Special Reviews

Aggresomes and Autophagy Generate Sites for Virus Replication

See allHide authors and affiliations

Science  12 May 2006:
Vol. 312, Issue 5775, pp. 875-878
DOI: 10.1126/science.1126766

Abstract

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.

Autophagy 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 (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), 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.

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).

Recent studies show that autophagy plays an important role in the removal of protein aggregates from cells. Protein aggregates—for example, those associated with neurodegenerative conditions such as Huntington's disease—are 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).

Fig. 2.

Autophagy, aggresome, and virus assembly pathways converge at the MTOC. Step 1: Protein aggregates bind dynein motors and are transported to the MTOC on microtubules (yellow lines) and accumulate within aggresomes surrounded by vimentin. Step 2: Viruses (hexagons) entering cells are recognized by dynein motors and transported to aggresomes. Step 3: Isolation membranes are transported by dynein to aggresomes, where they engulf aggregated proteins. It is not known whether they also engulf viruses. The autophagosomes fuse with lysosomes, leading to degradation of protein aggregates and possibly of viruses.

Large Cytoplasmic DNA Viruses Replicate in Factories that Resemble Aggresomes

The factories generated by large cytoplasmic DNA viruses such as vaccinia virus, iridoviruses, 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).

Fig. 3.

African swine fever virus replicates in aggresomes. Cells infected with ASFV were labeled with antibody to the capsid protein (green). Viruses are assembled within a cage of vimentin (red) located next to the nucleus (blue). [Photo reproduced from (47) by permission of Blackwell Publishing; see also (12, 14, 17)]

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 iridovirus frog virus–3 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, μ2 (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 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 and use kinesin to enter the cytoplasm (25). Recognition by kinesin needs to be linked to the final stages of virus assembly to ensure that assembly intermediates do not move away from the supply of structural proteins present in the factory before assembly is completed. ASFV lacking the pE120R protein from the outer capsid does not move from factories, which suggests a role for pE120R in recruitment of kinesin (26); however, direct binding of pE120R to kinesin has not been demonstrated. For vaccinia virus, recruitment of kinesin motors is regulated during envelopment by the trans-Golgi network and requires the viral membrane protein A36R, which binds to the TPR domain of kinesin light chain (27, 28). This binding may be regulated by a second envelope protein, A33R, and/or by phosphorylation of A36R (29).

Replication of Viruses in Nuclear Aggresomes

Protein aggregation in the nucleus has been linked to neurodegenerative misfolded protein diseases such as Huntington's disease and spinocerebellar ataxias. Protein aggregates accumulate in nuclear aggresomes; because nuclear aggresomes recruit chaperones, ubiquitin, and proteasomes and expand in the presence of proteasome inhibitors, they may be sites specialized for protein degradation. Nuclear aggresomes are associated with nuclear domain 10 (ND10) bodies (30), which are also called promyelocytic leukemia (PML) bodies or promyelocytic leukemia oncogenic domains (PODs) and are involved in chromatin metabolism and DNA repair. ND10 structures are also associated with the genomes and/or replication complexes of herpesviruses, adenoviruses, parvoviruses, papovaviruses, and papillomaviruses (31). Thus, virus replication in the nucleus also takes place in structures that accumulate protein aggregates.

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 suggests that ND10 is recruited to viral replication complexes. It will be interesting to follow up these studies, using infected cells containing nuclear inclusions, to determine whether viral genomes and protein aggregates colocalize in ND10. Recent studies show that autophagy markers are not recruited to nuclear aggresomes and that the turnover of aggregated protein is unaffected by conditions that modulate autophagy (33). Autophagy is unable to clear misfolded proteins from nuclear aggresomes, and this may be why ND10 bodies are favored as sites of virus replication in the nucleus.

Autophagosomes as Sites of Viral Replication

The Nidovirales and Picornaviridae are positive-stranded RNA viruses that assemble a replication complex containing the RNA polymerase, as well as proteins with helicase and nucleotide triphosphate activity, on the cytoplasmic face of cellular membranes. The Nidovirales constitute the arteriviruses and coronaviruses, and they replicate in association with double-membraned vesicles. The vesicles are smaller than cellular autophagosomes (diameter 500 to 1000 nm); their diameters range from 80 nm for the arteriviruses up to 100 to 300 nm for the coronaviruses (3436). Arterivirus vesicles are connected to the ER, and those produced by the mouse hepatitis and SARS coronaviruses colocalize with Atg8 (35). Because coronavirus yields fall by a factor of 1000 in the absence of Atg5, the data suggest that the vesicles are related to autophagosomes. The vesicles can be produced by expression of the arterivirus Orf 1a encoding nonstructural proteins 2 to 7, which suggests that the transmembrane proteins and replicase enzymes encoded on Orf 1a stimulate production of autophagosomes (34).

The poliovirus replication complex also forms on double-membraned vesicles that originate either from coat protein complex II– 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 virus-induced 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 positive-sense 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 high-affinity 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 ubiquitin-proteasome 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 factories—for 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.

References and Notes

View Abstract

Navigate This Article