11. Tonning, A. et al. Dev Cell 9, 423–430 (2005). 12. Tour, E. & McGinnis, W. Cell 126, 448–449 (2006). 13. Swanson, L. E. & Beitel, G. J. Curr. Biol. 16, R51–R53 (2006). 14. Tonning, A., Helms, S., Schwarz, H., Uv, A. E. & Moussian, B. Development 133, 331–341 (2005). 15. Devine, W. P. et al. Proc. Natl Acad. Sci. USA 102, 17014–17019 (2005).

Anchoring microtubules at the spindle poles Anne Paoletti and Phong T. Tran In the cell, microtubule organizing centers (MTOCs) dynamically nucleate microtubules and arrange them in functional patterns, but microtubule anchoring to MTOCs is not well understood. A novel fission-yeast protein that anchors the γ-tubulin-containing nucleating complex (γ-TuC) at the spindle pole during mitosis has now been described. The work highlights the complex regulation of microtubule anchoring.

The architecture of the microtubule network is dynamically regulated, allowing it to perform specific functions at different stages of the cell cycle. For example, interphase linear microtubule arrays are required for cellular morphogenesis and cell-polarity establishment, whereas in mitosis, radial arrays of microtubules form the bipolar mitotic spindle and have crucial roles in chromosome segregation. In addition to their role in microtubule nucleation, MTOCs are also essential for organizing microtubule arrays and for microtubule anchoring. Microtubules are primarily nucleated by the γ-TuC of conserved proteins1. Nucleation produces a dynamic microtubule plus end that grows distally from the γ-TuC, and a microtubule minus end that is stabilized by the γ-TuC. In animal cells, γ-TuCs primarily localize to the centrosome, the major MTOC. How γ-TuCs, and indirectly microtubules, are anchored at Anne Paoletti is in the Institut Curie, Centre de Recherche, Paris, F-75248, France and the CNRS, UMR 144, Paris, F-75248, France. Phong T. Tran is at the University of Pennsylvania, Cell & Developmental Biology, 421 Curie Blvd Rm 1009, Philadelphia, PA 19104, USA. e-mail: [email protected]; [email protected]

the centrosome throughout the cell cycle is not well understood, although several centrosomal proteins (such as pericentrin, ninein, AKAP450, NLP, ASP and GGP-WD) have been implicated in the process2. The fission yeast, Schizosaccharomyces pombe, because of its simple microtubule network, genetic tractability and relative amenability to microscopic analysis, has proven useful in dissecting molecular mechanisms of microtubule organization3. In addition to the spindle-pole body (SPB) that is equivalent to the mammalian centrosome, fission yeast also have interphase (i) MTOCs. During interphase, both iMTOCs and the SPB are anchored to the nuclear envelope. During mitosis, the iMTOCs disassemble and the duplicated SPBs are embedded in the nuclear envelope and organize the mitotic spindle, which is composed of spindle and astral microtubules4,5. Spindle and astral microtubules are nucleated by γ-TuCs and anchored at the SPBs. On page 646 of this issue, Toya et al. describe a novel fission yeast coiled-coil protein, msd1p (mitotic spindle disanchored 1), which anchors spindle microtubules to the SPBs in an M-phasespecific manner6. Their analysis of msd1p reveals

NATURE CELL BIOLOGY VOLUME 9 | NUMBER 6 | JUNE 2007

the complicated regulation of microtubule anchoring to the SPB. Msd1–GFP only localizes to the SPBs during mitosis. Cells carrying a deletion of the msd1 gene (msd1Δ cells) assemble a spindle during mitosis, suggesting that microtubule nucleation by γ-TuCs is not affected by the loss of this protein. However, the mitotic spindle displays striking structural defects — for example, in addition to the normally anchored spindle microtubules emanating from the two SPBs, there are frequent additional spindle microtubules that are not attached to the SPBs. Instead, these microtubules abnormally extend beyond the SPB, protruding outward and deforming the nuclear envelope (Fig. 1A). The outward protruding tips of these microtubules do not display the typical fast dynamics of microtubule plus ends and do not contain the plus end-tracking protein mal3p (the fission yeast homologue of mammalian EB1). As alp4p (the fission yeast homologue of mammalian GCP2), an essential core component of the γ-TuC, was detected at these tips, they seem to be capped by γ-TuCs and represent microtubule minus ends that are not anchored to the SPB. Therefore, msd1p is a protein that anchors microtububles to the mitotic spindle poles. Msd1p binds directly to the γ-TuC 619