SPECIALSECTION

through preventing host recognition, either by using humanized proteins or by masking the recognition surface through attachment of poly(ethylene glycol) Epegylation (28)^. One can even imagine taking advantage of the immune system to enhance the utility of viral capsids through the rapid clearance of cages not localized at the cell or tissue target. This potentially would reduce the deleterious effects of exposure to nontargeted therapeutic agents. There are other attributes of viruses that have not yet been exploited but are well within the realm of possibility. Besides delivering nucleic acids, many viruses deliver other cargo, including catalysts and regulatory molecules. One can envision the targeted delivery of synthetic catalytic capacity or designer regulatory molecules. Engineered viral capsids can be envisioned as Trojan horses, remaining quiescent until a cellular signal causes release of their cargo. The ability to sequester a cargo within a protein architecture is not exclusive to viruses; similar architectures are common in the biological world, including ferritins (29) and heat shock proteins, and these have also been used for pharmaceutical and materials applications. Synthetic approaches in materials science often use extreme conditions, and the biological world has been traditionally viewed as a limited source of raw materials due to the relatively narrow temperature and chemical environments in which they exist. This perception is being challenged by the amazing array of environments that support life and their associated viruses. For example, the recent discovery of viruses from extreme thermal environments (Fig. 2G) potentially expands the synthetic window in which virus architectures may be used for materials applications (30). An extension of the principles exemplified by viruses to biomimetic approaches with the full range of protein templates will open the range of synthetic possibilities. We have turned a corner from viewing viruses only as hostile enemies to seeing and using them as a potentially vast and beneficial resource. Ultimately, the utility of viral capsids will be limited only by our own creativity.
References and Notes
1. C. A. Suttle, Nature 437, 356 (2005). 2. C. M. Shepherd et al., Nucleic Acids Res. 34, D386 (2006). 3. Y. Zhu, B. Carragher, D. J. Kriegman, R. A. Milligan, C. S. Potter, J. Struct. Biol. 135, 302 (2001). 4. T. Douglas, M. J. Young, Nature 393, 152 (1998). 5. T. Douglas et al., Adv. Mater. 14, 415 (2002). 6. M. Knez et al., Nano Lett. 3, 1079 (2003). 7. E. Gillitzer, D. Willits, M. Young, T. Douglas, Chem. Commun. 2390 (2002). 8. J. D. Lewis et al., Nat. Med. 12, 354 (2006). 9. Q. Wang, T. W. Lin, J. E. Johnson, M. G. Finn, Chem. Biol. 9, 813 (2002). 10. M. L. Flenniken et al., Chem. Commun. 2005, 447 (2005). 11. J. Tang et al., J. Struct. Biol. 15, 59 (2006). 12. J. Bancroft, G. Hills, R. Markham, Virology 31, 354 (1967). 13. B. Bothner et al., Virology 334, 17 (2005). 14. J. A. Speir, S. Munshi, G. Wang, T. S. Baker, J. E. Johnson, Structure 3, 63 (1995). 15. M. A. Allen et al., Magn. Reson. Med. 54, 807 (2005). 16. M. L. Flenniken et al., Chem. Biol. 13, 161 (2006). 17. A. Chatterji et al., Bioconjugate Chem. 15, 807 (2004). 18. M. T. Klem, D. Willits, M. Young, T. Douglas, J. Am. Chem. Soc. 125, 10806 (2003). 19. Q. Wang, T. W. Lin, L. Tang, J. E. Johnson, M. G. Finn, Ang. Chem. Int. Ed. 41, 459 (2002). 20. K. S. Raja, Q. Wang, M. G. Finn, ChemBioChem 4, 1348 (2003). 21. E. Strable, J. E. Johnson, M. G. Finn, Nano Lett. 4, 1385 (2004). 22. Q. Wang et al., J. Am. Chem. Soc. 125, 3192 (2003). 23. T. L. Schlick, Z. Ding, E. W. Kovacs, M. B. Francis, J. Am. Chem. Soc. 127, 3718 (2005). 24. R. Pasqualini, E. Ruoslahti, Nature 380, 364 (1996). 25. S. R. Whaley, D. S. English, E. L. Hu, P. F. Barbara, A. M. Belcher, Nature 405, 665 (2000). 26. C. B. Mao et al., Science 303, 213 (2004). 27. M. T. Klem et al., Adv. Funct. Mater. 15, 1489 (2005). 28. K. S. Raja et al., Biomacromolecules 4, 472 (2003). 29. F. C. Meldrum, V. J. Wade, D. L. Nimmo, B. R. Heywood, S. Mann, Nature 349, 684 (1991). 30. G. Rice et al., Proc. Natl. Acad. Sci. U.S.A. 101, 7716 (2004). 31. We thank J. E. Johnson for a long-standing collaboration in developing this field and for use of cryoelectron micrograph images. Much of this work has been made possible by funding from NIH (RO1GM61340, R21EB005364, and RO1EB000432), Office of Naval Research (N00014-04-1-0672), and NSF (DMR011636 and MCB0132156). 10.1126/science.1123223

REVIEW

Aggresomes and Autophagy Generate Sites for Virus Replication

Thomas Wileman The replication of many viruses is associated with specific intracellular compartments called virus factories or virioplasm. These are thought to provide a physical scaffold to concentrate viral components and thereby increase the efficiency of replication. The formation of virus replication sites often results in rearrangement of cellular membranes and reorganization of the cytoskeleton. Similar rearrangements are seen in cells in response to protein aggregation, where aggresomes and autophagosomes are produced to facilitate protein degradation. Here I review the evidence that some viruses induce aggresomes and autophagosomes to generate sites of replication. utophagy is a cellular response to starvation as well as a quality control system that can remove damaged organelles and long-lived proteins from the cytoplasm. Autophagy is involved in several developmental pathways and disease processes (1, 2) and may provide defense against pathogens (35). In resting cells, autophagy is inhibited by the TOR

School of Medicine, Health Policy and Practice, University of East Anglia, Norwich NR4 7TJ, UK. E-mail: t.wileman@ uea.ac.uk

(target of rapamycin) kinase and is triggered by events such as starvation or by the presence of rapamycin, either of which leads to dephosphorylation and inactivation of TOR. Autophagy begins with the sequestration of an area of the cytoplasm within a crescent-shaped isolation membrane (Fig. 1). Isolation membranes mature into large double-membraned vesicles (diameter 500 to 1000 nm) called autophagosomes, which eventually fuse with endosomes and lysosomes (6). Isolation membranes contain Atg5, Atg8 (also called LC3 in mammals), SCIENCE VOL 312

and Atg12 proteins. The Atg8 protein remains associated with the autophagosome and can be used to track the fusion of autophagosomes with endosomes and lysosomes. Recent studies show that autophagy plays an important role in the removal of protein aggregates from cells. Protein aggregatesfor example, those associated with neurodegenerative conditions such as Huntington_s diseaseare first delivered to the microtubule organizing center (MTOC) by dynein-dependent retrograde transport along microtubules (Fig. 2, step 1). When the degradative capacity of proteasomes is exceeded, protein aggregates accumulate in perinuclear inclusions called aggresomes (7). Aggresomes are surrounded by vimentin filaments and recruit chaperones, proteasomes, and mitochondria, suggesting a site specialized for protein folding and degradation. Many protein aggregates that cannot be refolded or degraded by proteasomes are eventually removed from aggresomes by autophagy, allowing delivery to lysosomes for degradation (8, 9) (Fig. 2, step 3). Large Cytoplasmic DNA Viruses Replicate in Factories that Resemble Aggresomes The factories generated by large cytoplasmic DNA viruses such as vaccinia virus, irido-

www.sciencemag.org

12 MAY 2006

875

VIROLOGY
viruses, and African swine fever virus (ASFV) contain viral DNA and structural proteins concentrated within inclusions that closely resemble aggresomes (Fig. 3). In common with aggresomes, factories are located close to the MTOC and recruit vimentin, cellular chaperones, ubiquitin, and mitochondria (1014). For ASFV, factories are absent in cells expressing the dominant negative dynein motor protein p50 dynamitin, indicating a role for the dynein motor in factory formation; also in common with aggresomes, these factories are dispersed by drugs that depolymerize microtubules (14). The close structural similarity between aggresomes and virus factories raises the possibility that aggresomes offer an innate defense against infection that confines viruses within inclusions at the MTOC in preparation for degradation by proteasomes and/ or autophagy (Fig. 2, step 2). The dual role played by dynein (in the formation of virus factories and in the delivery of protein aggregates to aggresomes) suggests that cells may see viruses as protein aggregates and transport them to the MTOC (14, 15). Many viral core particles are of similar size (60 to 100 nm) to the aggregates that are transported to aggresomes, and the number of viruses that are recognized by dynein motors soon after entry into the cell is impressive (16); examples include vaccinia virus, ASFV, herpesvirus, adenovirus, and canine parvovirus. Recognition by dynein may be an innate response to viral particles. For many viruses, transport to the MTOC for storage and eventual degradation may protect cells from infection. For the large cytoplasmic DNA viruses, however, recognition by elements of the aggresome pathway may provide a site for replication. Do Aggresomes Offer Advantages as Sites of Replication? To demonstrate that aggresomes facilitate replication, it is necessary to show that replication is slowed in the absence of aggresomes. ASFV replication is blocked by p50 dynamitin, which suggests that delivery of incoming viruses to the MTOC is important to initiate virus replication (14). Replication of ASFV and iridoviruses also appears to require rearrangement of vimentin. During the early stages of ASFV infection, vimentin is transported by microtubules to the MTOC. Once viral DNA replication is initiated, vimentin is phosphorylated on serine by calcium/calmodulin-dependent protein kinase II (CaMKII) and is redistributed to the edge of the factory, where it forms a cage (17). Inhibition of CaMKII prevents vimentin cage formation and prevents virus DNA replication. Similarly, early studies showed that temperature-sensitive mutants of the irido-

Fig. 1. Production of autophagosomes as replication sites for poliovirus and coronaviruses. Step 1: Autophagy is activated by dephosphorylation of TOR, which results in recruitment of Atg8, Atg5, and Atg12 onto crescent-shaped isolation membranes. Step 2: Isolation membranes release Atg5 and Atg12 and mature into double-membraned autophagosomes that engulf the cytosol, which may contain aggregated proteins (stacked crescents) or viruses (hexagons). The autophagosome provides a platform for the assembly of replication complexes by picornaviruses and nidoviruses (orange spheres). Step 3: Autophagosomes fuse with lysosomes, leading to degradation of content. Viruses and proteins that survive in lysosomes may be delivered to the cell surface (step 4). virus frog virus3 that are unable to phosphorylate vimentin cannot rearrange vimentin and cannot proceed to late gene expression (10, 18). Vimentin rearrangement may facilitate replication by providing a scaffold during the assembly of the virus factory (12), and the vimentin cage made later during infection may prevent diffusion of viral components into the cytoplasm (14, 17). This would allow structural proteins to be concentrated within one site in the cell, with a steady supply of viral components and host proteins delivered from the cytoplasm along microtubules. Such a mechanism may be particularly important for complex large DNA viruses, such as vaccinia and ASFV, that are assembled from as many as 50 different structural proteins. It is not always necessary for aggresomes to facilitate replication. It is possible that some viruses may be restricted to aggresomes because of innate cellular defenses against infection, but would replicate just as well if they remained in the cytoplasm. Rotaviruses replicate in cytoplasmic virus factories, which are globular, as in strain T3D, or filamentous, as in strain T1L. The globular factories coalesce and migrate to the MTOC, which suggests involvement of the aggresome pathway, whereas the filamentous factories align along microtubules in the cytoplasm, which are stabilized by a minor viral structural protein, m2 (19, 20). It is possible that immobilization of the factory on microtubules resists recruitment of rotavirus proteins into aggresomes (19). Because the filamentous strains of rotaviruses make greater quantities of viral proteins during infection and show increased virulence and broader tropism, escape from VOL 312 SCIENCE aggresomes may offer a selective advantage to rotaviruses. How Do Viruses Avoid Degradation in Aggresomes? One possibility is that the degradative arm of the pathway is blocked in infected cells. This could involve inhibition of transport of isolation membranes and autophagosomes into virus factories and/or inhibition of subsequent fusion with lysosomes. A second possibility is that isolation membranes are recruited into the factory and are then used as a source of membranes for virus envelopment. The membranes used for the envelopment of vaccinia virus and ASFV are poorly characterized but appear similar to the membranes used to form isolation membranes and autophagosomes (6). Both membranes are derived from specialized extensions of the endoplasmic reticulum (ER) and/or de novo membrane synthesis, producing perinuclear virioplasm or preautophagous structures, respectively. The crescent-shaped structures that form the inner envelope of vaccinia virus and ASFV are not dissimilar from the crescents formed by isolation membranes. When cells lacking Atg5 are infected with vaccinia virus, morphogenesis is normal (21), making it unlikely that the internal envelope of vaccinia originates from the isolation membrane. Vaccinia virus and ASFV use microtubules to leave virus factories. Movement of vaccinia virus from factories requires intact microtubules and the viral proteins A27L and F12L, but the motor driving transport has not been identified (2224). ASFV particles recruit the cargo-binding domain (TPR domain) of conventional kinesin light chain to factories

876

12 MAY 2006

www.sciencemag.org

SPECIALSECTION
Aggregated proteins are continuously delivered to nuclear aggresomes and are relatively mobile. It is not yet understood how protein aggregates are recruited to, or recruit, ND10. ND10 structures are relatively static in the nucleus but can exchange their content with the rest of the nucleoplasm. When herpesvirus capsids are delivered to one side of the nucleus, ND10 is recruited to the site of incoming viruses, which suggests that viruses recruit ND10 components as they enter the nucleus (32). Recruitment requires association of viral DNA with viral proteins required for translation and/or replication with the genome, which Fig. 2. Autophagy, aggresome, and virus assembly pathways converge at the MTOC. Step 1: Protein suggests that ND10 is recruited to aggregates bind dynein motors and are transported to the MTOC on microtubules (yellow lines) and viral replication complexes. It will accumulate within aggresomes surrounded by vimentin. Step 2: Viruses (hexagons) entering cells are be interesting to follow up these recognized by dynein motors and transported to aggresomes. Step 3: Isolation membranes are transported by studies, using infected cells condynein to aggresomes, where they engulf aggregated proteins. It is not known whether they also engulf taining nuclear inclusions, to deterviruses. The autophagosomes fuse with lysosomes, leading to degradation of protein aggregates and possibly mine whether viral genomes and of viruses. protein aggregates colocalize in ND10. Recent studies show that autophagy work and requires the viral markers are not recruited to nuclear aggresomes membrane protein A36R, which and that the turnover of aggregated protein is binds to the TPR domain of unaffected by conditions that modulate autophkinesin light chain (27, 28). This agy (33). Autophagy is unable to clear misbinding may be regulated by a folded proteins from nuclear aggresomes, and second envelope protein, A33R, this may be why ND10 bodies are favored as and/or by phosphorylation of sites of virus replication in the nucleus. A36R (29). Autophagosomes as Sites of Viral Replication Replication of Viruses in Nuclear Aggresomes The Nidovirales and Picornaviridae are positiveProtein aggregation in the nucle- stranded RNA viruses that assemble a replicaus has been linked to neuro- tion complex containing the RNA polymerase, degenerative misfolded protein as well as proteins with helicase and nucleodiseases such as Huntingtons dis- tide triphosphate activity, on the cytoplasmic ease and spinocerebellar ataxias. face of cellular membranes. The Nidovirales Fig. 3. African swine fever virus replicates in aggresomes. Cells Protein aggregates accumu- constitute the arteriviruses and coronaviruses, infected with ASFV were labeled with antibody to the capsid protein (green). Viruses are assembled within a cage of late in nuclear aggresomes; and they replicate in association with doublevimentin (red) located next to the nucleus (blue). [Photo because nuclear aggresomes re- membraned vesicles. The vesicles are smaller reproduced from (47) by permission of Blackwell Publishing; cruit chaperones, ubiquitin, and than cellular autophagosomes (diameter 500 proteasomes and expand in to 1000 nm); their diameters range from 80 nm see also (12, 14, 17 )] the presence of proteasome in- for the arteriviruses up to 100 to 300 nm for the hibitors, they may be sites spe- coronaviruses (3436). Arterivirus vesicles are and use kinesin to enter the cytoplasm (25). cialized for protein degradation. Nuclear connected to the ER, and those produced by Recognition by kinesin needs to be linked to aggresomes are associated with nuclear do- the mouse hepatitis and SARS coronaviruses the final stages of virus assembly to ensure main 10 (ND10) bodies (30), which are also colocalize with Atg8 (35). Because coronavirus that assembly intermediates do not move called promyelocytic leukemia (PML) bodies yields fall by a factor of 1000 in the absence of away from the supply of structural proteins or promyelocytic leukemia oncogenic do- Atg5, the data suggest that the vesicles are present in the factory before assembly is mains (PODs) and are involved in chro- related to autophagosomes. The vesicles can completed. ASFV lacking the pE120R pro- matin metabolism and DNA repair. ND10 be produced by expression of the arterivirus tein from the outer capsid does not move structures are also associated with the ge- Orf 1a encoding nonstructural proteins 2 to 7, from factories, which suggests a role for nomes and/or replication complexes of which suggests that the transmembrane proteins pE120R in recruitment of kinesin (26); how- herpesviruses, adenoviruses, parvoviruses, and replicase enzymes encoded on Orf 1a stimever, direct binding of pE120R to kinesin has papovaviruses, and papillomaviruses (31). ulate production of autophagosomes (34). not been demonstrated. For vaccinia virus, Thus, virus replication in the nucleus also The poliovirus replication complex also recruitment of kinesin motors is regulated takes place in structures that accumulate pro- forms on double-membraned vesicles that during envelopment by the trans-Golgi net- tein aggregates. originate either from coat protein complex II www.sciencemag.org SCIENCE VOL 312 12 MAY 2006

877

VIROLOGY
coated vesicles, which move proteins from the ER to the Golgi (37), or from autophagosomes (38). They are again smaller than cellular autophagosomes and range from 200 to 400 nm in diameter (39). A role for autophagy during replication of poliovirus is nonetheless supported by colocalization of Atg8 with virusinduced vesicles and by lowered virus yield in the absence of Atg12 and Atg8 (40). In common with coronaviruses, double-membraned vesicles are produced by coexpression of the membrane-targeted components of the replication complex, in this case 3A and 2BC (38, 40). Viruses may induce autophagy to generate a scaffold for the replication complex; alternatively, replication complexes may be restricted to autophagosomes as a result of innate defenses against infection. Although relatively few viruses have been studied to date, virus yields fall after loss of Atg proteins, which suggests that autophagosomes facilitate replication (35, 40). The factor of 1000 fall for coronavirus is more striking than the factor of 20 fall seen for poliovirus; hence, viruses may vary in dependence on autophagosomes and/or Atg proteins. Surprisingly, suppression of Atg12 and Atg8 by RNA interference reduced extracellular poliovirus yield more than it reduced intracellular levels of virus, indicating a selective effect on virus release (40). Enteroviruses such as poliovirus are relatively resistant to low pH and proteases and may survive in autophagosomes and lysosomes. Viruses engulfed by autophagosomes (Fig. 1, step 2) may be released from cells after fusion of lysosomes with the plasma membrane (Fig. 1, step 4). A similar pathway leading to the extracellular delivery of protein aggregates may be responsible for the deposition of amyloid plaques associated with misfolded protein diseases. Again, the pathways followed by viruses and protein aggregates converge. Two other families of positive-stranded RNA viruses, the alphaviruses and flaviviruses, replicate inside invaginations (diameter 50 nm) in cellular membrane compartments called spherules (41, 42). The replication complexes are assembled inside spherules that are connected by a neck to the cytosol, allowing positivesense genomes to be delivered into the cell. For poliovirus, most of the double-membraned vesicles appear to be sealed, which suggests that replication occurs on the outside of membrane vesicles, allowing direct delivery of new genomes into the cytosol. The RNA polymerase of poliovirus self-assembles into large flat ordered arrays (43) that are believed to coat the surface of membranes, forming a catalytic shell that may be critical for highaffinity binding of RNA and RNA elongation. The replicase complexes of alphaviruses, on the other hand, contain a low relative concentration of RNA polymerase. Replication efficiency may be enhanced by the neck of the spherule, which slows the diffusion of negative-sense RNA into the cytosol, increasing its availability as a template for genome replication. The most striking difference between spherules and autophagosomes is their mechanism of production. Spherules are produced from existing cellular membranes by active assembly of viral proteins. This process has been likened to the coordinated assembly of viral proteins on membranes that leads to capsid assembly, genome packaging, and budding (44, 45). Because autophagy is suppressed in resting cells by the TOR kinase, production of autophagosomes requires activation of signaling pathways. The structure of the vesicle is then determined by host Atg proteins rather than by the assembly of viral proteins. Conclusions It is clear that some viruses replicate on (or within) structures used by cells to remove protein aggregates. This may be important for virus replication, because virus yields fall if the structures are not made or if viruses are denied access to them. Aggresomes and autophagosomes are present at low levels in resting cells and are stimulated in response to specific signals. Understanding how virus infection stimulates these pathways to produce new structures for virus replication is a challenge for the future. Protein aggregation causes global inhibition of the ubiquitinproteasome system and signals the formation of aggresomes (46). It will be interesting to see whether large cytoplasmic DNA viruses also modulate the ubiquitin-proteasome system to generate inclusions for replication. Virus factories share many structural features with aggresomes, but direct evidence that they are the same structure has been lacking. It will be important to demonstrate a functional link between aggresomes and factoriesfor example, by showing that virus factories can recruit misfolded proteins, or that viruses assemble in preformed aggresomes. Autophagosomes are formed in cells expressing the membrane-targeted components of replication complexes. These proteins provide an excellent opportunity to unravel how these viruses generate autophagosomes. Finally, it is likely that many viruses stimulate these pathways during infection and are eliminated (4, 5). An understanding of how some viruses survive will provide a better understanding of the role played by autophagy and aggresome pathways during infection, leading to valuable tools for further dissection of these processes. VOL 312 SCIENCE
References and Notes
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. T. Shintani, D. J. Klionsky, Science 306, 990 (2004). B. Levine, D. J. Klionsky, Dev. Cell 6, 463 (2004). V. Deretic, Trends Immunol. 26, 523 (2005). Z. Talloczy et al., Proc. Natl. Acad. Sci. U.S.A. 99, 190 (2002). K. Kirkegaard, M. P. Taylor, W. T. Jackson, Nat. Rev. Microbiol. 2, 301 (2004). F. Reggiori, D. J. Klionsky, Curr. Opin. Cell Biol. 17, 415 (2005). J. A. Johnston, C. L. Ward, R. R. Kopito, J. Cell Biol. 143, 1883 (1998). A. Iwata, B. E. Riley, J. A. Johnston, R. R. Kopito, J. Biol. Chem. 280, 40282 (2005). B. Ravikumar et al., Nat. Genet. 37, 771 (2005). M. Chen, R. Goorha, K. G. Murti, J. Gen. Virol. 67, 915 (1986). A. Ploubidou et al., EMBO J. 19, 3932 (2000). C. Risco et al., J. Virol. 76, 1839 (2002). A. Schepis, B. Schramm, C. A. M. de Haan, J. K. Locker, Traffic 7, 308 (2006). C. M. Heath, M. Windsor, T. Wileman, J. Cell Biol. 153, 449 (2001). B. Sodeik, J. Cell Biol. 159, 393 (2002). K. Dohner, C.-H. Nagel, B. Sodeik, Trends Microbiol. 13, 320 (2005). S. Stefanovic, M. Windsor, K. Nagata, M. Inagaki, T. Wileman, J. Virol. 79, 11766 (2005). D. B. Willis, R. Goorha, A. Granoff, Virology 98, 328 (1979). J. S. Parker, T. J. Broering, J. Kim, D. E. Higgins, M. L. Nibert, J. Virol. 76, 4483 (2002). T. J. Broering et al., J. Virol. 78, 1882 (2004). H. Zhang et al., Autophagy 2, 91 (2006). C. M. Sanderson, M. Hollinshead, G. L. Smith, J. Gen. Virol. 81, 47 (2000). J. Rietdorf et al., Nat. Cell Biol. 3, 992 (2001). M. Hollinshead et al., J. Cell Biol. 154, 389 (2001). N. Jouvenet, P. Monaghan, M. Way, T. Wileman, J. Virol. 78, 7990 (2004). G. Andres et al., J. Virol. 75, 6758 (2001). B. M. Ward, B. Moss, J. Virol. 78, 2486 (2004). U. F. Greber, M. Way, Cell 124, 741 (2006). T. P. Newsome, N. Scaplehorn, M. Way, Science 306, 124 (2004); published online 5 August 2004 (10.1126/ science.1101509). L. Fu et al., Mol. Biol. Cell 16, 4905 (2005). R. D. Everett, Cell Microbiol. 8, 365 (2006). R. D. Everett, J. Murray, J. Virol. 79, 5078 (2005). A. Iwata et al., Proc. Natl. Acad. Sci. U.S.A. 102, 13135 (2005). K. W. Pedersen, Y. van der Meer, N. Roos, E. J. Snijder, J. Virol. 73, 2016 (1999). E. Prentice, W. G. Jerome, T. Yoshimori, N. Mizushima, M. R. Denison, J. Biol. Chem. 279, 10136 (2004). R. Gosert, A. Kanjanahaluethai, D. Egger, K. Beinz, S. C. Baker, J. Virol. 76, 3697 (2002). R. C. Rust et al., J. Virol. 75, 9808 (2001). D. A. Suhy, T. H. Giddings Jr., K. Kirkegaard, J. Virol. 74, 8953 (2000). S. Dales, H. J. Eggers, I. Tamm, G. E. Palade, Virology 26, 379 (1965). W. T. Jackson et al., PLoS Biol. 3, e156 (2005). J. Mackenzie, Traffic 6, 967 (2005). R. R. Novoa et al., Biol. Cell 97, 147 (2005). J. M. Lyle, E. Bullitt, K. Bienz, K. Kirkegaard, Science 296, 2218 (2002). M. Schwartz et al., Mol. Cell 9, 505 (2002). M. Schwartz, J. Chen, W.-M. Lee, M. Janda, P. Ahlquist, Proc. Natl. Acad. Sci. U.S.A. 101, 11263 (2004). E. J. Bennett, N. F. Bence, R. Jayakumar, R. R. Kopito, Mol. Cell 17, 351 (2005). P. Monaghan, H. Cook, P. Hawes, J. Simpson, F. Tomley, J. Microsc. 212, 62 (2003). Work in the authors laboratory was funded by the Biotechnology and Biological Sciences Research Council.

30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48.

10.1126/science.1126766

878

12 MAY 2006

www.sciencemag.org