Structure, Function and Intracellular Dynamics of Alphavirus Replication Complexes

Institute of Biotechnology Department of Biological and Environmental Sciences Division of Microbiology Faculty of Biosciences University of Helsinki and Viikki Graduate School in Biosciences

Structure, function and intracellular dynamics of alphavirus replication complexes

Giuseppe Balistreri

ACADEMIC DISSERTATION

To be presented, with the permission of the Faculty of Biosciences of the University of Helsinki, for public examination in Room 2041 at Biocenter Viikki (Viikinkaari 5), Helsinki, on 4th August 2010.

Helsinki, 2010

To my astonishing beautiful wife Laura, the Sun in my life, and my little princesses Emilia, Sara and Kiira, who make my days such an unstoppable explosion of joy!

Supervisor

Professor Emeritus Leevi Kääriäinen and Docent Tero Ahola Institute of Biotechnology, University of Helsinki Viikinkaari 9, 00790 Helsinki, Finland

Reviewers

Professor Pirjo Laakkonen Molecular Cancer Biology Research Program, Institute of Biomedicine, University of Helsinki, Helsinki, Finland and A.I. Virtanen Institute for Molecular Medicine, University of Kuopio, Kuopio, Finland

Docent Vesa Olkkonen Minerva Foundation Institute for Medical Research Biomedicum 2U, Tukholmankatu 8 00290 Helsinki, Finland

Opponent

Professor Kai Simons Max-Plank-Institute of Molecular Cell Biology and Genetics Pfotenhauerstrasse 108, 01307 Dresden, Germany

ISBN 978-952-10-6377-0 (pbk.) ISBN 978-952-10-6378-7 (PDF)

Helsinki University Printing House Helsinki 2010

Abstract

Intracellular membrane alterations are hallmarks of positive-sense RNA (+RNA) virus replication. Strong evidence indicates that within these ‘exotic’ compartments, viral replicase proteins engage in RNA genome replication and transcription. To date, fundamental questions such as the origin of altered membranes, mechanisms of membrane deformation and topological distribution and function of viral components, are still waiting for comprehensive answers. This study addressed some of the above mentioned questions for the membrane alterations induced during Semliki Forest virus (SFV) infection of mammalian cells. With the aid of electron and fluorescence microscopy coupled with radioactive labelling and immuno-cytochemistry techniques, our group and others showed that few hours after infection the four non structural proteins (nsP1-4) and newly synthesized RNAs of SFV colocalized in close proximity of small membrane invaginations, designated as “spherules”. These 50-70 nm structures were mainly detected in the perinuclear area, at the limiting membrane of modified endosomes and lysosomes, named CPV-I (cytopathic vacuoles type I). More rarely, spherules were also found at the plasma membrane (PM). In the first part of this study I present the first three-dimensional reconstruction of the CPV-I and the spherules, obtained by electron tomography after chemical or cryo-fixation. Different approaches for imaging these macromolecular assemblies to obtain better structure preservation and higher resolution are presented as unpublished data. This study provides insights into spherule organization and distribution of viral components. The results of this and other experiments presented in this thesis will challenge currently accepted models for virus replication complex formation and function. In a revisitation of our previous models, the second part of this work provides the first complete description of the biogenesis of the CPV-I. The results demonstrate that these virus- induced vacuoles, where hundreds of spherules accumulate at late stages during infection, represent the final phase of a journey initiated at the PM, which apparently serves as a platform for spherule formation. From the PM spherules were internalized by an endocytic event that required the activity of the class I PI3K, caveolin-1, cellular cholesterol and functional actin-myosin network. The resulting neutral endocytic carrier vesicle delivered the spherules to the membrane of pre-existing acidic endosomes via multiple fusion events. Microtubule based transport supported the vectorial transfer of these intermediates to the pericentriolar area where further fusions generated the CPV-I. A signal for spherule internalization was identified in one of the replicase proteins, nsP3. Infections of cells with viruses harbouring a deletion in a highly phosphorylated region of nsP3 did not result in the formation of CPV-Is. Instead, thousands of spherules remained at the PM throughout the infection cycle. Finally, the role of the replicase protein nsP2 during viral RNA replication and transcription was investigated. Three enzymatic activities, protease, NTPase and RNA-triphosphatase were studied with the aid of temperature sensitive mutants in vitro and, when possible, in vivo. The results highlighted the interplay of the different nsP2 functions during different steps of RNA replication and sub-genomic promoter regulation, and suggest that the protein could have different activities when participating in the replication complex or as a free enzyme.

Acknowledgments

I would like to express my deep gratitude to all the wonderful persons I have come across during these years at the Institute of Biotechnology. I will start with my beloved colleagues and friends in the SFV group, with whom I have had the privilege to work. To Pirjo Spuul, Andrey Golubtsov, and Maarit Neuvonen: there is not much space, you see, in this thesis book, but I know you will squeeze out of this THANK YOU all the wonderful time we have had together and will continue to have. I truly love you all! I would like to extend my gratitude to the rest of the former and present members of the group, including Julia Magden, Anne Salonen, Peter Sarin, Pia Salomaa, Nana Nordman, Airi Sinkko, Javier Caldentey, Leena Pohiola, and the new members Kirsi Hellström, Katri Kallio, Yaseen Syed-Basha and Antti Ahde. I thank Dr. Tero Ahola for giving me the opportunity to finalize my research in his group and for his help and support during these years. I am particularly grateful to the former and current directors of the Institute of Biotechnology, Professor Mart Saarma and Professor Tomi Mäkelä, for being always so supportive and for creating such a productive working environment. I am also thankful to Professor Timo Korhonen, head of the Division of Microbiology, for his positive and encouraging attitude towards my studies. I am grateful to Ritva Virkola and Anita Tienhaara for guiding me through the practical arrangements on the completion of my graduate studies. Part of the merit for the completion of this PhD thesis is also of the Viikki Graduate School in Biosciences, for the top- level education provided, and the excellent work of the former and current coordinators Eeva Sievi and Sandra Falk. A special thank goes to the members of my “follow-up” group, Professor Marc Bauman and Jaana Bamford, for their constructive critics towards my work and for useful advice. For making the reviewing process of this thesis one of the most educative experiences of my PhD training, I would like to express my gratitude to Professor Pirjo Laakkonen and Docent Vesa Olkkonen. I will make treasure of the fruitful discussions with all the group leaders and professors on the campus, especially Dr. Johan Peränen, Claudio Rivera, Pekka Lappalainen, Jussi Jantti, Sirkka Keränen, Ulla Pirvola, Maria Vartiainen, Leonard Kirugh, Sarah Butcher and Dennis Damford. A big, big hug to all my colleagues and friends in the 5th and 4th floor, I already miss you! It has been a pleasure to collaborate with Dr. Eija Jokitalo and all the members of the fantastic EM unit, particularly Helena Vihinen, Mervi Lindman.and Aria Strandell. As to how did I end up in Finland? This was evidently a conspiracy between my wife Laura and Professor Marja Makarov, who reasoned the way to keep me here for a long, long time: they brought me at the office door of Professor Leevi Kääriäinen…the rest is in this thesis! It has been a long way but it feels as we started yesterday. I would like to thank you both for all you have done for me; I hope I have deserved it. Working with you Leevi is a trilling experience. I have found in your contagious enthusiasm and perfectionism an endless source of motivation. I wish you a bright future and I am sure we will have the opportunity to soon work together again. Finally, I would like to thank all the members of my big and wonderful family, who are always supporting me and helping in moving forward.

Helsinki, 2010 2

Contents

Abstract 4

Acknowledgements 5

List of original publications 8

Abbreviations 9

1 Introduction 10

1.1 +RNA virus replication in association with cellular membranes 10

1.2 Alphaviruses 14

1.3 SFV replication cycle 15

1.4 General composition of alphavirus replication complex 18

1.5 The spherules of alphavirus: ‘mini-organelles’ for viral RNA replication 21

1.6 General principles of endocytosis 22

Endocytosis assisted by clathrin 23

Caveolae mediated endocytosis: caveolins, cavins and flotillins 24

Macropinocytosis and phagocytosis 25

Other types of endocytosis 26

2. Aims of the study 27

3. Methods 28

4. Results 31

4.1 Approaching the high resolution 3D-structure of alphavirus replication complex: from conventional EM to correlative cryo-electron tomography (Unpublished) 31

4.2 Dynamics of SFV replication complexes (I, II) 36

4.2.1 Spherules arise at the plasma membrane (I) 36

4.2.2 Internalization of newly formed spherules from the PM (I, II) 40

4.2.3. The signal for spherule internalization resides in the replicase component nsP3 (II) 45

4.2.4 Intracellular transport of SFV spherules: biogenesis of CPV-I (I) 46

4.3 SFV infection induces down-regulation of transferrin receptor and redistribution of EEA-1 (unpublished) 47

4.4 Addressing the function of nsP2 with ts-mutants (III) 49

Discussion 53

Structure of SFV spherules 53

Spherule formation 55

RNA replication and the multifunctional protein nsP2 60

The viroplasms of SFV: spherules and CPVs 62

Comparison with viroplasms induced by other +RNA viruses 65

Internalization of spherules from the PM: spherules as unconventional endocytic cargoes 68

References 72

List of original publications

This thesis is based on the following articles, which are referred to in the text by their roman numerals:

I. Spuul P.*, Balistreri G.*, Kääriäinen L., Ahola T. (2010) PI3K-, actin- and microtubule-dependent transport of Semliki Forest virus replication complex from the plasma membrane to modified lysosomes. J. Virol. 84, 7543-7557. II. Balistreri G.*, Spuul P.*, Vihiven H., Belevich I., Jokitalo E., Peränen J., Ahola T. (2010) The multi-caveolar transport of Semliki Forest virus replication complex spherules from the plasma membrane defines an unconventional endocytic event. Manuscript. III. Balistreri G., Caldentey J., Kääriäinen L., Ahola T. (2007) Enzymatic defects of the nsP2 proteins of SFV temperature-sensitive mutants. J. Virol. 81, 2849-2860. Unpublished data are also presented * Equal contribution

Abbreviations

+RNA viruses positive-sense RNA viruses aa amino acid BHK baby hamster kidney Cav1 caveolin 1 CCP clathrin coated pit CET correlative electron tomography CLEM correlative light electron microscopy CME clathrin mediated endocytosis CPV cytopathic vacuole DMV double membrane vesicle EEA1 early endosomal antigen 1 EM electron microscopy FHV flock house virus HCV hepatitis C virus nsP(s) non-structural protein(s) NTPase nucleotide triphosphatase ORF open reading frame PFA paraformaldehyde PFU plaque forming unit PI3K phosphoinositol 3 kinase PM plasma membrane PV polio virus RC replication complex RdRP RNA-dependent RNA polymerase RTPase RNA-triphosphatase SARS severe acute respiratory syndrome SFV Semliki Forest virus SIN Sindbis virus TC transcription complex TIR-FM total internal reflection fluorescence microscopy Ts temperature sensitive wt wild type

1 Introduction

1.1 +RNA virus replication in association with cellular membranes

Viruses with single-stranded RNA genome of positive polarity (+RNA viruses) represent over a third of known virus genera, and include important plant, animal and human pathogens. +RNA viruses can be very different in morphology, genome size, and life cycle. However, a large amount of evidence accumulated during the last two decades has underlined a common characteristic: viral RNA synthesis occurs in association with intracellular membranes. Disruption of such association by site directed mutagenesis of virus replicase proteins involved in membrane anchoring is lethal for the virus. Based on morphological studies by EM, and colocalization between cellular markers and viral components, it appears that different viruses utilize membranes of different organelles to support replication. In all reported cases, not only are viral replicase proteins and newly synthesized RNAs found in close association with membranes, but the formation of impressive membrane rearrangements is readily detected in electron micrographs of infected cells. It has been shown that although these structures can be morphologically very different, depending on the virus family, all appear to consist of densely packed membrane vesicles (or spherules), where viral components are concentrated and possibly protected from host defence mechanisms in the cytoplasm. In the case of +RNA viruses that infect animals, the vesicles accumulate in the perinuclear area. As opposed to the cytoplasmic space, these specialized compartments have been designated ‘viroplasms’. In the following text a functional distinction will be made between vesicles (or spherules) as the site of virus RNA synthesis, and viroplasms as the cytoplasmic area where vesicles are found concentrated at late stages of infection. In Table 1, representative examples of virus-induced membrane alterations are described for various members of +RNA viruses, including Picornaviridae (Poliovirus), Togaviridae (Semliki Forest virus), Flaviviridae (West Nile, Dengue, Hepatitis C virus), Coronaviridae (SARS coronavirus), and Nodaviridae (flock house virus); (reviewed in (1-4)). Plant and insect viruses are known to modify the membranes of ER, peroxisomes and mitochondria (5-7). Among + RNA viruses that infect animals, Togaviridae (discussed later in this review) are the only one group known to replicate in association with the membranes of the endo/lysosomal compartment (8-10). For the remaining viruses so far studied, viroplasms seem to originate from ER membranes (11-15). The morphology of these compartments is complex, with clustered single- or double-membrane vesicles of different size, and convoluted membranes forming a dense network. Recently, advances in fluorescence and electron microscopy have allowed researchers to reveal the three dimensional organization of such complex compartments. A combination of immunolabelling and electron tomography (16) has been applied to study the membrane alterations induced in cells infected by flock house, SARS and Dengue virus (17-19). In these three studies, viral replicase proteins and newly synthesized RNA (labelled in different ways) 10

were localized within the membrane rearrangements. Importantly, in the case of Dengue and flock house virus (which utilize the membranes of ER and mitochondria, respectively) three dimensional reconstructions revealed that the interior of all the imaged vesicles was connected to the cytosol through ‘pore-like’ structures of approximately 10 nm in diameter (17;19). Similar structures had been found for different viruses in other studies using conventional EM. The finding constitutes the basis for the model of spherule functioning during viral RNA synthesis: the interior of the spherules would concentrate (and protect) replicase proteins and RNA templates, whereas the pore would allow import of cellular molecules required for RNA synthesis (e.g. nucleotides), and export of newly synthesized viral RNAs (e.g. for translation and packaging) in the cytoplasm (Figure 1). To date, the formation of such structures in vitro, using merely viral proteins, RNAs and membranes, has never been reported. On the contrary, an increasing amount of evidence suggests that the biogenesis of such unique ‘organelles’ is assisted by cellular proteins (20-23). However, the identity and precise biological role of viral and cellular components involved in the process are not yet clear. Basic questions that remain to be answered include: 1. What is the source of membranes? 2. Which viral and/or cellular components constitute the minimum requirement for the formation of these structures? How do they work together? 3. How are the viral replicase proteins and RNAs targeted to the membranes? 4. How are the different steps of virus genome replication/transcription orchestrated in the context of the membranous structures? In our research group we use Semliki Forest virus (SFV) (genus Alphavirus within Togaviridae family) as a model system to address these questions.

Fig 1. Schematic model for + RNA virus genome replication within membrane vesicles. The interior of the vesicle is connected to the cytosol through a pore-like structure. Viral replicase proteins and RNAs are concentrated inside the vesicle, and the pore allows the entry of molecules required for RNA synthesis (e.g. NTPs), and release of newly synthesised RNA into the cytoplasm for translation and packaging. 11

1.2 Alphaviruses

Alphaviruses (Togaviridae) are distributed throughout the world and can infect both invertebrate and vertebrate organisms. A persistent and asymptomatic infection is established in insects (typically mosquitoes), which are used as vectors to infect different vertebrate hosts such as birds and mammals. Most infections are cleared by the organism or result in mild symptoms like fever and rash. However, some alphaviruses can also cause painful arthralgia and arthritis, long lasting fever, and even fatal encephalitis when the virus passes the blood brain barrier and infects the central nervous system (24;25). Recently, an outbreak of Chikungunya virus started in the coasts around the Indian Ocean and reached Europe, resulting in more than two millions human infections and several hundred casualties (26). No vaccines or effective antiviral treatments are currently available against alphaviruses.

Fig 2. Genome organization of SFV and schematic representation of viral proteins and their functions. The genome is capped and polyadenylated, and has the same coding polarity as cellular messenger RNAs. The first open reading frame encodes for the replicase polyprotein precursor, P1234. The second gene encodes for the precursor polyprotein of the structural proteins. Both polyprotein are proteolyticaly processed by viral and cellular proteases. Cleavages are indicated by scissors of different colours depending on the protease involved. Functional domains are indicated for each of the ns-proteins. Different shades of colours indicate functionally distinct regions in the respective protein. C=capsid, E=envelope glycoprotein.

The genus Alphavirus contains at least thirty members which can be grouped in four lineages: the ‘Old World’ group contains two lineages of which SFV and Sindbis (SIN) virus are the prototype members respectively. Eastern, Western and Venezuela equine encephalitis viruses constitute the ‘New World’ lineage. The forth lineage includes only the recently described Salmonid alphavirus (SAV). In the following text I will use SFV as a representative of the genus. Like other alphaviruses SFV is a small enveloped virus with a diameter of 70 nm. The genome is a single stranded RNA molecule of approximately 11.5 kb, with a 5’-cap structure, a 3’-polyadenylate sequence and the same coding polarity as cellular messenger RNAs (positive polarity). The 42S RNA genome is packed by 240 copies of a single capsid protein (C) and further enveloped by a membrane acquired during the budding process and containing the spike glycoproteins (E1, E2 and E3). The virion has icosahedral symmetry with a triangulation number of four (27). The genome contains two open reading frames (ORFs). The 5’-two thirds encode the replicase polyprotein, while the 3’-one third codes for the virion structural proteins (24) (Fig 2).

1.3 SFV replication cycle

At the beginning of infection, alphaviruses bind to a receptor on the surface of host cells. The identity of this receptor has been elusive although recent data suggest that it could be DC-SIGN (dendritic cell-specific ICAM-3-grabbing non-integrin) (28). The binding of the virion to the receptor induces clathrin-mediated endocytosis, resulting in the internalization of the virus in intracellular carrier vesicles, which deliver the cargo to early endosomes (Fig 3, step 1). The progressive acidification of the endosomal lumen triggers a change in the conformation of the envelope spike glycoproteins, which mediate the fusion of the viral envelope with the limiting membrane of the endosome, releasing the nucleocapsid in the cytoplasm of the infected cell (29). Interaction of nucleocapsid and ribosomes seems to assist the genome uncoating (30;31). Once in the cytoplasm, the 5’- first ORF of the viral messenger-RNA-genome is translated giving rise to the nonstructural polyprotein P1234, the precursor of the viral replicase (Fig 3, step 2) (24;32). In SFV, three sequential proteolytic cleavages yield four proteins, nsP1-nsP4, all involved in RNA replication as discussed below. The cleavages, catalyzed by the C- terminal moiety of nsP2, are highly regulated and the cleavage intermediates play a crucial role during different steps of RNA synthesis (Fig 3, step 3) (33-35). The first cleavage releases nsP4, the actual RNA dependent RNA-polymerase (RdRP), and the precursor P123. This first event produces the early replicase, which copies the 42S genomic RNA into a complementary strand, by definition designated minus-sense RNA (32;36;37). The second cleavage releases nsP1 and the precursor P23. This event is crucial because it switches the replicase from negative to mainly positive-sense RNA production: the recognition of a promoter sequence in the 3’-terminus of the minus strand results in synthesis of genome-size 42S RNAs that serve as messenger for further synthesis of the P1234 polyprotein capable for initiating further rounds of replication. At later stages of infection, these molecules will be packaged into new virions. The cleavage of P23 releases 15

nsP2 and nsP3. This event seems to redirect the replicase specificity to an internal subgenomic promoter (SGP) in the minus sense RNA, resulting in the synthesis of a shorter 26S positive sense RNA molecule corresponding to the 3’-one third of the genome. Like the genomic 42S RNA, this molecule is also capped and poly-adenylated, and serves as messenger RNA for the production of the structural proteins. Only the 42S RNA can be packaged into new virions due to a packaging signal within the nsP2 encoding sequence (24;32). Not all the synthesized nsPs are engaged in RNA synthesis; a sizable fraction accumulates in the cytoplasm where the single nsPs interact with and influence different cellular components. During the first hours of infection, genome size negative- and positive- sense RNAs are synthesized. Within 4 to 6 hours, depending on the multiplicity of infection, the synthesis of minus-strand is shut down, and no new replication complexes are formed. This is due to the combined effect of three events: i) cleavage-mediated conversion of the replicase specificity from minus to plus RNA synthesis; ii) accumulation of nsP2 in the cytoplasm results in faster cleavage of nascent P1234, preventing the formation of new RCs iii) progressive inhibition of host transcription and translation, the later resulting in inhibition of new P1234 synthesis. Shut-down of host macromolecular synthesis occurs at least in part through nsP2 (38-40). Whereas the mechanism of RNA synthesis inhibition is still unclear, the block of translation was attributed to the nsP2-mediated induction of eIF2a phosphorylation (41). Under these conditions, however, the production of structural proteins still continues, due to the 5’-end of the 26S RNA, which serves as an eIF2a- independent translational enhancer (41-43). The virus-specific RNA synthesis continues in the previously formed replication complexes at a linear rate throughout the infection cycle. In the absence of newly synthesised replicase proteins, 42S positive sense genomes become available for encapsidation ensuring production of new progeny virions (24;44). As for other +RNA viruses, at least some of the above mentioned events take place in association with intracellular modified membranes (4). In the following, I will focus on the still poorly understood interplay between viral replicase components, RNA replication and intracellular membranes.

Fig 3. Replication cycle of SFV in the cytoplasm. (1) After clathrin mediated endocytosis and delivery of the virion to endosomes, the nucleocapsid of SFV is released into the cytoplasm by pH triggered fusion of viral and endosomal membranes. (2) In the cytoplasm, ribosomes assist uncoating, followed by translation of viral genomic mRNA. The replicase precursor polyprotein P1234 is produced and undergoes auto-proteolysis. (3) The cleavage products P123 and nsP4 (the RdRP) form the ‘early’ replicase, capable of recruiting the genomic +RNA and synthesizing genome-size minus-strand RNA. It is assumed that this process leads to the formation of a double stranded RNA intermediate. Further cleavages are required to form the ‘late’ replication complex, which can use the minus strand as template to make more plus strands. The products of the fully processed polyprotein are also able to recognize in the minus-strand RNA an internal subgenomic promoter for the production of 26S +RNA, which is used as messenger RNA to produce structural proteins. The newly synthesized 42S +RNAs are used for further rounds of translation and replication, and at later stages of infection (4) packaged into new progeny virions.

1.4 General composition of alphavirus replication complex

As mentioned above, the four nsPs work as a complex. The main pieces of evidence supporting the existence of a ‘replication complex’ are listed below: 1. mutations that influence RNA synthesis are found in each of the four nsPs (45-47) 2. mutations in one of the four nsPs that affect RNA synthesis can be attenuated or suppressed by mutations in other nsPs (24;32) 3. all the four nsPs can be co-immunoprecipitated, after non-ionic detergent solubilisation of membranes, using anti-sera specific for any of them (48) 4. a fraction of the four nsPs colocalize with each other and with RNA replication- intermediates during immuno-cytochemistry experiments in infected cells (49;50) 5. when imaged by immuno-electron microscopy, each of the nsPs colocalizes with bromouridine in the proximity of alphavirus-specific membrane alterations (49) 6. expression of nsP4 in cells does not results in efficient template-dependent RNA synthesis. The reaction is rescued by the concomitant expression of the polyprotein P123, which also provides membrane targeting (8;37;51;52) Although nsP4 has been shown to catalyze template-dependent RNA synthesis in the absence of other nsPs in vitro (53), this reaction is inefficient in vivo (our unpublished data). Moreover, in cell-based experiments, RNA replication cannot be initiated if the nsPs are provided separately. The polyprotein precursor is essential for targeting the complex to the correct membranes and, presumably, for RNA-template recruitment (8;51). On the contrary, each of the nsPs localises to different compartments when expressed alone: nsP1 associates with the PM where it induces the formation of extensive and branched filopodia-like protrusions (54); nsP2 can be detected in the cytosol and in the nucleus (9). NsP3 forms large aggregates that colocalize with markers of cellular inclusion bodies (8). NsP4 is cytosolic and rapidly degraded by the proteasome according to the N-end rule (55). All this data support but do not demonstrate the existence of a macromolecular complex containing the four replicase proteins. A molecular model of such a complex based on structural information is to date missing. We can therefore only infer the role that each of the nsPs plays during RNA replication based on the above mentioned information and their biochemical properties as single proteins. In fact, much knowledge has been gained on each of the nsPs by the characterization of their enzymatic activities (Fig 2) (reviewed in (32)).

NsP4 (614 aa) is the RdRP. Although the presence of common RdRP sequence motifs has been known for long a time (56), and genetic evidence that this protein was involved in RNA synthesis has been reported more than two decades ago (57), only recently this activity was demonstrated: i) unspecific terminal transferase activity was shown for the purified C-terminal domain of the protein (58); ii) using crude cellular extracts obtained from cells expressing nsP4 and P123, a fraction containing only nsP4 did support template dependent RNA synthesis (53). These experiments confirmed that the core polymerase is an active enzyme but do not provide information on how the other nsPs regulate this

function, as indicated by the large amount of genetic evidence in vivo. NsP4 does not have membrane binding properties.

NsP1’s (537 aa) N-terminus domain comprises the methyl-transferase region involved in an RNA capping reaction specific for alphaviruses: the enzyme first methylates GTP and then (presumably) transfers it to the 5’-beta phosphate of the RNA via an m7GMP- enzyme intermediate (59-64). NsP1 does not have RNA triphosphatase activity; instead nsP2 harbours this function as explained below. Interestingly, the activity of nsP1 is abolished by detergents and restored by the addition of phospholipids, phosphatidylserine being the strongest activator, indicating that this protein is active when associated with membranes of appropriate composition (65). Indeed, membrane association has been demonstrated for this protein, which is the only one among the nsPs that can serve as a membrane anchor. Two regions are particularly important for the membrane binding: an ‘amphipathic peptide’ in the middle of the methyltransferase region plays the major role, whereas palmitoylation of cysteines 418-420 seems to have an accessory role which is not crucial for membrane binding or virus viability. In contrast, single point mutations in the amphipathic peptide can disrupt membrane binding in vitro, change the localization of nsP1 or un-cleavable P123 precursors in cell systems, and result in nonviable viruses (54;66-68). Moreover, point mutations in the same region also inactivated the methyltransferase activity of the protein (65). This finding, together with the requirements of phospholipids in vitro, suggests that during infection the capping reaction could be prevented in the cytosol, and that the amphipathic peptide could play a regulatory role serving as a membrane-sensor. Indeed, the peptide changes its conformation during membrane binding (69).

NsP2 (799) is a multifunctional protein; protease- (70-72), NTPase- (71;73), RNA- triphosphatase- (71;74) and helicase activities (75) have been demonstrated. Evidence also supports a role for this protein in sub-genomic promoter regulation and virus pathogenicity (76-78). The sequence of nsP2 is generally divided into two regions: the N- terminal half contains sequence motifs typical for the SF1 family of helicases (79). The C- terminus of the protein represents the protease (80). The protease activity is responsible for the three cleavages of the polyprotein P1234. Site directed mutagenesis has confirmed that nsP2 is a thiol-protease (32). The recombinant purified C-terminal half of the protein was shown to retain efficient proteolytic activity for the nsP3/nsp4 recombinant cleavage site (70). The cleavages of the other sites (nsP1/nsP2 and nsP2/nsP3) seem to be more complicated, and have different requirements in different alphaviruses (35;71;81). Our group has recently characterized the molecular mechanisms that regulate these events, and highlighted similarities as well as important differences during the polyprotein processing of two distantly related alphaviruses (Dr. Andrey Golubtsov’s still unpublished results; thesis 5, 2008). Using recombinant purified proteins, it was also shown that the N-terminal fragment of nsp2 possesses NTPase- and RNA-triphosphatase activities, but only the full protein displayed moderate helicase activity (73-75). These reactions are assumed to be important

for RNA capping and synthesis, although no direct evidence has been provided so far. Consistently, mutations of the NTPase active site are lethal for the virus. A large array of temperature sensitive mutants was generated by pioneering studies during 1970s using both SINV and SFV (45-47). Of those that impaired RNA synthesis, many were subsequently mapped to the nsP2 region (82). Of particular interests is the fact that mutations in each of the two domains of the protein resulted in similar phenotypes: when cells were infected by mutant viruses at the restrictive temperature no RNA synthesis occurred. However, if the infection was carried out at the permissive temperature for more than 5 hours, subsequent shift to the restrictive temperature could not stop genomic RNA synthesis or virus production, but severely hampered 26S RNA production. Therefore a mutation in either of the two main domains of nsP2 could cause a change in sub-genomic RNA production. In one of their elegant studies, Simmons and Strauss reported that an RNase sensitive region existed at about two thirds of the genome RNA sequence in a fraction of the replicative intermediate (RI) double-stranded RNA molecules extracted from SINV infected cells (83). Later Sawicki et al. used SFV and reported of a temperature-sensitive mutant for which the RNA sensitive region was lost when infections were shifted at the restrictive temperature. This coincided with the loss of sub genomic RNA synthesis, while synthesis of 42S genomic RNA continued. This phenotype was reversible: when cultures were shifted back to permissive temperature 26S RNA restarted, and after RNA extraction the RNase sensitive region reappeared (84). From these data the authors proposed a model whereby a viral factor, different from the polymerase core, could interact in a temperature sensitive manner with a region of the minus strand preceding the initiation of 26S RNA, thereby preventing polymerization from the 5’ to proceed and allowing initiation of 26S RNA synthesis. Thus, this interaction would have produced a single stranded region sensitive to RNase after extraction. The ts-mutation was subsequently mapped by the same group in the sequence of nsP2 (76). Different models could be proposed to explain this phenomenon, but in the absence of direct experimental evidence the issue remains a matter of speculation. In fact, whether nsP2 specifically binds to regions overlapping with or close to the SGP has never been demonstrated.

NsP3 (482 aa) is the smallest of the nsPs; although it is involved in RNA replication its biological role remains mysterious (85). This protein is comprised of three main regions: the N-terminal 170 residues contain the “macro domain”, the structure of which has been recently resolved for Chikungunya and Venezuela Equine encephalitis virus (86). This region is highly conserved among alphaviruses but it also displays a high degree of similarity with proteins from other viruses and even eukaryotic cells. Deletions in this region are lethal for alphaviruses but its exact role remains unknown. The central part of the protein is specific for alphaviruses; no structure homology exists for this region. Recent biochemical experiments have shown that this part of the protein in important for polyprotein processing. The C-terminus is not conserved even among alphaviruses and it has been predicted to be unstructured. This region is highly phosphorylated at clusters of Ser and Thr residues, and it contains many proline repeats and acidic residues (87;88).

Short deletions, insertions or mutations in this region are tolerated but often result in attenuated or nonpathogenic viruses (89;90). The role of nsP3 as a pathogenicity factor is also supported by the fact that the differences between a pathogenic and nonpathogenic strain of alphaviruses often reside in the C-terminus of this protein (25). Although some of these changes result in reduced level of replication and virus production in cell culture, some do not (91), thus suggesting that beside the rate of RNA replication and virus production, other mechanism exist that regulate pathogenicity and nsP3 plays a crucial role. Nsp3 also influences the localization of the replication complex: whereas expression of recombinant un-cleavable polyprotein precursor P12 results in plasma membrane localization, addition of nsP3 results in the targeting of endo-lysosomal membranes (8). Therefore in the context of replication complexes, nsP3 could act as a regulator of membrane targeting, providing signals for intracellular trafficking of the nsP1-membrane- anchored RCspherules. In this thesis I will provide evidence that this is indeed the case.

1.5 The spherules of alphavirus: ‘mini-organelles’ for viral RNA replication

The spherules, as commonly named by ‘alpha-virologists’, are the smallest membrane alterations detected during alphavirus infections. They were first reported in two pioneering electron microscopy studies published by different groups in 1967. The authors provide electron micrographs showing 50-70 nm membrane invaginations (“vesicles”) on the limiting membrane of 0.2-2.0 µm electron translucent vacuoles present in the cytoplasm of infected cells. “The vesicles are bounded by a single membrane, which in some cases is continuous with the membrane of the surrounding vacuole. In addition, the vesicles show an irregular dark central spot in an otherwise lightly staining interior” (from pag. 139 in ref.(92) ). The large vacuoles where spherules accumulate at late stages of infection were later designated as cytopathic vacuoles type one, CPV-I (Grimley, Berezesky and Friedman, JVI, 1968 (93); the term ‘spherule’ first appeared in the same article). For convenience, the abbreviation CPV for CPV-I will be used in the following text. Already at this early stage, the CPVs were associated with virus specific (Actinomycin-D resistant) RNA synthesis. More specifically, Grimley and co-workers used autoradiography of infected cells pulse-labelled with 3H-uridine to localize the sites of viral RNA synthesis. The label was associated with CPVs but, interestingly, also the plasma membrane (93). Twenty years later, Froshauer et al. demonstrated that the CPVs contained markers of cellular endosomes and lysosomes (10). In this important contribution, the authors also provided electron micrographs where spherules were characterized by finer ultrastructural details, showing that: i) spherules clearly represented invaginations of the limiting membrane of the larger vacuoles, the CPVs; ii) inside the spherules an electron-dense, star-shaped structure with thin spokes radiating from the central mass was reproducibly detected; iii) the interior of the spherule was connected to the cytoplasmic space by a pore, 21

or membranous ‘neck’ (outer diameter 19-20 nm; inner diameter 8-9 nm); iv) thread-like electron-densities were apparently coming out of these pores. Ribosomes and nucleocapsids were often associated with the filamentous material. Immuno-cytochemistry confirmed also the presence of viral replicase components. Taken together, these and the previous data strongly indicated that the spherules represented the site of viral RNA replication. Spherules were supposed to form on the limiting membrane of pre-existing cellular endosomes. More specifically, at early stages of infection the CPVs had characteristics of endosomes but later they behaved more like lysosomes. In fact, at late stages of infection, virtually all the lysosomes of the cell had been converted to CPVs. In a series of follow-up studies, our group showed, using replicon-RNA transfections, that the formation of spherules was dependent on the synthesis of nsPs and that virus entry and the presence of structural proteins were not necessary for spherule and CPV formation (94). The endo-lysosomal nature of the CPVs was confirmed; using EM, spherules were also found inside endocytic processes at the PM (49). EM-immunocytochemistry of bromouridine-pulsed infected cells showed that newly synthesized viral RNA and replicase proteins were unambiguously colocalized at the limiting membrane of the CPVs, in close proximity of the spherules (49). In this study, it was proposed that the spherules could originate at the PM and that CPVs could form as the result of spherule internalization. By expressing different combinations of the nsP polyprotein in mammalian cell culture systems, it was also shown that the membrane binding protein nsP1 or the combination nsP1-nsP2 localized at the plasma membrane, whereas the addition of nsP3 moiety resulted in targeting of P123 to both plasma membrane and endosomes (95).

1.6 General principles of endocytosis

The term endocytosis appeared in the early 1960s, along with a plethora of similar terminology (e.g. intracytosis, cytosis, etc.). To encompass the multitude of morphological varieties of this basic cellular process, de Duve in 1963 referred to endocytosis as the general process of internalization of extracellular material within an invagination of the plasma lemma (rephrased from (96) ). In light of the current knowledge, this definition had to be adjusted. In one of the most recent reviews on the topic, endocytosis is described as “de novo production of internal membranes from the plasma membrane lipid bilayer. In so doing, PM lipids and proteins, and extracellular molecules become fully internalized into the cell” (97). Indeed, it is now clear that the process of PM internalization not only serves to internalize extracellular material. It is also a means for the cell to regulate the composition of the PM, which plays crucial roles in a variety of fundamental processes such as mitosis, cell migration or antigen presentation. Thus, a PM receptor can be endocytosed (or ‘downregulated’) in the presence or absence of its ligand. In this way cells can modulate their responsiveness to different extracellular stimuli.

Pathogens have evolved various strategies to exploit different endocytic routes. Many viruses use endocytosis to access the intracellular space (98;99). Furthermore, it is becoming apparent that after entering cells, at least some viruses can activate different endocytic events as a means to modify the composition of the PM, thereby interfering with cellular defence mechanisms (e.g. antigen presentation) (100;101).

Endocytosis is a highly regulated event, in which a large number of cellular proteins sequentially interact with each other and with the membranes, to coordinate in time and space the internalization event. Historically, the morphology of the membrane invagination, which can be visualized by EM, has been used as the first criterion to define the identity of a specific endocytic event. In this way, PM protrusions are distinguished from invaginations, and vesicles from tubules. Further distinctions have been made based on the presence or absence of electron dense materials surrounding the membranes of certain vesicles at the stage of invagination. These layers correspond to protein coats, which play important roles in vesicle formation and stabilization of membrane curvature. As described below, not all protein coats are easily visualized by conventional EM. Moreover, certain structures can be very sensitive to fixation conditions, which may have hampered their discovery. An accurate morphological analysis not only provides means for classification, but it is also useful for the construction of models to describe the sequence of molecular events which drive membrane deformation. Live-cell imaging techniques, on the other end, provide information on the dynamics of the process. In the following, I will introduce the major endocytic routes so far recognized. The scope of this short review is to describe the general principles that lead to the formation and internalization of an endocytic vesicle, highlighting common features and main differences between the pathways.

Endocytosis assisted by clathrin

A myriad of molecules are internalized by clathrin-mediated endocytosis (CME), including low-density lipoprotein (LDL) receptor, tyrosin-kinase receptors (RTK), anthrax toxin (Doherty and McMahon, 2009, Conner and Schmid, 2003) (97;102). SFV was the first virus shown to enter cells via CME (98;103). It has been estimated that in fibroblasts thousands of clathrin coated vesicles (CCV) can be internalized from the plasma membrane every minute (104). The clathrin coat assembles at the cytoplasmic side of the plasma membrane (clathrin coated pits) at sites where internalization will occur (hot- spots). The major component of the coat is the clathrin heavy chain (190 kD), and the clathrin light chain (25 kD), which assemble in three-legged structures called triskelia. Each triskelion constitutes the structural unit of the clathrin coat, which at the end of the polymerization process forms a basket surrounding the membrane of endocytic vesicles with typical diameters of ~100 nm (105). The clathrin coat provides a scaffold for membrane deformation and budding, although theoretical considerations indicate that clathrin polymerization alone is not sufficient for inducing membrane curvature (106). In order to form at correct sites, clathrin coats must be recruited by adaptor protein

complexes (e.g. AP2, b-arrestin, epsin1). Other accessory proteins participate in different steps of CCV formation, either as scaffolds to stabilize the different components, or by adding membrane curvature (e.g. eps15, amphiphysin, dynamin, etc.) (97;107). The adaptors recognize short amino acids sequences, or motifs, on the cytoplasmic side of PM receptors. With the aid of accessory proteins, receptors are clustered and internalized by the CCV. At the end of the process, the large GTPase dynamin mediates the scission of the vesicle, which after coat disassembly delivers its cargo by fusing with endosomes (108-110). Thus, CME could be classified as a localized event, occurring through progressive clathrin polymerization at precise sites of the PM, where adaptor proteins interact with their cargo molecules. The estimated time for the internalization of a CCV in cell culture systems is ~1 minute. Recently, different modes of CME, characterized by slower kinetics of internalization and larger clathrin-coated plaques, have been described. These mostly occur at the PM attached to the substrate, and are strictly dependent on actin polymerization (111).

Caveolae mediated endocytosis: caveolins, cavins and flotillins

Caveolae are flask-shaped invaginations of the PM, with a diameter of 60-70 nm. The main difference from CCV is that caveolae exist at the PM in the absence of a stimulus (or cargo). There, they remain static and with a relatively wide opening (as wide as the diameter of the vesicle) to the extracellular space (ref Helenius article and JCS artile) (112;113). Caveolae are classified as coated structures. The main components of the coat are the proteins caveolin-1 (cav1) and cavin-1 (or PTRF), present in a 1:1 ratio (114). It has been estimated that each caveola contains ~150 cav1 proteins (115). Differently from the clathrin coat, when imaged by conventional EM the membrane of caveolae appears smooth, although in some micrographs a thin spike-like coat has been shown (116), and by scanning-EM (SEM) the cytoplasmic side of caveolae appears surrounded by filamentous material arranged in spirals (117). Multi-caveolar (or ‘grape-like’) structures have been described, and are often present in regions of the PM rich in caveolae (118). Their number and size can be increased by treatments with phosphatase inhibitors (e.g. okadaic acid) (119). The physiological roles as well as the mechanism of formation of caveolae are still a matter of debate. They have been implicated in different cellular processes, including cell signaling, cell adhesion and cancer, lipid regulation, and transcytosis (perhaps the best characterized event) (120). Live-cell imaging studies have provided evidence that at least certain molecules (e.g. viruses, toxins and membrane proteins) can be internalized by caveolae-mediated endocytosis (121-123). However, the molecular events that induce caveolae internalization are much less characterized that those of CME. It appears that upon binding of cargoes to PM receptors, a ‘signal’ is sent into the cytoplasm resulting in the activation of Src and other kinases, local actin remodeling and caveolae internalization (124). As in the case of CCV, the scission of the vesicle is also dependent on dynamin (125). Interestingly, and differently from CCV, internalized caveolae recycle back to the PM, and caveolins do not seem to disassemble from the vesicle during this traffic. Directed transport to more internal endocytic

compartments occurs upon engagement of long distance microtubule based transport. Kinases have been shown to regulate the switch between the recycling and the long distance transport events (115). Whether fusion with endosomes precedes the long distance transport has not been investigated. The role of cavin proteins has not been fully characterized yet, although they seem to be involved in caveolae biogenesis (114;126;127). Recently another class of proteins, flotillins, has been reported to form endocytic structures morphologically similar to caveolae (128-131). Flotillins do not colocalize with caveolins or clathrin, and thus it has been proposed that these proteins could mediate a different endocytic event (128). More experiments will clarify the role of these proteins and the interplay with caveolins. Whether caveolae endocytosis is a localized event (one piece of cargo, one internalization event) or whether the activation of kinases mentioned above can result in the induction of internalization events over large areas of the PM is not firmly established.

Macropinocytosis and phagocytosis

Cells do not only intake portions of their PM through invagination processes. Some endocytic events start with the formation of PM protrusions supported by actin polymerization, which eventually fuse (with each other or with the rest of the PM) resulting in the formation of large cytoplasmic vacuoles (macropinosomes and phagosomes) (132). The term macropinocytosis refers to the uptake of large amounts of liquid from the extracellular space. Phagocytosis, instead, is used for the internalization of large particles, like bacteria or even other cells. In both cases the PM protrusions extend for up to several micrometers, depending on the process. Active macropinocytosis is visualized by the formation of PM ‘ruffles’, which can take the form of blebs, planar folds (lamellipodia) or cup-shaped membrane extensions (133). These structures are characterized by large variations in size and shape, which do not depend on the size of the cargo. In fact, macropinocytosis can be triggered by the attachment of hormones (e.g. growth factors) to their receptors. No protein coat is involved in the process. On the other hand, the induction of phagocytosis, and the size of the resulting phagosomes, strictly depends on the presence and size of the particle to be internalized. The PM protrusions, or ‘phagocytic cup’, extend following the contour of the particle; this process is driven by the coordination of actin polymerization (to push the membrane) and the interactions of numerous PM receptors over the entire surface of the ingested particle (133). Thus, cargo (fluid) uptake through macropinocytosis is non-specific, whereas particle internalization through phagocytosis is specific. All cells could potentially activate macropinocytosis, while only specialized cells (e.g. macrophages) can phagocytose particles. Importantly, whereas phagocytosis is a localized event, macropinocytosis can be triggered by the binding of a few (perhaps one) ligands with a receptor, and involves actin remodeling and ruffle formation over the entire surface of the PM (134). The scission of macropinosomes and phagosomes does not involve dynamin. In both cases, actin remodeling is initiated after a multi-branched cascade of signaling events triggered by the binding of cargoes (hormones, or larger particles) to their receptors. For instance, in the case of

macropinocytosis, upon binding with the cognate ligand, RTKs dimerize and activate kinases (including Src and PI3K), Rho GTPases (Rac1), and proteins involved in membrane trafficking (e.g. Rab5 and Arf6). These events occur sequentially, and regulate membrane protrusion, closure and intracellular trafficking of the macropinosomes. Similar but localized events are involved during phagocytosis, although apparently different Rho GTPases are involved in different stages of the process (132;133).

Other types of endocytosis

Endocytic events which differ from those mentioned above, both morphologically and in terms of cellular proteins involved in their regulation, have been described. These include the internalization of IL-II receptor, mediated by Rac1, RhoA, dynamin and cortactin (135-138). The morphology of this event has not been characterized. Internalization of glycosyl-phosphatidyl inositol (GPI) anchored proteins is been shown to occur, at least in part, through tubule-vesicular invaginations of the PM induced after activation of the Rho- GTPase Cdc42 (128;139;140) and regulated by the GTPase-activating protein GRAF1 (141;142). Recently, the infectious entry route of SV40, previously thought to be mediated by caveolae endocytosis, has been attributed to a ‘zipper-like’ invagination process mediated by stoichiometrical interactions between the viral capsid protein VP1 and the GM1 receptors (143), thus not involving any of the above mentioned endocytic pathways. Intracellular trafficking of the internalized virion is currently under investigation (Ari Helenius personal communication). Other mechanisms of PM internalization have been proposed and more are likely to appear in the future, as the network of interacting proteins is rapidly being deciphered.

2. Aims of the study

1. to develop a method to obtain a high resolution 3D-structure of RC spherules 2. to determine the role of different intracellular membranes in the formation and maturation of RC spherules 3. to characterize the molecular events that lead to the formation of the CPV-I 4. to determine the role of nsP2 in the context of RCs and RNA replication

3. Methods

Most of the methods used in this study are described in detail in the original articles as indicated in the following table.

Article Cell culture BHK-21 cells I II III Hela cells I II MEFs I Huh7 cells II Cav1-GFP cell line II Virological and cell culture methods Infections with VRPs I Growth curve studies I Infections and drug treatments I II Viral RNA labelling I III Transfections and siRNA studies II Immunological methods Western blotting I II Indirect immunofluorescence I II Immunoprecipitation III Microscopy Light microscopy I II TEM I II SEM I CLEM I II Tomograpy II Protein studies Protein expression and purification III Protease assay III NTPase assay III RTPase assay III Protein labelling III Preparation of P15 fraction III DNA and RNA methods DNA cloning I III RNA synthesis III Data analysis ImageJ I II

ImagePro I TILLVisION I AutoQuant I II Imaris Bitplane I II Aida Image Analyzer III Fuji BAS-1500 III

Antibodies and reagents Article anti-nsP1 III anti-nsP2 III anti-nsP3 I II anti-dsRNA (J2) I II anti-α-tubulin I anti-γ-tubulin I anti-β-actin II anti-clathrin heavy chain II anti-caveolin-1 II anti-EEA1 II anti-LBPA II anti-GM130 I phalloidin-AF568 I II Tf-AF647 II CTxB-AF647 II Lysotracker Red DND-99 I Filipin complex II (-)Blebbistatin I Wortmannin I LY294002 I PI-103 I Nocodazole I Methyl-β-CD II ROCK inhibitor Y-27632 II IPA-3 II Genistein II Dynasore II PP2 II Bisindolylmaleimide (GF109203X) II Akt inhibitor VIII II RacI inhibitor II

Plasmid constructs Article pSFV4 I II pSFV1-ZsG I II pSFV4-ZsG I II pSFV4_nsP3Δ50 II

Other methods

EEA-1 and TfR localization by immunocytochemistry (Fig 11): BHK or HeLa cells were infected with SFV (moi 50) in MEM containing 0.2% BSA. After fixation with paraformaldehyde (PFA) cells were permeabilized with 0.1% TritonX-100 in PBS, for one minute at room temperature. EEA-1, TfR were detected by incubation with a rabbit polyclonal anti-EEA-1 antibody (kindly provided by Dr. Marvin Fritzler, University of Calgary, Alberta, Canada) and a human monoclonal anti-TfR antibody (MAbs B3-25, BoehringerMannheim). Alexa fluor-conjugated secondary antisera raised against rabbit and human antisera (Sigma), respectively, were used in fluorescent microscopy to visualize the proteins.

Tomography (Fig 10): SFV infected BHK cells (moi 50) were either fixed at 6 hpi in 0.2% glutaraldehyde (GA) and processed for conventional transmission electron microscopy as described in (II), or high-pressure frozen using a Leica EM Pact High Pressure Freezing device (Leica Mikrosysteme Gmbh, Austria) and freeze-substituted in a Leica EM AFS Freeze substitution unit (Leica Mikrosysteme Gmbh, Austria). For tomography, specimens were imaged essentially as described in (II) at increasing tilt angles (-65 to +65 degrees), with step increments of 1 degrees. Images were processed with the Imod software package (http://bio3d.colorado.edu/imod/) and 3D-models obtained with Amira (Visage Imaging).

RNA labelling and autoradiography (Fig 12): This experiment was performed as described in III)

Correlative cryo-electron microscopy (Fig 7 and 8): A plasmid containing the gene of nsP1 fused with the GFP coding sequence was transfected in BHK cells using Lipofectamine LTX (Invitrogen) for 12 hours according to the manufacturers’ instructions. The cells were grown on perforated-carbon coated EM grids for correlative microscopy (C-flat, CF-2/1-2AU-F1, EMS-Protoships). Cells were imaged first by fluorescence microscopy (I) and subsequently frozen by plunging in liquid ethane. cryo- EM imaging was performed using a Tecnai-20 electron microscope (FEI) operating at 200 KV, and equipped with a cryo-holder.

4. Results

4.1 Approaching the high resolution 3D-structure of alphavirus replication complex: from conventional EM to correlative cryo- electron tomography (Unpublished)

In order to obtain a high resolution structure of the spherules, a ‘zoom-in’ approach was taken: first the general 3-dimensonal distribution and morphology of the membranous structures was obtained by electron tomography of chemically fixed samples stained with osmium tetroxide and uranyl acetate. Two serial ‘thick’ sections (250 nm) were obtained from BHK cells infected with SFV replicon particles (VRPs; 50 infectious units per cell) and fixed at six hours post infection. Numerous CPVs were identified in each section. A representative CPV, having a large number of lining spherules and located in close proximity to ER membranes, was chosen for tomographic reconstruction (Fig 4 A). The CPV had a diameter of approximately 600 nm along the sectioning plane, and more than 500 nm on the vertical axis, such that the bottom and top of the modified endosome were not included in the two sections. This was an advantage because it allowed the electron beam to better penetrate the section, leading to higher quality images. The lumen of the CPV was translucent and did not contain internalized vesicles, although electron dense filamentous material was observed. The limiting membrane of the organelle was saturated with spherules, each appearing, after tomography, as an invagination of the CPV membrane. Importantly, a clear opening, or pore, connecting the spherule interior to the cytoplasmic space, could be distinguished for all the imaged spherules (Fig 4 A and B, arrowheads). Although generally spherical, none of the spherules was a perfect sphere. Moreover, the diameter also fluctuated, having an average of 60 nm (outer diameter) and fluctuations of up to 30%. Thus, under these imaging conditions, the limiting membrane of one spherule could not be perfectly superimposed with the membrane of another spherule. The size of the pore was also not constant, ranging from 10 to 15 nm. The interior of most of the spherules had electron-dense material. The centre often appeared as a dark core. In some cases, these densities could be traced throughout the pore of the spherule and continued in the cytoplasmic space (Fig 4 C). Spherules that did not contain internal densities often had a much wider pore. On rare occasions, two or more spherules were found fused sideways, such that the interior of one continued with that of the next. The CPV was almost completely surrounded by ER membranes (Fig 4 D). To assess how chemical fixation affected the morphology and staining of the spherules, the same experiment was repeated after cryo-fixation obtained by high-pressure freezing (ref (144)). To achieve higher contrast, specimens were stained with osmium, after hydrated-acetone substitution of the amorphous ice generated during fast freezing (see methods). Thin sections of resin embedded specimens were analysed. Different cellular compartments (including RER, mitochondria and nuclei) and cytoskeletal components displayed an overall higher morphological preservation. A CPV was imaged from a 90 nm section (Fig 5 A). The reduction in sample thickness allowed better electron 31

penetration and resulted in higher resolution. Given that the average diameter of spherules was 60 nm, and that they were equally distributed on the surface of a mature CPV, a 90 nm section was thick enough to contain many complete structures. Some of the spherules were only partially represented. The major difference compared with the chemically fixed sample was the better resolution of the dense core inside the spherules, which appeared as a convoluted filamentous material often denser in the central part (Fig 5 A-C). Also, the pore structure was much narrower, such that the membranes delimiting the pore were almost juxtaposed (Fig 5 B, arrowheads). The overall morphology was more regular than that obtained in the chemically fixed samples, with smaller variations in spherule size and changes in the curvature of the spherule membrane. None of the spherules were now found connected with each other, indicating that this could be an artefact induced by aldehyde fixation.

Fig 4. CPV and spherules imaged by EM and tomography after chemical fixation. A) BHK cells infected with SFV VRPs for 6h where processed for conventional EM after chemical fixation. A CPV with numerous spherules and surrounded by ER processes was chosen for tomography. The arrowheads indicate two spherules, each having an open neck to the cytoplasm. B) Higher magnification of a portion of the CPV in figure A shows several spherules. Four digital tomographic sections are shown. The arrowheads indicate the pore of the spherules. Note that the pores appear open only in some sections. C) 3D modelling of one of the spherule shown in B (white arrowhead). The spherule is shown in different orientations. In red the membrane of the spherule, in yellow the electron densities inside and outside the spherule. D) 3D-model of the CPV and ER processes shown in A. In the model only few spherules have been modelled (arrowhead). The membranes of CPV and spherule are shown in red, in yellow the membrane of the ER. 32

According to our model, spherules would contain at least viral proteins (nsPs), RNA (genome-size minus-strand at least) and cellular membranes. Since compact protein lattices and highly organized RNA-protein complexes are readily visualized by conventional EM as electron densities, spherules were qualitatively compared to viruses at different stages of budding. SFV and HIV-1 were included in this comparison. Under these conditions, the nucleocapsid of SFV reproducibly appears as a dense core of 30 nm, which contacts the PM at the site of budding. Stoichiometrical interactions with the cytoplasmic residues of the spike glycoproteins, already localized at the PM, are probably the most prominent inducers of membrane curvature. Thus, NCs are progressively enveloped and eventually bud into the extracellular medium (Fig 6 A). In the case of retroviruses such as HIV-1 or SIV-I, instead, monomers of the capsid protein Gag are membrane associated and by lateral interaction form a lattice that progressively bends membranes at the site of budding. This protein layer is visualized by a very strong osmium staining (Fig 6 B). Differently from the structures presented above, spherules do not have electron dense layers on either side of their membrane. They cannot be recovered from the extracellular medium, but remain connected to the PM or endosomal membrane through a narrow membranous neck. The electron dense filamentous material coming out of the neck is one of the most reproducible features of the spherules (Fig 6 C). On fortuitous sections, ribosomes and nucleocapsids are often associated with the cytoplasmic portion of the filamentous material. Thus the organization of proteins and RNAs within the spherules is very different from the highly ordered structure of SFV or immature SIV-1 virion.

Although different in many respects, both of the above mentioned methodologies can induce artefacts. To date, the only method by which biological materials can be imaged in their native state is cryo-electron microscopy (cryo-EM) (145). In this technique, hydrated specimens are fixed by fast freezing in liquid ethane, and imaged directly using low doses of electrons in a dedicated electron microscope in which the sample is maintained under liquid nitrogen temperature. Thus, no chemicals are used at any stage to fix or stain the material. For instance, cells can be grown in a monolayer on special EM grids, and subsequently processed for cryo-EM. This technique, in conjunction with tomography (cryo-electron tomography, CET) (146), could give accurate and reliable information on the distribution of cellular components, as all the electron densities represent biological material and not chemical staining. The major limitation is that only very thin (< 500 nm) samples can be imaged at high resolution. Thus, in order to image spherules using this method, either they should be purified or, if imaged in infected cells, they should be contained in a portion of the cell, which is thin enough to allow the electron beam to penetrate the sample. Most spherules are found on the surface of the CPVs. These modified endosomes have diameters ranging from 600 to 2000 nm, too large to be imaged by cryo-EM at high resolution. However, CPVs with smaller diameters (200-400 nm) have been detected.

Fig 5. CPV and spherules imaged by transmission electron microscopy and tomography after high pressure freezing, acetone substitution and osmium staining. A) A CPV is shown from a tomographic digital section. The membrane of the CPV is saturated with spherules, at a certain point connected to the limiting membrane of the CPV through a ‘neck’ structure (red arrowhead). Filamentous structures are visible inside and outside the spherules. Numerous electron densities in the cytoplasm close to the CPV resemble, in size and morphology, viral nucleocapsids (white arrowhead). B) Four spherules (1-4) were chosen from the tomogram for more detailed analysis. Four digital sections are shown. The neck of the spherules appears only on few sections, as indicated for spherule 1 (red arrowhead). At this higher magnification the interior of the spherule does not appear to be connected to the cytoplasmic space. The neck is a granular electron dense structure rather than a pore. C) 3D-models of the filamentous material inside the four spherules showing heterogeneous morphology.

Moreover, the fact that spherules can be also found at the PM suggested the possibility to image these structures directly in infected cells. In the first attempts, the major limitations of the technique became evident: i) only a very limited area of the cell cortex was thin enough to allow imaging (Fig 7 A); ii) the abundance of cellular organelles and cortical actin made an unambiguous identification of spherules and CPVs virtually impossible (Fig 7 A); iii) CPV preparations obtained after cell fractionation and differential centrifugation were still heavily contaminated by smooth membranes and vesicles of different sizes, again making the interpretation of the images extremely difficult (Fig 7 B and C). To better identify spherules and have a means to locate spherules in specific regions of the cell (or to distinguish CPVs obtained after cell fractionation from other co-purified cellular membranes), we planned to use correlative fluorescence cryo-electron tomography (CET), (147). To this end, a recombinant virus was created in which one of the replicase components (nsP3) was fused with a fluorescent protein (ZsGreen). The idea was to fast-freeze infected cells (or purified CPVs), and follow the fluorescence signal of nsP3_ZsGreen to obtain a ‘map’ that could be used in the electron microscope. This meant adapting a fluorescence microscope to work under liquid nitrogen temperatures, and image the frozen sample first by fluorescent microscopy and subsequently by cryo-EM and tomography (Fig 8 A). The feasibility of the approach was tested by imaging filopodia- like structures induced after expression nsP1-GFP chimera in BHK cells. After fluorescence imaging, filopodia could be relocated at the cryo-EM. The technique allowed good visualization of membranes and a prominent actin cytoskeleton (Fig 8 B). For imaging RCs, the first step was to choose a time point during infection when spherules would be more abundant at the PM, or alternatively, when small CPVs could be isolated from cell lysates. This prompted us to start a systematic analysis of the ‘intracellular-dynamics of spherules’. Recently obtained results, which I present below, have allowed us to establish a protocol for purification of CPVs of limited size, or relocation of spherules in infected cells with an accuracy of approximately 0.5-1 micrometres. Preliminary encouraging results have been obtained and the structural work is ongoing.

4.2 Dynamics of SFV replication complexes (I, II)

4.2.1 Spherules arise at the plasma membrane (I)

At late stages of infection the replication complexes of alphaviruses are localized on the limiting membranes of modified endosomes, the CPVs. However, spherules can also be detected at the PM. To obtain an overview of the distribution of RCs during infection, BHK cell monolayers were infected with SFV at a multiplicity of infection (moi) of 500 plaque forming units (PFU) per cell. Cells were fixed at different time points starting one hour after infection, and processed for immuno-cytochemistry to stain for replicase components (using antibodies raised against nsP1-4) and virus specific RNA-replication intermediates in the form of double stranded (ds) RNA (using a monoclonal anti-dsRNA antibody). The result of this experiment demonstrated that the localization of RCs was time dependent: they first accumulated at the PM, followed by appearance of dsRNA signal on punctate structures scattered in the cytoplasm, at the periphery of the cell. Only at later time points did the viral nsPs and dsRNA colocalize on larger perinuclear vacuoles, the CPVs. This coincided with the disappearance of both the small scattered puncta in the cytoplasm and the signal at the PM, suggesting that the CPVs are formed through a maturation process that involves intracellular transport of RCs (Fig 9) (I, Fig 1). Quantitative image analysis confirmed these results, using different moi, and different cells lines (BHK, HeLa, MEF). Using recombinant SFV in which the gene of the replicase component nsP3 was fused with the coding sequence of a fluorescent protein (the coral reef fluorescent protein ZsGreen) the intracellular transport of RCs could be followed by live cell imaging. Importantly, since nsP3 does not have intrinsic membrane binding properties, its association with cellular membranes (plasma membrane or endosomes) indicated its participation in the RCs. Colocalization with other replicase components and dsRNA after cell fixation confirmed this assumption. Moreover, correlative microscopy (CLEM) (147) allowed visualizing nsP3_ZsGreen labelled RCs at the plasma membrane of live cells, using a combination of confocal microscopy and subsequent imaging of the same cell by EM. This experiment confirmed the presence of thousands of spherules in regions of the PM that were previously located by fluorescence imaging (I, Fig 2). Thus, the plasma membrane served as a platform for spherule formation, and nsP3 was a good marker to locate membrane-bound RCs in different cellular compartments.

Fig 6. Morphological comparison of SFV, HIV-1 and spherules after conventional EM. A) SFV budding from BHK cells; nucleocapsids are progressively enveloped by the PM during the budding process. B) SIV-1 at different stages of budding; the association of gag proteins, viral RNA and membrane of the forming virions results is strong osmium staining (image kindly provided by Mark Marsh and Annegret Pelchen-Matthews, Cell Biology Unit, MRC Laboratory of Molecular Cell Biology, University College London, U.K.). C) Spherules on the surface of a CPV have similar diameters as SFV particles but appear very different; the membrane of the spherules has the same intensity and thickness as the other cellular membranes (red arrowheads). A denser core is often detected in the interior of the spherules (otherwise translucent), sometimes projecting to the cytosol trough an open pore-like structure (white arrowheads).

Fig 7. SFV infected cells and purified CPVs imaged by cryo or conventional transmission electron microscopy. A) The cytoplasm at the periphery of cryo-fixed BHK cells is imaged using a cryo- electron microscope operating at 200 kV. Four representative pictures are presented. Due to sample thickness and insufficient power of the microscope, not many features can be resolved. B- C) CPVs purified from infected cells, after cell fractionation and differential centrifugation, imaged by cryo electron microscopy after fast freezing (B), or conventional EM after chemical fixation and osmium staining and sectioning (C). In C, the CPVs (arrowhead) can be readily identified, due to the characteristic appearance of spherules after sectioning. In B, the low contrast of images after cryo-imaging does not allow straightforward identification of the same structures.

Fig 8. Correlative fluorescence cryo-electron tomography. A) Schematic illustration of the experimental plan: recombinant viruses are generated where the gene of the fluorescent protein ZsGreen is fused with the sequence of replicase component nsP3. After infection of cells cultured on special EM grids, fast freezing in liquid ethane ensures morphological preservation of intracellular structures. The localization of RCs is determined by fluorescence imaging, using a fluorescent microscope modified to operate at liquid nitrogen temperatures. Frozen grids are then transferred to the cryo-electron microscope. Using the reference letters and numbers on the grid (B) and the fluorescent map previously obtained, the area of interest is relocated (‘correlated’) in the same cell, and imaged by tomography. Digital image processing is used to obtain 3-D models and, when possible, enhance resolution. B) A pilot experiment was made with cells expressing fluorescent versions of viral replicase component nsP1 (nsP1-GFP). NsP1-induced filopodia-like protrusions were first imaged by fluorescence microscopy and, after transfer to cryo-EM, successfully relocated (arrow). As visualized at high magnifications, the protrusions are supported by thick bundles of actin fibres (tomography was not applied).

4.2.2 Internalization of newly formed spherules from the PM (I, II)

When using a high multiplicity of infection, thousands of spherules accumulated at the PM during the first two hours of infection. If the accumulation of spherules on the membrane of CPVs was due to their removal from the PM, due to an endocytic-like event, the possibility to inhibit this process was considered as a means to ‘freeze’ the spherules at the PM and prevent the formation of the CPVs. Among various drug treatments, the use of different inhibitors of class I PI3K, an enzyme known to regulate specific endocytic events, resulted in a virtually complete inhibition of RC-spherule internalization and, concomitantly, prevented the formation of CPVs. This effect was reversible, since several hours after washing out the drug, the signal of dsRNA that had accumulated at the plasma membrane was replaced by the appearance of large perinuclear vacuoles, similar to those found in non-treated infected cells at late stages of infection (I, Fig 3). Spherules remain connected to the cytoplasm, but during their transport from the PM to the surface of the CPV, they could be studied as an atypical form of endocytic cargo. To characterize this unconventional endocytic event, different approaches were taken: i) spherules at the stage of internalization were imaged using CLEM and tomography; ii) the contribution of cellular proteins involved in endocytosis was assessed by immunocytochemistry and iii) siRNA studies; iv) colocalization with different endocytic markers was used to follow the internalized spherules in the cytoplasm of fixed or live cells; v) the role of cholesterol and different cellular proteins in RC internalization was analysed using small molecule inhibitors. i) Spherules at the stage of internalization imaged using CLEM and tomography (II) BHK cells were infected with SFV_nsP3-ZxG (moi 500) for 2 h 30 min before chemical fixation. Using confocal fluorescence microscopy, an area of the PM was chosen based on the nsP3_ZsGreen signal. The presence of fluorescence in several confocal-optical planes indicated that a large area of the PM contained viral components. After relocalizing the same cell with the aid of dedicated imaging dishes (see Methods in article II), the same area of the PM was imaged and processed for tomography (II, Fig 1 A). Four 250 nm thick serial sections, covering an area of the PM of approximately 12.5 mm2, were aligned and analyzed. A portion of one tomogram (2.25 mm2) was selected for 3D modelling. Spherules at the PM reached local densities of more than one hundred per mm2 (78 in the model) (II, Fig 1 B). Analysis of the four obtained tomograms showed fifteen spherules in PM processes resembling endocytic vesicles at the stage of internalization. The morphology of these carrier vesicles was similar to that of caveolae, multi-caveolar (‘grape-like’) and caveosome structures (II, Fig 1 B and Fig 2). Differently from coated vesicles (e.g. clathrin coated pits indicated in Fig 1B and Fig 2B; II), no extra densities were seen close to their membranes to support the presence of a coat-like lattice. Each of these endocytic vesicles had a diameter of approximately 80-90 nm, and contained one spherule. The membrane of the spherule continued with the membrane of the endocytic

Fig 9. Localization of SFV RC at different times of infection. BHK cells are infected with SFV (moi 500) and fixed at indicated times. RCs are labelled by indirect immuno-fluorescence using dsRNA antibodies. At 1h30min pi, the RC localize at the PM; at 2.5 hpi the dsRNA signal is distributed in small and scattered cytoplasmic dots and at the PM; at 4hpi the dsRNA signal localizes in large dots in the perinuclear area. Higher magnifications (Zoom) of selected areas (white boxes) are shown for each time point

vesicle (II, Fig 2B). Each of the internalizing spherules had a diameter of approximately 50-70 nm, and a pore connecting the interior to the cytoplasmic space. In many instances, electron dense material was detected inside and coming out of the spherules. Similar endocytic structures without spherules were found throughout the analyzed PM area, most of them being flask-shaped, some more tubular (Fig 10 A). Interestingly, spherules were found close to but never inside clathrin coated vesicles (II, Fig 1B and Fig 2B). PM protrusions resembling macropinocytic processes were not evident, although in the imaged area the overall curvature distribution of the PM was quite irregular. Spherules at the PM had similar diameters as those found in CPVs, but a sizable fraction was significantly smaller, with diameters ranging from 40 to 20 nm; is some cases tiny spherules were observed (diameter 12-15 nm), suggesting that these could represent spherules in the process of formation. Despite the different size, all the spherules had pore structures with similar diameters (8-10 nm). In figure 10 B, spherules with different sizes are shown. Their pore-like structures are indicated by arrowheads.

ii) Colocalization of dsRNA with clathrin and caveolin-1 early in infection (II) The above results indicated that spherules could be internalized by caveolae rather than clathrin coated pits. To better characterize this endocytic event, 2 h 30 min after infection RCs were colocalized with different endocytic markers. Polyclonal antibodies, raised against the heavy chain of clathrin or cav1, were used in immunocytochemistry to distinguish vesicles internalized by clathrin- or caveolae-mediated endocytosis, respectively. Confocal imaging showed that the fluorescent spots of dsRNA-labelled RCs at the PM were segregated from the fluorescent signal of clathrin. No colocalization was detected also in the cytoplasm at the periphery of the cell. In control experiments, clathrin colocalized with the endocytic marker transferrin (II, Fig 2 A and B). When the dsRNA signal was compared with the distribution of cav1 or cholera toxin B subunit (CTxB), it appeared that the markers colocalized in large areas of the PM (II, Fig 2 C D). However, the localization of cav1 in our immunocytochemistry experiments was less reproducible than that of clathrin, and changed dramatically depending on the fixation and permeabilization protocol used. Thus, to confirm the results, we repeated the colocalization in HeLa cells stably expressing caveolin-1-EGFP (cav1-GFP) chimera. The distribution of dsRNA and cav1-GFP was very similar at early (II, Fig 4, 3h) but also at late times of infection (II, Fig 4, 6h), when large cytoplasmic vacuoles contained both dsRNA antibody and cav1-EGFP.

iii) RC internalization after siRNA mediated knock down of cav1 and in Huh7 (cav1 - /-) cells (II) A pool of validated siRNA against cav1 mRNA was used to test the effect of cav1 knock down on spherule transport from the PM. A non-targeting control siRNA was also included. HeLa cells were incubated with siRNAs (20 nM) for 48 or 72 hours prior to infection with SFV (moi 500). Immuno-cytochemistry and western blotting were used to validate the effect of each siRNA (II, Fig 5 A). Similarly to non treated infected cells, in the presence of the non-targeting siRNA, replication complexes stained with dsRNA- antibody localized in large punctate structures around the perinuclear area, the CPVs (II,

Fig 10. Morphological characterization of spherule internalization from PM using EM tomography. 2h 30 min after SFV infection, BHK cells were fixed and processed for transmission electron microscopy. A) Three dimensional modelling of a portion of the PM (light blue) with spherules (red). A digital section form the original tomogram is shown on the right panel. Spherules at the PM and endocytic processes resembling caveolae and multi-caveolar structures are indicated by red and black arrowheads, respectively. One spherule is found inside a multi-lobed structure in the cytoplasm. Front and back view of this structure are shown in the model. B) The same structure is shown in the four digital sections from the original tomogram; the red arrowhead indicates the position of the spherule. C) Top view of the PM shows spherules of different size grouped in clusters. The three dimensional modelling of the imaged area highlights the size difference. D) Digital sectioning of the same area shows that despite the different size, spherules are connected to PM through pores with similar diameters (arrowheads). 43

Fig 5 B, a and b). Instead, downregulation of cav1 dramatically reduced the formation of CPVs, and resulted in accumulation of dsRNA in small punctate structures at the PM, or in the proximity of the PM (II, Fig 5 B, c). The role of cav1 was confirmed in Huh7 cells, which do not express caveolin-1 and caveolin-2 (II, supplementary figure 1, A). In these cells, the distribution of dsRNA at the PM was different that that typically seen in BHK or HeLa cells. Confocal imaging showed that most RCs remained clustered at the PM at the bottom of the cell, even at 8 h.p.i. (II, supplementary figure 1, A and B). Only at very late time points (24 h), could a few dsRNA positive vacuoles be found in the perinuclear area (not shown).

iv) Colocalization of dsRNA with markers of recycling, early and late endosomes (II) We quantified the degree of colocalization between dsRNA- (or nsP3-) labelled RCs and endocytic markers, e.g. transferrin (recycling endosomes), early endosomal antigen 1 (EEA-1) and lysobisphosphatidic acid (LBPA, late endosomes), by immunocytochemistry and confocal imaging of BHK cells infected for 2 h 30 min. The EEA-1 and short incubation (4 min) with transferrin (Tf) were used to visualize early endosomes. Longer pulses (11 min) of Tf allowed identification of recycling endosomes. The late endosomal compartment was localized using antibodies against lysobisphosphatidic acid (LBPA) in fixed cells (or by lysotracker, a pH-sensitive marker of acidic organelles, in live cells). The results indicated that the cytoplasmic vesicles carrying internalized spherules did not colocalize with conventional endocytic markers at this early time of infection (II, Fig 6 A), while in control experiments, the EEA-1 colocalized with endocytosed Tf (4 min uptake) (II, Fig 6 B). At later time points, the RCs labelled with anti-nsP3 antibodies colocalized with the late endosomal marker LBPA in large vacuoles. EM analysis of parallel samples indicated that the vacuoles corresponded to CPV (II, Fig 6 C).

v) Role of cholesterol and effect of small molecule inhibitors on spherule internalization (II) Endocytosis of caveolae is strictly dependent on membrane cholesterol. We used methyl- β-cyclodextrin (MβCD) to test whether cholesterol depletion could inhibit spherule internalization. Since high concentrations of MbCD are known to inhibit also clathrin mediated endocytosis, we used amounts that allowed at least 60% levels of Tf internalization compared to untreated non-infected cells (II, supplementary figure 2, A, B and C). The level of MbCD induced cholesterol depletion was monitored by filipin staining (II, supplementary figure 2, A). The integrity of the actin cytoskeleton (labelled with fluorescently conjugated phalloidin) was also monitored (II, supplementary figure 2, A). Confluent infected cells were incubated with the drug (2 mM in serum-free medium) starting at 20 min or 1 h 30 min after infection. At 4 h post infection cells were fixed and the localization of dsRNA in treated and non treated infected cells was monitored by immunocytochemistry. Strikingly, in cells deprived of cholesterol, RCs remained at the plasma membrane even at late stages of infection, in contrast to the accumulation in the perinuclear compartment observed in non-treated cells (II, Fig 7).

EM analysis indicated that spherules were still formed (not shown). Thus, cholesterol was apparently important for spherule endocytosis. On the other hand, cholesterol depletion did not prevent spherule formation. Other drugs were tested for their ability to inhibit RC endocytosis (II, Fig 8). For all the compounds, the unspecific effect of each treatment on the actin cytoskeleton was monitored by phalloidin staining in control non-infected cells (II, Fig 8 B and C). Notably, inhibiting the function of two Rho GTPases, i.e. by using Rac1 and ROCK inhibitors (ROCK being an effector of RhoA), did not result in accumulation of RCs at the PM (II, Fig 8 A and C); nor did inhibition of Pak1 and Pak2 proteins (Rac1 effectors), suggesting that macropinocytosis is not the main endocytic pathway for spherule internalization (II, Fig 8 A and C). Interestingly, inhibition of ROCK enhanced CPV formation, as the dsRNA signal accumulated in large vacuoles in the perinuclear area, similar to those typically detected at very late stages of infection (e.g. 8 hpi) (II, Fig 8 A). Pharmacological inhibition of Dynamin-II by means of Dynasore (50 µM) did have a strong effect on RC transport from the PM. However, it should be noted that although the effect of Dynasore was consistent with the involvement of Dynamin-II (dynII), its use also affected the actin cytoskeleton in control experiments (II, Fig 8 A, B and C). Thus, care should be taken in interpreting the results when this compound is used, and further experiments carried out to assess the contribution of actin cytoskeleton in endocytosis. Indeed, a role for actin in the early steps of spherule internalization was confirmed indirectly by using blebbistatin, a compound that, by inhibiting the catalytic activity of non-muscle myosin-II, induces a strong and virtually instantaneous collapse of the actin network. When infected cells were treated with blebbistatin from 1.5 to 4 h p.i., the formation of CPVs was inhibited and dsRNA accumulated at or in the proximity of the PM in small punctate structures (I, Fig 5 B-D).

In summary, the data so far obtained strongly argue that the RC spherules of SFV form mainly at the PM, where they accumulate during the first two-three hours of infection. Morphological analysis, siRNA and pharmacological inhibition studies indicated that internalization of spherules from the plasma membrane occurred via PI3K-, cholesterol-, cav1- and apparently dynII-dependent endocytosis. A functional actin-myosin network seemed to be required to assist this ‘unconventional’ endocytic-like event.

4.2.3. The signal for spherule internalization resides in the replicase component nsP3 (II)

Caveolae endocytosis is highly regulated by signalling events. If the internalization of spherules was trigged by similar signals, the question was raised on which part of this peculiar structure could interact with the cellular endocytic machinery. Our group had previously shown that the membrane localization of the replicase polyprotein-intermediates was influenced by the presence of nsP3. Specifically, while expression of nsP1 or uncleavable nsP1-nsP2 chimeras resulted in a very specific PM

localization, addition of nsP3 retargeted a fraction of the recombinant polyprotein to the cytoplasmic side of endosomes. Thus, nsP3 could represent a signal for spherule internalization. Of particular interest was a heavily phosphorylated sequence of 50 amino acids in the C-terminal half of the protein, the function of which has remained obscure. Deletion of this region results in viable viruses (SFV-Δ50), thus allowing us to test its effect on spherule trafficking. To our surprise, the RC spherules of SFV-Δ50 remained at the PM through the infection cycle (II, Fig 9). Consistently, no CPVs were formed in the cytoplasm as judged by distribution of dsRNA in immuno-cytochemistry experiments and by EM analysis of parallel samples (not shown). This unexpected result was confirmed in different cell lines (BHK 21, HeLa, MEFs) (not shown).

4.2.4 Intracellular transport of SFV spherules: biogenesis of CPV-I (I)

Following their internalization from the PM (2.5 to 4.5 h.p.i.), RCs labelled by dsRNA antibody or nsP3-ZsGreen in live cells (I, Fig 3) were found as spots scattered throughout the cytoplasm (I, Fig 1 and 2). Using fluorescence live cell imaging in the presence of lysotracker, to visualize acidic vacuoles, it was possible to demonstrate that these early carrier vesicles were not acidic. Single particle tracking revealed a multidirectional movement confined to the cell periphery, in some occasions very close to the PM (I, Fig 4 A-D). EM analysis of thin sections obtained from parallel experiments confirmed the presence of numerous vacuoles randomly distributed in the cytoplasm, and with a diameter ranging from 150 to 400 nm (not shown). In each section, a few spherules could be detected on the limiting membrane of these organelles (I, Fig 7). A dramatic change in dynamics was observed when some of the RC-carrying vesicles docked to a larger acidic vacuole (I, Fig 4 E). In this case a long range directional movement displaced the hybrid vesicle from the cell periphery towards the perinuclear area. Quantitative image analysis after particle tracking showed that this represented the net movement for all the acidic vesicles containing nsP3 (I, Fig 4, and our unpublished data). Once ‘at destination’, the vesicles became more static and underwent slow fusion events resulting in larger vacuoles (Fig 4). Repetition of this process led to the formation of the previously characterized CPVs: large acidic organelles, typically detected from four hours post infection in the perinuclear area, and characterized by hundreds of spherules lining their limiting membrane (I, Fig 3D, 4E and 7). Disrupting the microtubule network with nocodazole abolished the long range movements and prevented RCs from reaching the perinuclear compartment (I, Fig 6 A-D). Docking events were not prevented, such that at five hours post infection, instead of being uniformly distributed on the limiting membrane of the CPVs, RCs appeared as patches on the surface of large acidic organelles distributed in different areas of the cytoplasm (I, Fig 6 B).

4.3 SFV infection induces down-regulation of transferrin receptor and redistribution of EEA-1 (unpublished)

The results with lysotracker showed that within four and six hours after infection, all the acidic vacuoles appeared ‘trapped’ in the pericentriolar area in the form of CPVs, which continued to undergo fusion events for many hours, resulting in very large static vacuoles (I). This prompted us to investigate whether other changes in the endocytic system occurred upon SFV infection. With the aid of immunocytochemistry and time course experiments, the distribution of the early endosomal marker EEA-1 and the transferrin receptor (TfR) was studied in BHK or HeLa cells. In control non-infected cells, EEA-1 antibodies stained numerous small vesicles at the periphery of the cell, and a perinuclear compartment strongly colocalizing with the recycling endosome marker transferrin (Fig 11 A, non-infected). Whereas a similar distribution was observed in infected cells at early (2.5 h.p.i.) stages of infection, already at 4 h p.i. the peripheral EEA-1 staining was lost, and replaced by a more perinuclear signal in large, tubular structures partially overlapping with dsRNA-staining (Fig 11 A, SFV). Similar changes were seen for the TfR, which during infection progressively disappeared from the cell periphery (Fig 11 B). Quantitative image analysis indicated a 60% decrease of the total amount of receptor per cell, and a concomitant similar decrease in Tf uptake (Fig 11 B and C).

Thus, along with spherule trafficking and CPV formation, the endocytic compartment and the PM undergo other important, large scale modifications.

Fig 11. SFV infection causes redistribution of early and recycling endosome markers. A) The distribution of the early endosomal marker EAA-1 is monitored by indirect immno-fluorescence in infected and non infected BHK cells. In control non infected cells the marker is distributed in small spots distributed throughout the cytoplasm and colocalizes with endosomes labelled by internalized transferrin. In infected cells, EAA-1 as similar distributions as in non infected cells, and does not colocalize with dsRNA. At 4 hpi, the small marker is redistributed mainly in the perinuclear area, where it colocalizes with dsRNA in large structures. Higher magnifications of selected areas are shown for each panel. B) The distribution of TfR and amounts of internalized Tf are analyzed in infected and non infected HeLa cells. In non infected cells, the receptor is distributed at the PM and in spots scattered throughout the cytoplasm, and colocalizes with internalized Tf. In SFV infected cells, at 6 hpi the TfR is mainly localized in the perinuclear area, and cells do not internalize Tf efficiently. C) Quantitation of the results presented in B; in infected cells, the total amount of TfR is 40% of that counted in non infected cells. This corresponds to a similar reduction in Tf uptake. 48

4.4 Addressing the function of nsP2 with ts-mutants (III)

The multifunctional protein nsP2 is required for RNA synthesis and virus viability. Its role as a protease in polyprotein processing has been firmly established for different alphaviruses, but for the other demonstrated enzymatic activities, i.e. NTPase, RNA- triphosphatase and helicase, the precise biological role during infection remains to be elucidated. A series of SFV temperature-sensitive (ts-) mutants was previously isolated based on their loss of viral RNA synthesis at the restrictive temperature (39 °C) (47;148). Many of these ts-mutants had a point mutation in nsP2 (149). For several mutants, temperature ‘shift-up’ experiments mainly showed that if infected cells were incubated at the permissive temperature (28 °C) for a period of 4-6 hours prior to shift to restrictive temperature, 42S RNA synthesis still continued but 26S sub-genomic RNA production was shut down. For one of these nsP2 mutants, this phenotype was reversible, as shifting cultures back to 28 °C restored 26S RNA synthesis. Thus, a specific role in the regulation of sub-genomic promoter was proposed for nsP2 (76;84). Based on sequence and biochemical information, nsP2 can be divided into two major domains. The N-terminal half contains SF1 family helicase motifs, while the c-terminal part includes the protease domain. Ts-mutations are distributed in both regions (III, Fig 1 A and B) We started by first asking: i) whether mutations in one domain could affect the activity of the other part of the protein; ii) given the suggested role during RNA capping and the helicase activity, we also asked whether different active sites were used to ‘fuel’ the two reactions, namely RNA-triphosphatase and NTPase, respectively; iii) the reversibility of 26S RNA shut-down phenotype was assessed in vivo for viruses with mutations in either of the two nsP2 domains. The properties of wt and ts-mutant variants of nsP2 were analysed in vitro by comparing the enzymatic activities of purified recombinant proteins at different temperatures. Histidine tagged versions of wt and mutant variants were obtained after over-expression in E. coli and purification by Nickel-affinity chromatography. (III, Fig 1C and D). The assays for three enzymatic activities, protease, NTPase and RNA triphosphatase were optimized in vitro at different temperatures using wt protein.

Protease. As the substrate for the protease assay, purified thioredoxin-conjugated peptides were produced containing the cleavage site sequence between nsP3 and nsP4. The release of the larger cleavage product (L) was measured after SDS-PAGE and analysis of Coomassie blue stained bands by densitometry, in time course experiments (III, Fig 2 B). The progress of the reaction was temperature dependent. Wt nsP2 had little activity at 39 °C, which was the restrictive temperature for the replication of the ts-mutant in infected cells. Thus, 24 and 35 °C were chosen as permissive and restrictive temperatures, respectively, for further comparisons with the mutant proteins (III, Fig 2 C). In this case, the temperature difference was 11 degrees, the same as that used in the in vivo studies but the range shifted by 4 degrees.

All of the protease-domain mutants (ts 4, 6, 11) had a ts-phenotype in vitro; one of the three helicase-domain mutants (ts-13a) also showed a strong effect at 35 °C. Analysis of the kinetic curves indicated that after 5 minute incubation at 35 °C the mutant enzymes became unstable. The rate of cleavage dropped dramatically such that enzyme and released substrate were no longer related by first order kinetics (III, Fig 3, B). This was confirmed when the enzymes were incubated at the restrictive temperature prior to the addition of substrate; in turn, the substrate was not sensitive to the same treatment. The effect of ts-mutations on the proteolytic processing of the three cleavage sites (nsP1-nsP2, nsP2-nsP3, nsP3-nsp4) were also studied in infected cell cultures. Cells were infected with wt or mutant viruses and incubated at 28 °C to allow protein and RNA synthesis. At 5h.p.i. cultures were shifted up to 39 °C or maintained at 28 °C as a control. After 30 minute temperature equilibration, [35S]-methionine was added to label proteins, followed by long (2 hours) or short (20 minutes) chases. Released nsPs and their polyprotein precursors were immunoprecipitated from denatured cell lysates, and analysed by SDS-PAGE followed by exposure to phosphoimaging plates. To resolve nsP1 and nsP3 bands, lysates were divided into two aliquots and nsPs immunoprecipitated using combinations of nsP1+nsP2 and nsP3+nsP4 antisera. Quantitative immunoprecipitation was obtained after optimization, allowing the estimation of molar ratios of the four nsPs and their precursors (III, Fig 7). In these experiments, accumulation of P1234 and P34 precursors pointed to a 3/4 cleavage site defect. As expected from the in vitro results using the analogous cleavage site substrate, the three protease mutants and the helicase mutant ts13a had a strong defect in the processing of this site. Ts9 and ts1 cleaved the 3/4 site like the wt protein. Although this results suggested that the helicase mutants had no protease defects, analysis of the other polyprotein precursors, P123, P12 and P23, demonstrated that the cleavages of sites 1/2 and 2/3 were severely and specifically impaired in SFV ts9 infected cells. Indeed, in vitro assays where a 2/3 cleavage site model substrate was compared with the ¾ site confirmed this result. The mutant ts13 had defects in cleaving both the 3/4 and 2/3 sites whereas ts1 had no defects at all. Thus, point mutations in the helicase domain could affect the protease activity of nsP2 in a mutant specific manner.

NTPase and RTPase. The release of phosphate was measured in both NTPase and RTPase reactions. Wt and mutant proteins were incubated with either [γ-32P]GTP or 5’[γ- 32P]GTP-labelled RNA (64 nucleotides long), respectively. In both experiments, the results were very similar. None of the protease mutants had a ts-defect whereas two out of the three helicase mutants, ts13a and 9, were inactive at the restrictive temperature. Interestingly, whereas ts13a had some activity at 24 °C, ts9 was inactive even at this permissive temperature (III, Fig 4 and Fig 5). This unexpected result was confirmed when the soluble fraction of nsP2 was obtained by immuno precipitation from cells infected with the respective mutant virus (SFV ts9) at 28 °C (III, Fig 6).

NTPase and RTPase share the same active site. Each of the helicase mutants, ts1, ts13a and ts9, behaved very similarly with respect to GTPase and RTPase reaction. Also sequence analysis of nsP2 indicates the presence of only one ‘Walker A’ motif responsible for GTP hydrolysis. Thus, similarly to the replicase proteins of other viruses (150-152), the same active site could be responsible for the two reactions. This hypothesis was confirmed as the RTPase reaction was competed by increasing amounts of unlabeled GTP, and completely inhibited by the addition of γ-S-GTP, a non hydrolysable analogue of GTP. The competition vas not due to inhibition caused by release of product, as addition of phosphate had no inhibitory effect (III, Fig 5 D)

Subgenomic RNA synthesis. The above data suggested that the two main domains of nsP2, the N-terminal helicase and C-terminal protease, could influence each other in a temperature sensitive manner. Consistent with this, it was previously shown that point mutations in both regions of the protein resulted in viruses with a similar phenotype: impaired sub-genomic RNA synthesis at the restrictive temperature. We wanted to test whether this phenomenon could be induced and then reversed, for both classes of mutants, by shifting infected cultured cells between restrictive and permissive temperatures. Replication complexes were allowed to assemble at 28 °C for 6 hours. The shutdown of 26S RNA synthesis was induced by shifting cultures to 39 °C, in the presence of cycloheximide and actinomycin-D, to stop protein and DNA-dependent RNA synthesis, respectively. A pulse of 3H-uridine was given from 7 to 8 hpi, to label newly synthesized viral RNAs. In control experiments, cultures were maintained at 28 °C. After extraction and separation by gel electrophoresis, the extent of synthesized 42S and 26S RNAs was quantified by autoradiography and densitometry. The results indicated that at the restrictive temperature the synthesis of genome size 42S RNA continued at the expense of subgenomic 26S RNA (Fig 12) (III, table 2). The reversibility of this ts-phenotype was tested in similar experiments, with the difference that after the 2 hour incubation at 39 °C (6 to 8 hpi), infected cultures were shifted back to 28 °C and RNA labelled from 9 to 10 hpi. The synthesis of 26S RNA was restored in the case of both protease (ts 4) and helicase mutants (ts 9 and ts 13a) (Fig 11) (III, table 2).

Thus nsP2 appeared to act as a molecular switch in the activation of genomic versus subgenomic promoter.

Fig 12. Temperature dependence of genomic and subgenomic RNA synthesis. A) Scheme of ‘shift-up’ experiments. BHK cells were infected at 28 °C with wt or ts-mutant variants of SFV, in the presence of actinomycin-D. At 6 hpi, cultures were shifted to 39 °C (controls were maintained at 28 °C) and protein synthesis was shut off by addition of cycloheximide. Cells were labelled with [3H]uridine from 6 to 7 hpi or from 7 to 8 hours post infection. Labelled viral RNAs were extracted, separated by electrophoresis and visualized by autoradiography. To test reversibility, after incubation at 39 °C (from 6 to 8 hpi), infected cultures were returned to 28 °C, and RNA labelled from 9 to 10 hpi. B) The outcome of a typical experiment where the products of RNA synthesis are analysed for wt, ts-4 and ts-9 infected cells. The synthesis of 26S subgenomic RNA in wt infected cells is only slightly affected by temperature changes, while in the case of ts-4 and ts-9 the 42S synthesis progressively increases at the expenses of 26S RNA. The effect can be reverted by subsequent shift to 28 °C for ts4 and, to a lesser extent, ts9 . The top row indicates the time when cells were lysed and RNA extracted as explained in A.

Discussion

Structure of SFV spherules

The idea that +RNA virus replication occurs within defined membrane structures raises interesting questions, such as what is the origin of the membranes or their mechanism of formation and chemical composition. It also represents a challenge for virologists, who are desperately trying to fit micrometers of viral genomes and fifty years of molecular biology of virus replication into a sphere of 50 nm. As one of them, I have had my share of the struggle. In fact, the existence of viruses such as Reoviruses (dsRNA) proves that it is possible to transcribe double stranded RNA molecules even when packaged into small structures. In the case of Bluetongue virus (BTV) (genus Orbivirus) for instance, the 10 dsRNA segments totalling 19.2 kb are organized in a highly structured core particle comprised of two protein layers, and surrounded by an outer capsid shell, forming an icosahedral virion of 86 nm in diameter (the core being 70 nm) (153). Whereas the capsid shell detaches after cell entry, the core of BTV remains intact in the cytoplasm and is transcriptionally active. The concentration as well as degree of organization of dsRNA and core proteins within the particle is so high that the structure of the BTV core could be determined by x-ray crystallography (154-157). In their study, Gouet and coworkers showed that the genomic dsRNA formed “ordered layers surrounding putative transcription complexex (TC) at the apices of the particle”. The authors describe the BTV core particle as a fascinating transcriptional machine. In figure 13 A, I provide a schematic model adapted from the original publication (156). In their model, which is also supported by various other studies that appeared shortly after (157), each of the 10 dsRNA segments inside the core is coiled around a TC positioned in front of one of the twelve apices at the 5-fold symmetry axis (two apices will not contain TC). Newly synthesized mRNAs would be extruded from the particle through a pore at or close to the 5-fold axes. Each dsRNA segment will undergo many rounds of transcription. Each TC harbours polymerase, helicase and methyltransferase activities. Molecules (e.g. nucleotides and divalent cations) required for RNA synthesis can access the core interior. Notably, the authors speculate that the dsRNA spirals inside the core are prevented from expanding only by steric interaction with neighbouring segments, thus they excluded strong chemical interactions with the components of the TC or core-shell proteins. It is suggested that the lack of strong protein-RNA interaction would facilitate the transcription reaction. If these assumptions are correct, one could infer that in the absence of lateral constrains imposed by other segments (i.e. if only one dsRNA segment would be present in the core) and replacing the outer protein shell with a membrane, the dsRNA would expand and become unfolded. In this case the transcriptional machinery would have a rather familiar structure: a membrane surrounding transcription complex and (ds)RNA, and exchanging newly synthesized mRNA and other macromolecules through a pore placed in front of or surrounding the RC. A schematic model in figure 13 A illustrates the functional comparison between the reovirus core particle and the spherules. Similar considerations could be made for 53

retroviruses such as HIV-1, which contain single-stranded RNA genomes and active polymerases inside the particle, and are able to initiate (reverse) transcription in the cytoplasm before virion disassembly. Thus, while engaged in RNA synthesis, spherules could be functionally and structurally analogous to dsRNA viruses or retroviruses (182). It is noteworthy that when imaged by conventional EM, both reoviruses and retroviruses are very different from spherules. The virion of reoviruses, similarly to that of SFV, appears as a single dark spot that fills all the space inside the envelope membrane. On the other hand, immature HIV-1 virions are readily identified at the PM by their electron densities underlying the membrane at the site of budding. Upon maturation, the particle undergoes dramatic conformational changes such that the electron densities are replaced by a cone-like structure in the centre of the virion (158). Spherules, in contrast, do not contain dark layers on either side of their membranes. An electron dense amorphous material is often visualized inside the lumen of the vesicle in chemically fixed samples. After high pressure freezing, acetone substitution and osmium staining, the interior of the spherule includes densities with a clear filamentous appearance. Similar features are found throughout the cytoplasm, suggesting that this could represent RNA. The pore structure that connects the interior of the spherule to the cytoplasm is better visualized after aldehyde fixation. Interestingly, under these conditions the pores are relatively large (8-15 nm), with fluctuations in diameter of up to 30%. Spherules with very wide pores are often translucent inside, indicating that the structure could be affected by the harsh procedure of conventional EM, which includes treatments with aldehyde, highly reactive salts (e.g. osmium tetroxide), dehydration and embedding in synthetic resins prior to sectioning and imaging. This suggests that the spherules are more labile structures than virions, and probably reflects less compact protein organization. On the other hand, high pressure freezing resulted in better morphological preservation of cellular compartments and spherule morphology. However, the chemical properties of osmium salts under the highly hydrophobic conditions imposed by acetone made the staining of membranes less reproducible and most likely favoured staining of proteins. This could explain why the pores were not discernible and instead, thin neck-like structures connected the interior of the spherule to the cytoplasm. It is noteworthy that most of the EM work on virus induced membrane alterations has been done under chemical fixation conditions. In these studies, similar pore-like structures could be identified after tomography (17; 19). For the structures induced by SARS coronavirus, Knoop and colleagues used cryofixation and substitution. Despite the high quality of their tomograms, pore structures could not be identified, leading to the dilemma of how RCs could function if they cannot access the cytoplasm (18). In my opinion both techniques in conjunction with tomography are very useful in giving an overview of the 3D-organization of large and complex structures. However, the intrinsic properties of the reagents used and the fact that images represent the distribution of heavy metals (e.g. osmium, uranium), which is dependent on the chemical environment used in each procedure, make their use unsuitable for high resolution studies (159). To my knowledge the only technique that could provide an accurate representation of RC organization is cryo EM. This technique has been extensively used to obtain high resolution structures of viruses with defined symmetry (160;161). The structure of SFV

was obtained by cryo EM and single particle averaging (162). Larger structures, which do not posses symmetries, are not suitable for particle averaging but a 3D reconstruction can be obtained by cryo EM and tomography (163;164). Although the resolution of this technique is by far lower than what can be achieved by averaging methods, it represents an important starting point. Structures like membraneous pores would be identified without ambiguity and most importantly, the distribution of intensities in each micrograph will represent biological materials since no chemicals are included (165-167). If under these conditions spherules would retain a regular structure and their labile molecular assemblies were preserved, they could be imaged by tomography and if sufficient numbers of structures were analysed, tomograms could be averaged leading to increased resolution. This method has been applied to nucleopores and smaller structures like the spike glycoproteins of irregular enveloped viruses (165;168-172). The major limitation is that the thickness of the sample has to be small (<300 nm). Among the RCs of other RNA viruses, SFV spherules belong to the most regular and smallest structures. However, although very abundant in the mature CPVs, their perinuclear localization makes direct imaging by cryo EM impossible. Nevertheless, by a systematic study of the intracellular dynamics of RCs, I could find ways of ‘freezing’ the spherules at the plasma membrane. The use of correlative microscopy and the availability of relatively flat cell lines (e.g. MEF, Huh7) (173-175) have opened the possibility to image the replication complex of an animal virus in its native state in infected cells. Although technically demanding, this possibility is feasible and will be pursued.

Spherule formation

The generation of membrane curvature requires energy. The energy required to bend a bilayer can be estimated from the elastic model of lipid membranes (176;177). For instance ~70 kcal/mole are required to generate a tubule of 60 nm in diameter and 60 nm in length. The formation of a spherical vesicle with similar diameter would require ~ 300 kcal/mole (178), and an equivalent distribution of tension in all the directions. ATP hydrolysis (~10 kcal/mole) could provide the required energy, and in principle even the highly asymmetric distribution of lipids across a bilayer could generate membrane curvature (179). However, biological membranes do not appear to have such local distributions of lipids to support the spontaneous formation of vesicles with curvature radii of few tens of nanometres (180;181). Considering their size and morphology, the structure of spherules is energetically unfavourable. Differently from cellular carrier vesicles (that generally dissipate their high membrane curvature by quickly pinching off and fusing with another membrane), spherules remain attached for many hours to the rest of the organelle membrane through a neck structure. In analogy with budding viruses, spherule formation could be explained by the accumulation of membrane binding proteins (i.e. nsP1, in the case of SFV) forming a coat at the cytoplasmic side of specific membrane sites. There, viral RNAs and polymerase would be recruited and wrapped by membranes. This mechanism has been

proposed by Dr. Ahlquist. In his thorough studies, the spherules induced by +RNA viruses are proposed to parallel the virions of retroviruses and the core of dsRNA viruses (5;182). To study the formation and composition of spherules induced by the distantly related plant alphavirus-like Brome mosaic virus (BMV), Schwartz et al. used a yeast expression system that sustained expression of the two replicase components 1A (Helicase and methyl transferase) and 2A (RdRP), as well as RNA molecules harbouring cis-acting elements required for recruitment and replication. A combination of quantitative EM- immunolabelling and biochemical assays allowed them to estimate the amounts of different replicase components and viral RNAs in yeast membrane extracts (5). It was shown that i) ectopic expression of protein 1A is sufficient to induce spherule formation; ii) approximately 50-100 1A molecules are contained in one spherule, while 2A is 25 times less abundant; iii) each spherule contained one copy of minus-strand RNA. Both gag and protein 1A show high but not absolute selectivity for the recognition of their cognate RNA. As the authors clearly state in the discussion, it cannot be excluded that cellular RNAs played a structural role in spherule formation since “non-specific RNA interaction may drive protein 1A-induced spherule formation in the absence of BMV RNAs”. From the EM micrograph shown in their work, it is difficult to make direct comparisons between the induced spherules and the structure of budding retroviruses. Nevertheless, my impression is that the spherules induced by BMV (or protein 1A overexpression) are different from the spherules of SFV; it appears that the inner side of the spherule membrane is darker than the outside, suggesting the presence of proteins. In some images the whole lumen is actually filled with electron densities. Moreover, in all the pictures where it is shown, the neck structure of the spherule is very long (about 30 nm, half the size of the diameter of the spherule). Indeed, it is known that expression of some Gag mutants induces very similar structures at the PM (183). Thus it is possible that in the case of BMV, the main scaffold of the spherules consists of a layer of protein 1A. Similar coats were proposed for the spherules induced on the outer membrane of mitochondria in Drosophila cells infected with Flock house virus (FHV, Nodavirus) (17). FHV has a bipartite genome (RNA 1 and 2, see table 1), and induces spherules which are very similar to SFV spherules. EM and tomography of chemically fixed FHV infected cells were used to obtain a 3D model of the modified mitochondria. Quantitative analysis of the amounts of viral replicase protein A (the only component of FHV replicase) and synthesized minus-strand RNA were used to estimate the composition of FHV replication complex spherules. It was estimated that each cell contains approximately 20.000 spherules, each consisting of ~100 protein-A molecules, one copy of RNA 1 and two of RNA 2. Whether ectopic expression of protein-A alone results in spherule formation has not been yet demonstrated; in fact concomitant expression of protein A and cognate viral RNA was necessary for the formation of spherules in yeast (184), indicating that the RNA could play an important (perhaps structural) role in this process. Interestingly, in the case of BMV expression of protein 1A did result in correct ER targeting and formation of spherules indistinguishable from dose induced during virus infection. These findings underline that the mechanisms of spherule formation could be different depending on the virus. Indeed, expression of viral replicase proteins involved in membrane binding often

results in similar but not identical membrane rearrangements as those induced by virus infection. This has been shown for instance for members of the Picornaviridae, Flaviviridae and Coronaviridae (1;3;185;186). In this respect, it is important to understand the sequence of events that lead to vesicle formation and activation of RNA synthesis. The concept of a preformed protein coat implies that the concentration of membrane binding proteins should be high enough to induce high membrane curvature. Moreover, other RC components will have to be transported specifically to sites of membrane deformation. Although this could be achieved at late stages of infection, when protein synthesis is very efficient, at the onset of infection these requirements would be difficult to fulfil; it is conceivable that different viruses have evolved different strategies to induce membrane alterations and coordinate RNA synthesis. This would at least be consistent with the fact that some +RNA viruses (e.g. flaviviruses) have very slow growth kinetics, while others are very fast (e.g. picornaviruses and alphaviruses) (summarized in (3)). SFV is a very efficient virus, and probably very few (perhaps one) particle can establish infection. Given the large area of cellular membranes, it is in my opinion unlikely that one mRNA molecule would provide sufficient proteins to support spherule formation. For instance, transfecting cells with high amounts viral genomic RNAs harbouring mutations that inhibit replication does not result in detectable amounts of nsP1 after immunocytochemistry or western blotting (Pirjo Spuul, personal communication). Rather, I prefer a model in which a few nsPs would recruit the genomic template soon after translation or co-translationally, and be directed to the target membrane as a ribonucleoprotein. There, spherule formation could commence. This would also parallel the assembly of dsRNA virus particles, which starts from single-stranded RNAs bound to proteins (187). Supporting this model, expression in susceptible mammalian cells of the membrane anchor protein nsP1, nsP1+nsP2 or polyprotein precursor P123, does not result in spherule formation (8). We are currently investigating the mechanism of spherule formation. We have optimized a ‘reconstituted’ system by which different combinations of nsP polyproteins and RNA templates can be produced in mammalian cells in trans, independently of virus replication. Using CLEM, we can test the efficiency of spherule formation, relocating specific areas of the cell with an accuracy of 0.5-1 µm2. Our preliminary results suggest that the expression of different combinations of the nsP polyprotein precursors is not sufficient for spherule formation, supporting the idea that a ‘pre-RC’ assembled on viral RNA could precede spherule formation. Since the system allows introduction of site-specific mutations, it will be possible to test the contribution of the enzymatic activities of the nsPs and the role of RNA template molecules provided separately. Clearly, the involvement of cellular proteins should not be excluded. Recent evidence has shown that the process of cellular vesicle formation can be very different from the classical concepts of coat protein assembly and consequent membrane curvature (e.g. clathrin- or COP-coated vesicles). The biogenesis of multivesicular bodies (MVB) provides an illustrative and, in terms of spherule formation, pertinent example. Inspired by the electron micrographs of Dr. John Heuser (188), different research groups have conceived an elaborate model by which some of the components of the ESCRT-III complex would assemble in spirals on the cytoplasmic surface of endosome membranes.

Due to a poorly defined ‘purse-string’ mechanism, the spirals would constrict, using energy from ATPases, thereby pushing membranes towards the intralumenal space of the endosome (189;190). This model has now been challenged by the finding that similar vesicles can be induced in in vitro systems without the previously reported protein components of the ESCRT-III complex. Instead, some subunits of the ESCRT-I and -II were sufficient. The model has been considerably revised, and now it is proposed that long bridges could form from one side to the other of a forming spherule, which will progressively invaginate and, with the aid of ESCRT-III, eventually be released into the lumen of the MVB (191). Although the question of MVB formation is clearly far from being solved, siRNA studies have indicated that components of this large macromolecular assembly are required for the formation of tombusvirus spherules (22), which form on the membranes of peroxisomes, and are very similar in size and morphology to SFV spherules. Interestingly, the assembly and budding of HIV-1 virions also depend on functions of the ESCRT machinery (158;192). ESCRTs mainly target endosomes and PM, and their involvement in SFV spherule formation is also under investigation in our research group. Both of the models so far proposed for ESCRT functioning assume that membrane vesicles with similar size and topology to spherules are generated without the involvement of a classical protein coat inside the mature vesicle (189). Given the variety and originality of such models, I felt it legitimate to propose a speculative model for spherule formation. This is an alternative to the model based on a protein-coat lattice inside the spherule, proposed by Dr. Ahlquist, and it is inspired by the recent work on biogenesis of MVB. Among the numerous possibilities, one of my favourites is that of ‘inflatable’ spherules (Fig 13 B): after translation of incoming viral RNA, a ribonucleoprotein complex consisting of one +RNA genome, several copies of P123 and fewer nsP4 would assemble at the 3’-end of the RNA forming a ring-like structure, the ‘pre-RC’. This complex would bind to the membranes through nsP1 interactions. Using the force generated by the nsP2 helicase and/or RNA polymerase activity of nsP4, the RNA is pumped (‘inflated’) against the membranes generating membrane curvature. NsP1 is anchored to membranes via monotypic binding of the amphipathic peptide moiety and palmitoylation at the c-terminus; thus in this model the lipids are supposed to move laterally upon RNA translocation, while the nsP1-achored RC must remain attached to the membranes, thereby maintaining the pore of the spherule throughout the process (Fig 12 B). Similarly to the packaging motors of certain bacteriophages (e.g. phi6 or phi12, (193;194)), in this model the ‘packaging’ of RNA into the spherules is dependent on the presence and enzymatic activities of helicase and/or polymerase. Synthesis of RNA would occur inside the spherule and the newly synthesized molecules extruded through the pore. The size of the spherules depends on the size of the RNA. This completely speculative model, which I propose as a framework to better formulate questions (and challenge other models), is at least supported by evidence from the morphological study presented in this work. EM tomography of the PM at early stages of infection showed spherules of different sizes. Many were very small but still connected to the cytoplasm by narrow pores, with similar diameters as those of larger spherules (Fig 10 B). This is evidence against a mechanism like the one proposed for e.g. HIV-1 virion assembly (fig 6) (158), clathrin coated vesicle formation or MVB biogenesis (Kirchhausen 2000; Mürk et al., 2003)

(105;195), in which forming vesicles must have wider openings that those in a more advanced stage of budding. More experiments will hopefully reveal the mechanism of spherule formation. I am confident that the information obtained from ongoing structural studies and the newly optimized reconstituted system will allow us to dissect the molecular events underlying this process.

Fig 13. A) Functional comparison of the reovirus core particle and SFV spherules. In both cases viral polymerase and genomic RNA are inside an enclosed compartment, and newly synthesized RNA is extruded into the cytoplasm through a pore (or channel). B) A speculative model for spherule formation: after translation, the viral replicase components recruit the genomic RNA, forming a ‘pre-RC’ which is directed to the PM. Upon membrane anchoring, mediated by the nsP1 moieties, conformational changes in the RC induce the activation of helicase and/or polymerase activity. The RNA is ‘pumped’ 59

through the complex against the membranes. This results in membrane curvature and eventually spherule formation. The size of the spherule depends on the size of the RNA. Once the genome is all inside the vesicle, replication occurs, and a second conformational change deactivates the helicase, opening a passive channel through which the newly synthesized RNA can pass into the cytoplasm. Inspired from the packaging motor of the bacteriophage phi12 (194).

RNA replication and the multifunctional protein nsP2

NsP2 is a multifunctional protein. Three enzymatic activities have been described by biochemical assays in vivo and in vitro: protease, NTPase/RNA triphosphatase and helicase. The role of the protease in the context of the RC and RNA synthesis has been characterized to some extent. From studies using SFV and SIN, it has been established that the cleavages regulate the activity of the RC (reviewed in (32)). The first cleavage releases nsP4 and activates the complex for synthesis of (-)strand. The second (nsP1/nsP2) redirects the synthesis to mainly 42S +RNA. The last (nsP2/nsP3) increases the efficiency of subgenomic RNA synthesis (196). In which way the protease can ‘sense’ the different stages of RNA synthesis, we are just starting to understand. Apparently the cleavages are modulated by RNA binding (197). The NTPase and helicase activities have been demonstrated in vitro with purified proteins and substrates (73;75). However, their exact biological role in vivo could not be investigated since mutations in the active site of the NTPase are lethal for the virus. This is true for all +RNA viruses that have helicases. The discovery of nsP2 ts-mutants with RNA synthesis negative phenotypes has provided a means to study the role of this protein during replication (47).

The protease activity: indications of RC organization In my study, I have purified different recombinant variants of nsP2 with mutations that caused a ts-phenotype in the respective viruses. After optimization, I was able to study the three enzymatic activities in vitro at different temperatures. The results showed that the enzymes retained a ts-behaviour. It could therefore be demonstrated that mutations in the protease domain resulted in impaired proteolytic cleavages of purified recombinant substrates. This was consistent with the data obtained in vivo, in which the processing of the polyprotein was studied in cells infected with mutant viruses. A surprising result was that mutations in the helicase domain of the protein also affected proteolytic cleavage, and in a mutant dependent manner. One of the helicase mutants (ts1) had no defects, while the other two, ts13a and ts9, had different phenotypes. Similarly to the protease mutants, Ts13a had a general defect in the processing of all the cleavage sites. Ts9, on the other hand, cleaved the nsP3/nsP4 site like wt, but could not cleave a model 2/3 substrate in vitro, and displayed delayed processing of this site in vivo. These experiments provide indication that the helicase domain of nsP2 is in contact with the protease domain. This would explain why, in vivo, the N-terminus of nsP2 must be released from nsP1 before the next cleavage (nsP2/nsp3) can take place (198). The ts-mutations can destabilize the 60

structure of the helicase domain thus disturbing the function of the C-terminal part of the enzyme. This interpretation seems to be confirmed by further investigations on protease activity of nsP2 carried in our laboratory (197). It appears that not only the helicase domain of nsP2 plays an important role in the 2/3 cleavage, but other regions of the polyprotein are also involved. All together, the results with ts-mutants presented in this study, and the thorough biochemical studies presented in Dr. Golubtsov’s work, provide the first indications on the structural organization of the replicase polyprotein. An important suggestion is that the definition of functional domains should not be deduced solely by the primary sequence of the nsPs. In fact, functional domains might not be contained in the single nsPs, but may span them.

Role of NTPase, RNA triphosphatase and RNA helicase Using competition assays, the analysis of NTPase and RTPase indicated that the same catalytic site is used for the two reactions. This is consistent with the data obtained for helicases from other +RNA viruses (e.g. flaviviruses and coronaviruses) (150-152), but poses interesting questions concerning the organization of the replication apparatus. The NTPase activity is supposed to fuel the helicase reaction, while the RTPase needs to act only once at the 5’-end of each newly synthesized RNA, to prepare the molecule for the capping reaction (catalyzed by nsP1). One possibility is that different nsP2 molecules are dedicated to each of the reactions. Alternatively, the helicase activity could be needed only during the synthesis of minus strand, which is not capped. We hoped that by analysing the ts-mutations located in the helicase region of nsP2, the role of helicase and RTPase could be specifically addressed. However, to our surprise mutations that did impair the two reactions also caused defects in polyprotein processing, which has not been reported for similar mutations in SIN. This is the general difficulty in using ts-mutants; it is not easy to discern how a given point mutation influences the biochemical properties of a protein. Moreover, how a mutation affects one nsP could differ from the case in which the nsP is still part of a polyprotein, which is a further layer of complication. For instance, recombinant ts9-nsP2 had no NTPase activity in vitro, even at the permissive temperature. This was not an artefact associated with the production of the protein in bacteria, since nsP2 extracted from ts9-infected cells by immunoprecipitation also had no activity over background levels. Still at the permissive temperature ts9 replicates to wt levels, and if infections are carried out at low temperatures for at least 6h, further shift up to restrictive temperature does not inhibit the synthesis of genomic RNA or diminish virus yields. In this case, the only explanation I can find is that when part of the polyprotein, ts9-nsP2 still retains NTPase activity, while the free enzyme is destabilized by the mutation. Further experiments will be needed to understand the role of the helicase and RTPase during SFV infection. Probably different approaches should be considered. Here again our hopes are in the reconstituted system. It allows us to introduce more ‘focused’ mutations, e.g. in the catalytic site of each enzyme, which cannot be done in the context of virus infection since they have lethal phenotypes.

Regulation of Subgenomic promoter: future prospective A crucial role of nsP2 in the regulation of subgenomic RNA synthesis has been demonstrated using ts-mutants by the groups of Gomatos and Kääriäinen (76;199). Based on their and previous results (83), it was proposed that two pools of nsP2 would function in different reactions: as part of the RC, nsP2 would provide protease, NTPase and helicase activities; the diffusible fraction of the protein could act as a transcription factor, binding in a region immediately upstream of the subgenomic promoter, and activating its transcription. In their model, the polymerization of full size genomic RNA, initiated at the 5’end of the RNA, would have been ‘jammed’ as the polymerase nsP4 reached the region where nsP2 was bound. In my study I provided indication that the loss of 26 S RNA synthesis coincides with increased levels of genomic RNA synthesis (Fig 12), which is consistent with the results obtained by Suopanki et al. (76). Although more experiments should be performed to confirm the results, this would support the model in which nsP2 acts as a repressor of 42S RNA synthesis and, concomitantly, activator of subgenomic RNA production. I also showed that not only the protease but also the helicase region of nsP2 is deeply involved in the process. Particularly interesting was ts1, a mutant that did not have any enzymatic defects, but exhibited a severe reduction of 26S RNA synthesis, more pronounced at the higher temperatures. This mutant is a good candidate for further studies on subgenomic promoter regulation, as its defect is specific. In vitro it is very difficult to test the binding specificity of nsP2 and RNA, since the purified protein is very basic (estimated isoelectric point 9.0) and tends to aggregate in the presence of any negatively charged molecules, even heparin (unpresented results) (197). It is intriguing to imagine how the process could take place in the context of the spherules. Would the diffusible nsP2 access the RNA from the cytoplasm or inside the lumen of the spherule? Several experiments could be imagined. If it is true that soluble nsP2 can regulate the subgenomic promoter, one could use membrane fractions from infected cells for in vitro RNA synthesis assays (a well established system for alphaviruses; (48)), and add purified nsP2. Both wt and mutant variants of the protein could be tested this way. Using radioactively labelled nucleotides, the products of the reaction (42S and 26S RNAs) would then be easily analyzed by gel electrophoresis and autoradiography. The same system could be used to test if the helicase activity is required for RNA synthesis (and /or subgenomic promoter regulation). This could be done by adding γ-S-GTP, a non hydrolysable analog of the nucleotide, which would inhibit the NTPase reaction, as shown in my article, but not the polymerase activity (which uses the alpha-phosphate). Similar experiments have been successfully carried out to analyze the properties of bacteriophage packaging motors (194). Clearly, how viruses orchestrate the different steps of RNA replication/transcription within the membrane alterations remains one of the most interesting questions to be answered.

The viroplasms of SFV: spherules and CPVs

The CPVs are the hallmark of alphavirus infection. They were identified by pioneering EM studies in the late 1960s and early 1970s, and are characterized by the presence of 62

numerous ~50 nm invaginations of their limiting membrane, the spherules (93). Initially, it was thought that the appearance of spherules on the surface of endosomes was related to the entry pathway of the virus. After clathrin mediated endocytosis and delivery to endosomes, pH-triggered fusion of virus and endosome membranes results in release of the nucleocapsid into the cytoplasm. Thus, the idea was that after uncoating and translation of incoming viral RNA (the template for the first step of replication), newly synthesized nsPs would still be in the proximity of the endosomal membrane, and could form the spherules. Structural proteins were also thought to be involved in the process (10). As often happens in biology, things turned out to be more complicated. The involvement of virus entry pathway and structural proteins was excluded by Peränen and Kääriäinen, who showed a few years later that CPVs still formed after transfection of cells with replicon RNAs (94). Thus the question of the biogenesis of the CPVs remained open. Their identity as modified endosomes was firmly established by colocalization studies using antibodies against various endocytic markers and antisera against the nsPs or incorporated bromouridine, as markers for the CPVs; short pulses of endocytosed BSA- gold and immuno-EM confirmed the results (8-10). In these studies it appeared that at early times of infection CPVs colocalized more with markers of early endosomes and only later acquired markers of late endosomes and lysosomes. Their size, as judged by fluorescence microscopy and EM, increased in a time dependent manner and, concomitantly, their number decreased. Thus, the basis for a maturation process had already been provided. The missing piece of the puzzle was the detection of spherules at the PM, which was reported even in the early studies, although not connected with the CPVs - in the drawings of Grimley et al. there are arrows connecting CPVs with everything except spherules at the PM (93). In Kujala et al., the authors proposed that spherules might originate from the PM; alternatively, the structures could have appeared there as the result of fusion of CPVs with the PM. Neither of the two possibilities was ruled out in their study (9). The origin of membranes is one of the fundamental questions in the field of virus induced membrane alterations. In the case of SFV, why was the role of the PM not noticed before? The answer might reside in technical considerations. After sectioning and conventional EM, the surface of a CPV is relatively small (400-600 nm in diameter) compared to the size of the spherules. Since already at 4 h pi, the CPVs are saturated with spherules, their identification in EM is straightforward. Instead, the area of the PM is very large, and after sectioning spherules appeared always sparse (scanning-EM was never used). The same applies to fluorescence microscopy: bright spots, as the CPV appear after nsP immuno-labelling, readily capture the attention of the viewer. Instead, signals from the PM are difficult to notice in widefield microscopy, and are more evident only by confocal microscopy, which is a more recent technique. Moreover, most of the EM and fluorescence studies have been carried out at 4 hpi or later. Among the results presented in this thesis, I have shown that by that time most of the RCs have already left the PM, and relocalize on the surface of the CPVs. A time course was attempted by Kujala et al., following from 2 hpi the localization of the nsPs by immunocytochemistry (9). In their work, a clear PM signal is shown in one image, but probably the concomitant detection of the cytoplasmic pool of nsPs not engaged in replication must have made the interpretation

of the results difficult. Although developed in the early nineties, the use of a monoclonal antibody against dsRNA has become more common among virologist only in recent years. It specifically reacts with sites of +RNA virus replication, resulting in a very bright and localized signal. In the present study, the use of dsRNA detection and high multiplicities of infection allowed us to study the localization of RCs at very early times of infection, with PM signals detectable already at one hour post infection. To follow the fate of RCs in live cells, we used fluorescently labelled nsP3 (nsP3_ZsGreen), which does not bind membranes as such (nsP1 is the membrane anchor). Colocalization with dsRNA and other nsPs in fixed cells, and with PM or acidic endosomes in live cell imaging, indicated its participation in RCs. Although a fraction of nsP3 does not colocalize with RCs, this marker was useful to follow the major movements of RCs from one compartment to the next. Using CLEM, we could correlate the localization of nsP3 at the PM with thousands of spherules at 2.5 h p.i. In those areas of the PM, spherules reached densities of more than 100 per µm2. Other areas of the PM that did not have nsP3 signal did not contain spherules. These interesting results were confirmed by both electron tomography and scanning electron microscopy (after wortmannin treatment). Spherules appeared concentrated in specific areas of the PM rather than being evenly dispersed. My interpretation is that as one spherule arises at the PM, the translation of the newly synthesized mRNA, coming out of the pore of the spherule, results in new nsPs and formation of a new spherule in the nearby area of the PM. This implies that spherules do not move laterally. Indeed, when the PM attached to the substrate-coverslip was imaged by total internal reflection microscopy (TIR-FM), the fluorescence of nsP3_ZsGreen appeared as defined spots, which approached from the cytoplasm and became immobile once at the PM (unpresented results). The size of the spots was below the diffraction limit of the microscope (~200 nm), thus they could represent single spherules. More experiments using fluorescent markers for viral RNA are needed to confirm these promising results. After removal from the PM, from 2 to 3 hpi, dsRNA and nsP3 appeared in small scattered spots in the cytoplasm at the periphery of the cell. Disruption of microtubules using nocodazole stopped spherule transport at this stage. In non-treated infected cells, live cell imaging in the presence of lysotracker (a pH sensitive dye) showed that at this time of infection a portion of the nsP3_ZsGreen signal localized in non-acidic spots. Some of them docked to larger acidic organelles and engaged in long range directional transport to the perinuclear area. It is conceivable that repetition of this process results in the formation of the large and static acidic vacuoles detected later in the perinuclear area, and surrounded by nsP3_ZsGreen in live cells, or dsRNA after fixation and immunostaining. After EM analysis these vacuoles correspond to CPVs. Strikingly, at 5 hpi, virtually all acidic endosomes appeared in the form of CPVs. Thus the formation of these unique acidic organelles corresponded to a major change in the cellular endocytic compartment, which is otherwise very dynamic and characterized by organelles that periodically ‘scan’ the cytoplasm by bidirectional microtubule based transport. By 12 hpi, CPVs had reached diameters of 2 µm, by far exceeding the size of lysosomes in non-infected cells (200). Similarly to the effect of nocodazole, the traffic of spherules could be stopped by treatment with brefeldin A (BFA) (results not shown), a fungal metabolite and potent

inhibitor of anterograde membrane transport, due to inhibition of specific guanine- nucleotide exchange factors (GEFs) which in turn regulate the activity of ADP ribosylation factors (ARFs - mainly ARF-1) (201). Although spherules still formed in the presence of the drug, they remained scattered in smaller vesicles in the cytoplasm. Thus, differently from other +RNA viruses (e.g. poliovirus) (202;203), components of the secretory pathway are not required for spherule formation. That BFA can influence endosome maturation has been known for a long time ((204); Ari Helenius, personal communication), although the mechanism is not clear. Whether ARF-1 is directly involved in the process, or some other uncharacterized effect of BFA is responsible for the inhibitory effect, will be interesting to investigate in the future. At late stages of infection, EM tomography of virus-replicon-particles (VRP)-transduced cells revealed prominent ER processes surrounding almost completely the imaged CPVs. This distribution of ER and CPVs was reproducible and appeared in many images of parallel experiments. Interestingly, in cells infected with viruses, the ER processes were still close to CPVs but less extensive. It was proposed that the association of ER membranes with sites of RNA synthesis could be related to the synthesis of envelope glycoproteins, which must be inserted into the ER lumen for maturation. Again, this was not the case, since the tomography was obtained from cells transduced with VRPs (the genes encoding for capsid and envelope are deleted from the genome). An interesting question is whether the ER processes seen in our tomograms have the same biological role as those typically detected in SFV infected cells. To this end, it is important to note that the ER membranes close to SFV induced CPVs are decorated by ribosomes (rough ER) (10;92), while in replicon transfected cells the membranes that wrap the CPVs appear rather smooth. I speculate that the latter could be a sign of a cellular defence mechanism, maybe similar to autophagocytosis. If this is true, the structural proteins could play a role in preventing this process. As previously noticed by several colleagues, nucleocapsids are readily found in the vicinity of the spherule necks (10). In this respect, the CPVs fit the definition given for viroplasms or virus factories: sites where viral RNA replication is coupled with the assembly of new virions (205). However, SFV buds from the plasma membrane, and thus at least this step of the virus life cycle seems to be disconnected from the viroplasm area.

Comparison with viroplasms induced by other +RNA viruses

Among the six families of +RNA viruses that contain human pathogens, the biogenesis of virus-induced membrane alteration has been more extensively addressed for members of the Picornaviridae, Flaviviridae, Coronaviridae and Togaviridae (1;3;4). The work on Astroviridae and Caliciviridae is much less extensive (11;12). Except for the Togaviridae (Alphaviruses and Rubiviruses), the rest seem to utilize membranes of the secretory compartment, e.g. ER and Golgi. The morphological characterization of these complex compartments is based on EM studies. Depending on the fixation method (chemical versus fast freezing) and orientation of sectioning, the membrane rearrangements may appear quite heterogeneous. However, recent studies where EM tomography was applied to the 65

viroplasms of SARS coronavirus and dengue flavivirus suggest that these compartments can be more similar to each other than previously thought (18;19). Indeed, the network of convoluted membranes and double membrane vesicles (DMVs) previously described in conventional EM studies, appeared as one interconnected compartment. Comparison of the 3D organization of vesicles (site of RNA synthesis) and the remaining membranes clearly shows that in both cases vesicles are contained in the lumen of the ER. In the case of dengue virus, the authors used chemical fixation and describe the pore of the vesicle as open toward the cytoplasm (19). The study on SARS coronavirus was done using fast- freezing and acetone substitution. In this case the pore was not identified (18). Assuming that a connection with the cytoplasm really exists (at least I believe so), the main difference between the flavivirus and coronavirus compartment is the size of the vesicles, larger in the case of coronaviruses (see table 1). In fact, the size of the spherules seems to correlate with the size of the virus genome which they presumably contain in their lumen. The topology is also the same, a vesicle forming as an invagination of the membrane of a cytoplasmic organelle, e.g. ER, PM, mitochondria, etc. Even for poliovirus-induced structures, previously described as single membrane ‘vesicle packets’ after chemical fixation and conventional EM, it has been recently shown that DMVs are formed in the lumen of the ER (206). Poliovirus genome is about 8 kb, smaller than the SFV genome (11.5 kb). The size of poliovirus vesicles varies from 70 to 400 nm, thus far larger than the vesicles induced by SFV, which are 50 to 70 nm. However, similarly to the CPVs of alphaviruses, the viroplasm induced by poliovirus (and in this respect by the rest of +RNA viruses that infect animals) seems also to involve fusion events. In the specific case of poliovirus, this ‘fusogenic’ activity has been attributed to the involvement of ARF, COP-I and COP-II components, which apparently are recruited by the replicase components to the modified membranes and could mediate fusion with other membranes bearing the appropriate tether (summarized in (3)). Similar mechanisms have been proposed for the formation of HCV induced membrane alterations (207). In the case of the CPV, our preliminary results indicate that the EEA-1, a master regulator of endosome fusion, seems to be recruited at late stages of CPV maturation, and this could explain why CPV continue to fuse reaching sizes of 2 µm at 12 hpi. More in general, the appearance of large compartments could in part be the result of artefacts caused by aldehyde-induced fusion of the highly packed membranes, similar to the lateral fusion of spherules that we see after similar treatments, in CPVs where these structures are highly packed side by side. Instead, SFV spherules are all separate after high pressure freezing and have regular sizes ranging from 50 to 60 nm, stressing the importance of sample preservation during morphological studies. A major finding presented in this thesis is that the appearance of ‘active’ spherules (sites of viral RNA synthesis) on the limiting membranes of the CPVs is the consequence of microtubule based transport, and apparently not of direct targeting of the nsPs and consequent spherule formation, as was thought previously. This finding highlights a striking similarity with the biogenesis of other +RNA virus viroplasms. The complex and extensive membrane alterations induced by poliovirus (Picornaviridae), Hepatitis C virus (HCV) (Flaviridae) and Mouse hepatitis virus (MHV) (Coronaviridae) consist of DMVs embedded in convoluted membranes, apparently of ER origin, which at late stages of

infection accumulate in the perinuclear area. Treatments with nocodazole at early times of infection result in a more dispersed distribution of RCs, as detected by dsRNA or nsPs labelling (208;209). At this stage, I would like to stress the distinction between spherules (or vesicles, as they are called for the other +RNA viruses mentioned above) and the remaining altered membranes, where these peculiar structures progressively accumulate during infection. The spherules/vesicles support RNA replication, whereas the formation of the final compartment is attributed to the facilitation of virion assembly, although this still remains an interpretation. There is no evidence that virion assembly really requires such compartmentalization. In fact, virion formation was not affected by microtubule disruption in the case of poliovirus, as elegantly shown by the pioneering and inspiring work of the Kurt Bienz and Denis Egger laboratory (209). Importantly, poliovirus is released after cell lysis, whereas for SFV and other viruses which require functional secretion, microtubules are also needed for other functions than RC trafficking (e.g. delivery of envelope proteins to the PM in the case of SFV). Thus, such experiments have to be limited to the detection of RNA synthesis and not the release of virion particles. Consistently with the results obtained for poliovirus, RNA synthesis was not affected by nocodazole treatment in the case of SFV (this work and our unpublished data), MHV (208) and HCV (210), as judged by the continuous accumulation of nsPs, viral RNA and, in the case of MHV, even virus release. It is important to notice that in the case of HCV, the redistribution of RCs after microtubule disruption was not as pronounced as in the other cited studies. Wölk at al. used a cell line stably expressing a HCV replicon (210), whereas all the other studies were performed using virus infection. Even in the absence of drugs, the RCs formed in HCV replicon cell lines never really accumulated in one area of the cell. In my opinion, the use of stable cell lines, in which RCs and induced membrane rearrangements are constantly in a ‘steady state’, is probably not appropriate if the scope of the study is to understand the formation of those structures. Moreover, the authors use only one of the nsPs, NS5A, as a marker for RCs. Nonstructural proteins are always produced in excess and the use of more specific markers for RNA replication should always be included, at least in parallel experiments. In the specific case of NS5A, the protein has membrane binding activity and it has recently been shown that it also binds to endosomes, which could have influenced the conclusions of Wölk et al. While an increasing amount of evidence indicates a role in viroplasm formation for proteins involved in membrane traffic and lipid metabolism, the major question still remains the formation of the vesicles (although many researchers do not make a difference between the two terms). The best way to address this question would be to have viruses that form vesicles which do not concentrate in complex structures in the perinuclear area. As we have shown, the combination of inhibitor treatments (e.g. nocodazole) and CLEM could provide one way to achieve this result. On the other hand, our discovery that sequences in one of the nsPs are responsible for spherule trafficking and not spherule formation, will hopefully open the way to similar findings in other +RNA viruses, which will undoubtedly make the studies of vesicle biogenesis easier. At least it will make easier the interpretation of electron-micrographs, and help virologists to find a common understanding of basic concepts such as the topology of these characteristic structures. This is fundamental in drawing correct models, which are the basis of any experimental

planning. To this end, I must make a comment on the ‘state of the art’ models proposed in recent articles and reviews on this topic. For all +RNA viruses that induce membrane rearrangements of ER origin (all the studied animal viruses except Togaviridae), components of the secretory pathway have been implicated in the formation of spherules/vesicles. These implications arise from colocalization studies, sensitivity to BFA and, more recently, siRNA studies (205). Not all viruses are affected by these treatments in the same way; some require a certain factor whereas others do not. Nevertheless, the models are all based on the same concept: recruiting proteins like ARFs or components of the coatomers (COPI and COPII) will help to induce membrane curvature, resulting in spherule formation (1). That these proteins could be involved in some important steps during RC assembly and vesicle formation (e.g. targeting of RCs to correct membranes) I have nothing against. Some pieces of evidence are quite clear in this respect (202;211;212). My only comment is that the vesicles induced during anterograde and retrograde transport are topologically opposite to those induced by viruses. The first bud towards the cytoplasmic space; the spherules away from it, towards the lumen of the organelle where they are forming. In the case of vesiculation mediated by protein coats, the topology of the process is dictated by the intrinsic properties of the coat protein (i.e. the shape of the protein, positioning of amphipathic helices, transmembrane domains, post translational modifications, lateral interactions, etc.) (178). To my knowledge, the same proteins have never been reported to bend membranes in the opposite direction. Recently, a role in double membrane vesicle formation for proteins involved in autophagosome maturation, originally proposed by George Palade in 1968, has been reconsidered by the Kirkegaard laboratory (Standford, USA) (23). Although the autophagosome model would account for the topology of the induced vesicles, the data are still very controversial. Some viruses are sensitive to siRNA-mediated depletion of proteins involved in authophagosome formation and maturation, while others are not (13;20;23;213). As the field of autophagocytosis is currently one of the hot topics in cell biology, hopefully in the next few years new findings will reveal the role of this process in viroplasm formation.

Internalization of spherules from the PM: spherules as unconventional endocytic cargoes

Spherules and caveolae. Our study of the intracellular dynamics of RCs (I) clearly showed that after accumulating at the PM, spherules are subsequently transferred to endosomes, where they appear as invaginations of the limiting membrane of the organelle. The lumen of an endosome is topologically analogous to the extracellular space. Thus spherules at the PM are oriented in the same way as those in the CPVs (they ‘bud’ away from the cytoplasm). The limiting membrane of the spherules continues with the PM, and the interior of the spherules remains connected with the cytosol through the pore structure. Can such a structure be compared with an endocytic cargo? The main body of the spherule has a similar size as a virus (SFV for instance). The topology is the same as that of a virus attached to a cell-surface receptor. We have used CLEM and tomography to visualize spherules at the stage of internalization (II). We found 15 of them inside smooth vesicles, 68

resembling in size and shape caveolae or multi-caveolar structures (116). In the imaged area, such endocytic processes were very abundant (not shown in the 3D model), and most did not contain spherules (fig 10). We also found several well defined CCV at late stage of internalization. Despite the high density of spherules at the PM, none of them was inside a CCV. Previous studies, in which conventional EM was used, showed spherules inside small endocytic vesicles. In Kujala et al., fortuitous sections are shown where spherules appear inside structures similar to those described in the present study (9). In Froshauer et al., one micrograph shows one spherule in a CCV, while in a nearby image several spherules are shown inside multi-lobed vacuoles, morphologically similar to the structures shown in our tomograms (10). So caveolae or clathrin mediated endocytosis? In the present work (II) we have started to address this question. It was found that at early stages of infection the distribution of dsRNA did not follow that of clathrin (and Tf) in immunocytochemistry assays. Instead, the RCs accumulated in areas of the PM that also contained cav1. Notably, at higher magnifications the dsRNA and cav1 signals could not be perfectly superimposed. This could be explained considering that the two markers label the outer membrane of caveolae and the interior of the spherules, respectively. Indeed, the cav1 staining was found on the cytoplasmic side of the PM and endosomes while the dsRNA signal was on the opposite side. Given that the majority of spherules are ‘sitting’ on the PM and only a minority of them is internalized per unit of time, the colocalization results at the PM were somewhat expected. Currently we are characterizing the role of cav1 in RC internalization using live cell TIR-FM imaging of HeLa cells stably expressing cav1-GFP. A functional role for cav1 during spherule endocytosis was confirmed by RNAi treatments. The results showed that depletion of cav1 induced a significant accumulation of dsRNA at the PM even at late stages of infection. A similar delay was found in infected Huh-7 cells, which do not express cav1. These preliminary results, although not quantitative, indicate that cav1 plays a functional role during spherule internalization. However, whether this is an active role is not easy to assess. In fact, the definition of caveolae as PM invaginations where cav1 localizes, can lead to erroneous conclusions as for the role of this protein during the endocytosis of a given cargo. Indeed, cav1 and caveolae are localized in areas of the PM which are rich in cholesterol (‘rafts’) (120). This is also true for other types of clathrin- independent endocytic events (128). Moreover, it has been proposed that caveolins could stabilize endocytically active raft-like domains of the PM (214) and even act as negative regulators of caveolae endocytosis (215). Thus, similar distributions of dsRNA and cav1, and the effect of cav1-siRNA could simply reflect an indirect effect of cav1 on a process that occurs in similar cholesterol enriched domains of the PM. This raises the crucial question of whether all morphologically defined caveolae in fact contain cav1. If yes, as indicated by the loss of these structures in non-muscle cells derived from cav1-/- mice (216-218), then spherules are internalized via cav1-positive-caveolae. If not, as stressed by Dr. Nichols (130;215;219), then a broader approach should be considered to pinpoint the factors responsible for the morphogenesis of the structures described in our tomograms.

Role of cholesterol. Acute cholesterol depletion affects both clathrin and caveolae dependent endocytosis (220-222). Depletion of cellular cholesterol, by means of

cyclodextrin treatments, resulted in a strong inhibition of spherule transport from the PM and consequent delay in CPV formation (II). Importantly, the amount of cyclodextrin used was kept to a minimum, such that at least 80% of CME, as measured by Tf uptake, still occurred in control non-infected cells. Thus the effect was specific. These results are consistent with caveolae-mediated endocytosis, but are not conclusive, since also other endocytic pathways are influenced by the levels of PM cholesterol (214;223). On the other hand, cholesterol was not needed for spherule formation as shown by dsRNA staining and EM analysis of cells treated with cyclodextrin starting from 20 min after infection. Consistent with an internalization of spherules from cholesterol-enriched PM domains, the accumulation of spherules in the CPVs resulted in a concomitant accumulation of cholesterol in these vacuoles, as measured by filipin staining.

NsP3 and the signal for spherule internalization. While more experiments will be required to better characterize the role of the cellular components (i.e. caveolins, cavins, flotilins, RhoGTPases, etc.) involved during spherule internalization, the finding that this event required the activity of the class I PI3K indicated a highly regulated event (I) (224). This enzyme plays a major role in macropinocytosis and phagocytosis (132;133;224), but to my knowledge its function has not been reported to be necessary for the initiation of caveolae/raft-mediated endocytosis. In fact, in a genome-wide siRNA screen of human kinases involved in CME and caveola/raft dependent endocytosis, this enzyme and its effector kinases were rather connected to CME (124). In this study, the authors used SV40 as a marker for caveolae/raft dependent endocytosis. The entry of this virus has now been attributed to non-caveolae endocytosis. Perhaps this could have influenced the results of Pelkmans and colleagues (124). In our hands, the use of a specific inhibitor of the class I PI3K, the recently developed IP-103 (225), or specific siRNA treatments resulted in a virtually complete block of RCs at the PM, while the same treatment did not affect the internalization of Tf in control non infected cells. Altogether, the results so far obtained indicated that the internalization of SFV spherules from the PM could represent an ‘exaggerated’ form of an otherwise low frequency process, regulated by phosphorylation and dephosphorylation events. The question was raised whether this unconventional process could be triggered by one of the replicase components. Salonen et al. had previously shown that the intracellular localization of the replicase precursor polyprotein was influenced by the presence of nsP3. Particularly, ectopic expression of nsP1+nsP2 resulted in PM localisation, whereas addition of nsP3 (P123) shifted the localization of the polyprotein to PM and endosomes (8). NsP3 is required for virus replication but its biological function is unknown. Particularly, the C-terminus of the protein is highly phosporylated, and deletions of this region result in viruses that are viable in cell cultures (and give same titers as wt viruses at 12 hpi), but are not pathogenic when introduced in mice (226). We wanted to test whether the same deletion had an effect on spherule endocytosis. The surprising result suggested that the amino acid sequence of nsP3 could be responsible for RC spherule internalization, as a 50 aa deletion in the C-terminal phosphorylated region of the protein resulted in a complete block of this process. However, the possibility that the deletion induced pleiotropic effects on some of the other nsPs (or cellular factors) is not excluded.

Mechanism of spherule endocytosis. Considering the known mechanisms of endocytosis, at least two different mechanisms for spherule internalization can be imagined: i) spherules interact directly with the endocytic machinery and are internalized with similar mechanisms as other endocytic cargoes; ii) spherules are not ‘sensed’ by the cell, and instead a global endocytic event is triggered by a cellular or viral factor, which induces the formation of the described endocytic processes over large surface areas of the PM. Spherules, which are very abundant at the PM, would be ‘trapped’ and internalized in such structures passively. For numerous reasons I prefer the second model. The fact that spherules accumulate at the PM during the first two hours of infection indicates that their internalization is a limiting step. The appearance of numerous cytoplasmic structures from 2 to 3 hpi suggests an activation process, which could be linked to the activity of PI3K. More importantly, the fact that spherules can remain at the PM, as shown for the nsP3 deletion mutant, indicates that these structures are not actually sensed by the cell to specifically induce their endocytosis. In my model the endocytic event would not be induced until sufficient amounts of phosphoinositol-3,4,5-triphosphate (PIP3), the product of the class I PI3K, are produced to create a platform for recruitment of the endocytic machinery. Such PIP3 thresholds have been proposed for the activation of phagocytosis and large scale macropinocytic events induced by growth factors and for vaccinia virus entry (132-134). The recruitment of the PI3K to the PM could be mediated by the pool of polyprotein P123 which is not participating in replication. The nsP1 moiety would provide membrane binding; while nsP3 could mediate directly or indirectly the recruitment of at least the catalytic subunit of the PI3K. Such a mechanism has been demonstrated for the accessory protein of HIV-1 Nef, which also induces unconventional endocytic events by recruiting the PI3K to specific sites of the PM (227;228). If this model is correct, ectopic expression of wt P123 should rescue the internalization of spherules in the case of the virus with deletions in nsP3. In fact, if spherule internalization is due to caveolae/raft-dependent endocytosis, as our data indicate, the same results could be achieved by addition of phosphatase inhibitors (e.g. okadaic acid), which are known to induced caveolae internalization and, interestingly, similar grape-like structures as those formed during SFV infection and described in our tomographic reconstructions (229;230). Such a massive and raft specific endocytic event could result in important changes in the composition of the PM and endosomes, which could affect their functions. Thus, the internalization of RCs would represent a marker of a larger event that SFV triggers at early stages of infection. Similarly to the role of Nef, this could be an important determinant of virus pathogenicity. While in vivo experiments should be performed to test the possibility, the finding that this process can be stopped by inhibitors of the class-I PI3K, opens the possibility to test similar compounds as antivirals for alphavirus infections. Many analogs already exist and are in clinical trials as anticancer (231), thus it should not be too difficult to test them in mice against SFV or the re- emerging Chikungunya virus.

References

(1) Miller S, Krijnse-Locker J. Modification of intracellular membrane structures for virus replication. Nat Rev Microbiol 2008 May;6(5):363-74. (2) Netherton C, Moffat K, Brooks E, Wileman T. A guide to viral inclusions, membrane rearrangements, factories, and viroplasm produced during virus replication. Adv Virus Res 2007;70:101-82. (3) Mackenzie J. Wrapping things up about virus RNA replication. Traffic 2005 November;6(11):967-77. (4) Salonen A, Ahola T, Kaariainen L. Viral RNA replication in association with cellular membranes. Curr Top Microbiol Immunol 2005;285:139-73. (5) Schwartz M, Chen JB, Janda M, Sullivan M, den Boon J, Ahlquist P. A positive-strand RNA virus replication complex parallels form and function of retrovirus capsids. Molecular Cell 2002 March;9(3):505-14. (6) Miller DJ, Schwartz MD, Ahlquist P. Flock house virus RNA replicates on outer mitochondrial membranes in Drosophila cells. J Virol 2001 December;75(23):11664-76. (7) Jonczyk M, Pathak KB, Sharma M, Nagy PD. Exploiting alternative subcellular location for replication: tombusvirus replication switches to the endoplasmic reticulum in the absence of peroxisomes. Virology 2007 June 5;362(2):320-30. (8) Salonen A, Vasiljeva L, Merits A, Magden J, Jokitalo E, Kääriäinen L. Properly folded nonstructural polyprotein directs the semliki forest virus replication complex to the endosomal compartment. J Virol 2003 February;77(3):1691-702. (9) Kujala P, Ikäheimonen A, Ehsani N, Vihinen H, Auvinen P, Kääriäinen L. Biogenesis of the Semliki Forest virus RNA replication complex. J Virol 2001 April;75(8):3873-84. (10) Froshauer S, Kartenbeck J, Helenius A. Alphavirus RNA replicase is located on the cytoplasmic surface of endosomes and lysosomes. J Cell Biol 1988 December;107(6 Pt 1):2075-86. (11) Hyde JL, Sosnovtsev SV, Green KY, Wobus C, Virgin HW, Mackenzie JM. Mouse norovirus replication is associated with virus-induced vesicle clusters originating from membranes derived from the secretory pathway. J Virol 2009 October;83(19):9709-19. (12) Guix S, Caballero S, Bosch A, Pinto RM. C-terminal nsP1a protein of human astrovirus colocalizes with the endoplasmic reticulum and viral RNA. J Virol 2004 December;78(24):13627-36. (13) Snijder EJ, van der MY, Zevenhoven-Dobbe J, Onderwater JJ, van der MJ, Koerten HK et al. Ultrastructure and origin of membrane vesicles

associated with the severe acute respiratory syndrome coronavirus replication complex. J Virol 2006 June;80(12):5927-40. (14) Mackenzie JM, Kenney MT, Westaway EG. West Nile virus strain Kunjin NS5 polymerase is a phosphoprotein localized at the cytoplasmic site of viral RNA synthesis. Journal of General Virology 2007 April;88:1163-8. (15) Moradpour D, Gosert R, Egger D, Penin F, Blum HE, Bienz K. Membrane association of hepatitis C virus nonstructural proteins and identification of the membrane alteration that harbors the viral replication complex. Antiviral Res 2003 October;60(2):103-9.

(16) Barcena M., Koster AJ. Electron tomography in life science. Seminars in Cell and Developmental Biology 20, 920-930. 5-8-2009.

(17) Kopek BG, Perkins G, Miller DJ, Ellisman MH, Ahlquist P. Three- dimensional analysis of a viral RNA replication complex reveals a virus- induced mini-organelle. PLoS Biol 2007 September;5(9):e220. (18) Knoops K, Kikkert M, Worm SH, Zevenhoven-Dobbe JC, van der MY, Koster AJ et al. SARS-coronavirus replication is supported by a reticulovesicular network of modified endoplasmic reticulum. PLoS Biol 2008 September 16;6(9):e226. (19) Welsch S, Miller S, Romero-Brey I, Merz A, Bleck CK, Walther P et al. Composition and three-dimensional architecture of the dengue virus replication and assembly sites. Cell Host Microbe 2009 April 23;5(4):365-75. (20) Prentice E, Jerome WG, Yoshimori T, Mizushima N, Denison MR. Coronavirus replication complex formation utilizes components of cellular autophagy. J Biol Chem 2004 March 12;279(11):10136-41. (21) Wang RYL, Nagy PD. Tomato bushy stunt virus co-opts the RNA-binding function of a host metabolic enzyme for viral genomic RNA synthesis. Cell Host & Microbe 2008 March;3(3):178-87. (22) Barajas D, Jiang Y, Nagy PD. A unique role for the host ESCRT proteins in replication of Tomato bushy stunt virus. PLoS Pathog 2009 December;5(12):e1000705. (23) Jackson WT, Giddings TH, Jr., Taylor MP, Mulinyawe S, Rabinovitch M, Kopito RR et al. Subversion of cellular autophagosomal machinery by RNA viruses. PLoS Biol 2005 May;3(5):e156. (24) Strauss JH, Strauss EG. The alphaviruses: gene expression, replication, and evolution. Microbiol Rev 1994 September;58(3):491-562. (25) Fazakerley JK. Pathogenesis of Semliki Forest virus encephalitis. J Neurovirol 2002 December;8 Suppl 2:66-74. (26) Her Z, Kam YW, Lin RT, Ng LF. Chikungunya: a bending reality. Microbes Infect 2009 December;11(14-15):1165-76.

(27) Jose J, Snyder JE, Kuhn RJ. A structural and functional perspective of alphavirus replication and assembly. Future Microbiol 2009 September;4:837-56. (28) Klimstra WB, Nangle EM, Smith MS, Yurochko AD, Ryman KD. DC-SIGN and L-SIGN can act as attachment receptors for alphaviruses and distinguish between mosquito cell- and mammalian cell-derived viruses. J Virol 2003 November;77(22):12022-32. (29) Helenius A, Marsh M. Endocytosis of enveloped animal viruses. Ciba Found Symp 1982;(92):59-76.

(30) Singh I, Helenius A. Role of ribosomes in Semliki Forest virus nucleocapsid uncoating. J Virol 1992 December;66(12):7049-58. (31) Wengler G. The regulation of disassembly of alphavirus cores. Arch Virol 2009;154(3):381-90. (32) Kaariainen L, Ahola T. Functions of alphavirus nonstructural proteins in RNA replication. Prog Nucleic Acid Res Mol Biol 2002;71:187-222. (33) Hardy WR, Strauss JH. Processing the nonstructural polyproteins of sindbis virus: nonstructural proteinase is in the C-terminal half of nsP2 and functions both in cis and in trans. J Virol 1989 November;63(11):4653-64. (34) Hardy WR, Strauss JH. Processing the nonstructural polyproteins of Sindbis virus: study of the kinetics in vivo by using monospecific antibodies. J Virol 1988 March;62(3):998-1007. (35) Vasiljeva L, Merits A, Golubtsov A, Sizemskaja V, Kaariainen L, Ahola T. Regulation of the sequential processing of Semliki Forest virus replicase polyprotein. J Biol Chem 2003 October 24;278(43):41636-45. (36) Lemm JA, Rice CM. Roles of nonstructural polyproteins and cleavage products in regulating Sindbis virus RNA replication and transcription. J Virol 1993 April;67(4):1916-26. (37) Lemm JA, Rumenapf T, Strauss EG, Strauss JH, Rice CM. Polypeptide requirements for assembly of functional Sindbis virus replication complexes: a model for the temporal regulation of minus- and plus-strand RNA synthesis. EMBO J 1994 June 15;13(12):2925-34. (38) Gorchakov R, Frolova E, Frolov I. Inhibition of transcription and translation in Sindbis virus-infected cells. J Virol 2005 August;79(15):9397- 409. (39) Garmashova N, Gorchakov R, Volkova E, Paessler S, Frolova E, Frolov I. The Old World and New World alphaviruses use different virus-specific proteins for induction of transcriptional shutoff. J Virol 2007 March;81(5):2472-84. (40) Gorchakov R, Frolova E, Williams BR, Rice CM, Frolov I. PKR-dependent and -independent mechanisms are involved in translational shutoff during Sindbis virus infection. J Virol 2004 August;78(16):8455-67.

(41) Ventoso I, Sanz MA, Molina S, Berlanga JJ, Carrasco L, Esteban M. Translational resistance of late alphavirus mRNA to eIF2alpha phosphorylation: a strategy to overcome the antiviral effect of protein kinase PKR. Genes Dev 2006 January 1;20(1):87-100. (42) Sjoberg EM, Garoff H. The translation-enhancing region of the Semliki Forest virus subgenome is only functional in the virus-infected cell. J Gen Virol 1996 June;77 ( Pt 6):1323-7. (43) Sanz MA, Castello A, Ventoso I, Berlanga JJ, Carrasco L. Dual mechanism for the translation of subgenomic mRNA from Sindbis virus in infected and uninfected cells. PLoS One 2009;4(3):e4772. (44) Garoff H, Sjoberg M, Cheng RH. Budding of alphaviruses. Virus Res 2004 December;106(2):103-16. (45) Hahn YS, Strauss EG, Strauss JH. Mapping of RNA- temperature-sensitive mutants of Sindbis virus: assignment of complementation groups A, B, and G to nonstructural proteins. J Virol 1989 July;63(7):3142-50. (46) Hahn YS, Grakoui A, Rice CM, Strauss EG, Strauss JH. Mapping of RNA- temperature-sensitive mutants of Sindbis virus: complementation group F mutants have lesions in nsP4. J Virol 1989 March;63(3):1194-202. (47) Keranen S, Kaariainen L. Isolation and basic characterization of temperature-sensitive mutants from Semliki Forest virus. Acta Pathol Microbiol Scand B Microbiol Immunol 1974 December;82(6):810-20. (48) Barton DJ, Sawicki SG, Sawicki DL. Solubilization and immunoprecipitation of alphavirus replication complexes. J Virol 1991 March;65(3):1496-506. (49) Kujala P, Ikaheimonen A, Ehsani N, Vihinen H, Auvinen P, Kaariainen L. Biogenesis of the Semliki Forest virus RNA replication complex. J Virol 2001 April;75(8):3873-84. (50) Gorchakov R, Garmashova N, Frolova E, Frolov I. Different types of nsP3- containing protein complexes in Sindbis virus-infected cells. J Virol 2008 October;82(20):10088-101. (51) Lemm JA, Rice CM. Assembly of functional Sindbis virus RNA replication complexes: requirement for coexpression of P123 and P34. J Virol 1993 April;67(4):1905-15. (52) Lemm JA, Bergqvist A, Read CM, Rice CM. Template-dependent initiation of Sindbis virus RNA replication in vitro. J Virol 1998 August;72(8):6546- 53. (53) Thal MA, Wasik BR, Posto J, Hardy RW. Template requirements for recognition and copying by Sindbis virus RNA-dependent RNA polymerase. Virology 2007 February 5;358(1):221-32. (54) Laakkonen P, Auvinen P, Kujala P, Kaariainen L. Alphavirus replicase protein NSP1 induces filopodia and rearrangement of actin filaments. J Virol 1998 December;72(12):10265-9.

(55) De Groot RJ, Rumenapf T, Kuhn RJ, Strauss EG, Strauss JH. Sindbis virus RNA polymerase is degraded by the N-end rule pathway. Proc Natl Acad Sci U S A 1991 October 15;88(20):8967-71. (56) Kamer G, Argos P. Primary structural comparison of RNA-dependent polymerases from plant, animal and bacterial viruses. Nucleic Acids Res 1984 September 25;12(18):7269-82. (57) Barton DJ, Sawicki SG, Sawicki DL. Demonstration in vitro of temperature-sensitive elongation of RNA in Sindbis virus mutant ts6. J Virol 1988 October;62(10):3597-602.

(58) Tomar S, Hardy RW, Smith JL, Kuhn RJ. Catalytic core of alphavirus nonstructural protein nsP4 possesses terminal adenylyltransferase activity. J Virol 2006 October;80(20):9962-9. (59) Laakkonen P, Hyvonen M, Peranen J, Kaariainen L. Expression of Semliki Forest virus nsP1-specific methyltransferase in insect cells and in Escherichia coli. J Virol 1994 November;68(11):7418-25. (60) Ahola T, Kaariainen L. Reaction in alphavirus mRNA capping: formation of a covalent complex of nonstructural protein nsP1 with 7-methyl-GMP. Proc Natl Acad Sci U S A 1995 January 17;92(2):507-11. (61) Ahola T, Laakkonen P, Vihinen H, Kaariainen L. Critical residues of Semliki Forest virus RNA capping enzyme involved in methyltransferase and guanylyltransferase-like activities. J Virol 1997 January;71(1):392-7. (62) Mi S, Stollar V. Expression of Sindbis virus nsP1 and methyltransferase activity in Escherichia coli. Virology 1991 September;184(1):423-7. (63) Mi S, Durbin R, Huang HV, Rice CM, Stollar V. Association of the Sindbis virus RNA methyltransferase activity with the nonstructural protein nsP1. Virology 1989 June;170(2):385-91. (64) Rozanov MN, Koonin EV, Gorbalenya AE. Conservation of the putative methyltransferase domain: a hallmark of the 'Sindbis-like' supergroup of positive-strand RNA viruses. J Gen Virol 1992 August;73 ( Pt 8):2129-34. (65) Ahola T, Lampio A, Auvinen P, Kaariainen L. Semliki Forest virus mRNA capping enzyme requires association with anionic membrane phospholipids for activity. EMBO J 1999 June 1;18(11):3164-72. (66) Spuul P, Salonen A, Merits A, Jokitalo E, Kaariainen L, Ahola T. Role of the amphipathic peptide of Semliki forest virus replicase protein nsP1 in membrane association and virus replication. J Virol 2007 January;81(2):872-83. (67) Peranen J, Laakkonen P, Hyvonen M, Kaariainen L. The Alphavirus Replicase Protein Nsp1 Is Membrane-Associated and Has Affinity to Endocytic Organelles. Virology 1995 April 20;208(2):610-20. (68) Laakkonen P, Ahola T, Kaariainen L. The effects of palmitoylation on membrane association of Semliki forest virus RNA capping enzyme. J Biol Chem 1996 November 8;271(45):28567-71.

(69) Lampio A, Kilpelainen I, Pesonen S, Karhi K, Auvinen P, Somerharju P et al. Membrane binding mechanism of an RNA virus-capping enzyme. J Biol Chem 2000 December 1;275(48):37853-9. (70) Vasiljeva L, Valmu L, Kaariainen L, Merits A. Site-specific protease activity of the carboxyl-terminal domain of Semliki Forest virus replicase protein nsP2. J Biol Chem 2001 August 17;276(33):30786-93. (71) Balistreri G, Caldentey J, Kaariainen L, Ahola T. Enzymatic defects of the nsP2 proteins of Semliki Forest virus temperature-sensitive mutants. J Virol 2007 March;81(6):2849-60.

(72) Merits A, Vasiljeva L, Ahola T, Kaariainen L, Auvinen P. Proteolytic processing of Semliki Forest virus-specific non-structural polyprotein by nsP2 protease. J Gen Virol 2001 April;82(Pt 4):765-73. (73) Rikkonen M, Peranen J, Kaariainen L. ATPase and GTPase activities associated with Semliki Forest virus nonstructural protein nsP2. J Virol 1994 September;68(9):5804-10. (74) Vasiljeva L, Merits A, Auvinen P, Kaariainen L. Identification of a novel function of the alphavirus capping apparatus. RNA 5'-triphosphatase activity of Nsp2. J Biol Chem 2000 June 9;275(23):17281-7. (75) Gomez dC, Ehsani N, Mikkola ML, Garcia JA, Kaariainen L. RNA helicase activity of Semliki Forest virus replicase protein NSP2. FEBS Lett 1999 April 1;448(1):19-22. (76) Suopanki J, Sawicki DL, Sawicki SG, Kaariainen L. Regulation of alphavirus 26S mRNA transcription by replicase component nsP2. J Gen Virol 1998 February;79 ( Pt 2):309-19. (77) Breakwell L, Dosenovic P, Karlsson Hedestam GB, D'Amato M, Liljestrom P, Fazakerley J et al. Semliki Forest virus nonstructural protein 2 is involved in suppression of the type I interferon response. J Virol 2007 August;81(16):8677-84. (78) Fazakerley JK, Boyd A, Mikkola ML, Kaariainen L. A single amino acid change in the nuclear localization sequence of the nsP2 protein affects the neurovirulence of Semliki Forest virus. J Virol 2002 January;76(1):392-6. (79) Koonin EV, Dolja VV. Evolution and taxonomy of positive-strand RNA viruses: implications of comparative analysis of amino acid sequences. Crit Rev Biochem Mol Biol 1993;28(5):375-430. (80) Gorbalenya AE, Koonin EV, Lai MM. Putative papain-related thiol proteases of positive-strand RNA viruses. Identification of rubi- and aphthovirus proteases and delineation of a novel conserved domain associated with proteases of rubi-, alpha- and coronaviruses. FEBS Lett 1991 August 19;288(1-2):201-5. (81) Lulla A, Lulla V, Tints K, Ahola T, Merits A. Molecular determinants of substrate specificity for Semliki Forest virus nonstructural protease. J Virol 2006 June;80(11):5413-22.

(82) Lulla V, Merits A, Sarin P, Kaariainen L, Keranen S, Ahola T. Identification of mutations causing temperature-sensitive defects in Semliki Forest virus RNA synthesis. J Virol 2006 March;80(6):3108-11. (83) Simmons DT, Strauss JH. Replication of Sindbis virus. I. Relative size and genetic content of 26 s and 49 s RNA. J Mol Biol 1972 November 28;71(3):599-613. (84) Sawicki DL, Kaariainen L, Lambek C, Gomatos PJ. Mechanism for control of synthesis of Semliki Forest virus 26S and 42s RNA. J Virol 1978 January;25(1):19-27.

(85) Kaariainen L, Ahola T. Functions of alphavirus nonstructural proteins in RNA replication. Prog Nucleic Acid Res Mol Biol 2002;71:187-222. (86) Malet H, Coutard B, Jamal S, Dutartre H, Papageorgiou N, Neuvonen M et al. The crystal structures of Chikungunya and Venezuelan equine encephalitis virus nsP3 macro domains define a conserved adenosine binding pocket. J Virol 2009 July;83(13):6534-45. (87) Peranen J, Takkinen K, Kalkkinen N, Kaariainen L. Semliki Forest virus- specific non-structural protein nsP3 is a phosphoprotein. J Gen Virol 1988 September;69 ( Pt 9):2165-78. (88) Vihinen H, Ahola T, Tuittila M, Merits A, Kaariainen L. Elimination of phosphorylation sites of Semliki Forest virus replicase protein nsP3. J Biol Chem 2001 February 23;276(8):5745-52. (89) Frolova E, Gorchakov R, Garmashova N, Atasheva S, Vergara LA, Frolov I. Formation of nsP3-specific protein complexes during Sindbis virus replication. J Virol 2006 April;80(8):4122-34. (90) Galbraith SE, Sheahan BJ, Atkins GJ. Deletions in the hypervariable domain of the nsP3 gene attenuate Semliki Forest virus virulence. J Gen Virol 2006 April;87(Pt 4):937-47. (91) Suthar MS, Shabman R, Madric K, Lambeth C, Heise MT. Identification of adult mouse neurovirulence determinants of the Sindbis virus strain AR86. J Virol 2005 April;79(7):4219-28. (92) Acheson NH, Tamm I. Replication of Semliki Forest virus: an electron microscopic study. Virology 1967 May;32(1):128-43. (93) Grimley PM, Berezesky IK, Friedman RM. Cytoplasmic structures associated with an arbovirus infection: loci of viral ribonucleic acid synthesis. J Virol 1968 November;2(11):1326-38. (94) Peranen J, Kaariainen L. Biogenesis of type I cytopathic vacuoles in Semliki Forest virus-infected BHK cells. J Virol 1991 March;65(3):1623-7. (95) Salonen A, Vasiljeva L, Merits A, Magden J, Jokitalo E, Kaariainen L. Properly folded nonstructural polyprotein directs the semliki forest virus replication complex to the endosomal compartment. J Virol 2003 February;77(3):1691-702.

(96) Besterman JM, Low RB. Endocytosis: a review of mechanisms and plasma membrane dynamics. Biochem J 1983 January 15;210(1):1-13. (97) Doherty GJ, McMahon HT. Mechanisms of endocytosis. Annu Rev Biochem 2009;78:857-902. (98) Mercer J, Schelhaas M, Helenius A. Virus Entry by Endocytosis. Annu Rev Biochem 2010 March 2. (99) Marsh M, Helenius A. Virus entry: open sesame. Cell 2006 February 24;124(4):729-40.

(100) Coscoy L, Sanchez DJ, Ganem D. A novel class of herpesvirus-encoded membrane-bound E3 ubiquitin ligases regulates endocytosis of proteins involved in immune recognition. J Cell Biol 2001 December 24;155(7):1265- 73. (101) Coscoy L, Ganem D. Kaposi's sarcoma-associated herpesvirus encodes two proteins that block cell surface display of MHC class I chains by enhancing their endocytosis. Proc Natl Acad Sci U S A 2000 July 5;97(14):8051-6. (102) Conner SD, Schmid SL. Regulated portals of entry into the cell. Nature 2003 March 6;422(6927):37-44. (103) Helenius A, Kartenbeck J, Simons K, Fries E. On the entry of Semliki forest virus into BHK-21 cells. J Cell Biol 1980 February;84(2):404-20. (104) Gaidarov I, Santini F, Warren RA, Keen JH. Spatial control of coated-pit dynamics in living cells. Nat Cell Biol 1999 May;1(1):1-7. (105) Kirchhausen T. Clathrin. Annu Rev Biochem 2000;69:699-727. (106) Nossal R. Energetics of clathrin basket assembly. Traffic 2001 February;2(2):138-47. (107) Kirchhausen T. Clathrin. Annu Rev Biochem 2000;69:699-727. (108) Mettlen M, Pucadyil T, Ramachandran R, Schmid SL. Dissecting dynamin's role in clathrin-mediated endocytosis. Biochem Soc Trans 2009 October;37(Pt 5):1022-6. (109) Bashkirov PV, Akimov SA, Evseev AI, Schmid SL, Zimmerberg J, Frolov VA. GTPase cycle of dynamin is coupled to membrane squeeze and release, leading to spontaneous fission. Cell 2008 December 26;135(7):1276-86. (110) Massol RH, Boll W, Griffin AM, Kirchhausen T. A burst of auxilin recruitment determines the onset of clathrin-coated vesicle uncoating. Proc Natl Acad Sci U S A 2006 July 5;103(27):10265-70. (111) Saffarian S, Cocucci E, Kirchhausen T. Distinct dynamics of endocytic clathrin-coated pits and coated plaques. PLoS Biol 2009 September;7(9):e1000191.

(112) Hommelgaard AM, Roepstorff K, Vilhardt F, Torgersen ML, Sandvig K, van Deurs B. Caveolae: stable membrane domains with a potential for internalization. Traffic 2005 September;6(9):720-4. (113) Thomsen P, Roepstorff K, Stahlhut M, van Deurs B. Caveolae are highly immobile plasma membrane microdomains, which are not involved in constitutive endocytic trafficking. Mol Biol Cell 2002 January;13(1):238-50. (114) Hill MM, Bastiani M, Luetterforst R, Kirkham M, Kirkham A, Nixon SJ et al. PTRF-Cavin, a conserved cytoplasmic protein required for caveola formation and function. Cell 2008 January 11;132(1):113-24.

(115) Pelkmans L, Zerial M. Kinase-regulated quantal assemblies and kiss-and- run recycling of caveolae. Nature 2005 July 7;436(7047):128-33. (116) Richter T, Floetenmeyer M, Ferguson C, Galea J, Goh J, Lindsay MR et al. High-resolution 3D quantitative analysis of caveolar ultrastructure and caveola-cytoskeleton interactions. Traffic 2008 June;9(6):893-909. (117) Rothberg KG, Heuser JE, Donzell WC, Ying YS, Glenney JR, Anderson RG. Caveolin, a protein component of caveolae membrane coats. Cell 1992 February 21;68(4):673-82. (118) Kiss AL, Botos E. Endocytosis via caveolae: alternative pathway with distinct cellular compartiments to avoid lysosomal degradation? J Cell Mol Med 2009 March 27. (119) Kiss AL, Botos E. Ocadaic acid retains caveolae in multicaveolar clusters. Pathol Oncol Res 2009 September;15(3):479-86. (120) Parton RG, Simons K. The multiple faces of caveolae. Nat Rev Mol Cell Biol 2007 March;8(3):185-94. (121) Tagawa A, Mezzacasa A, Hayer A, Longatti A, Pelkmans L, Helenius A. Assembly and trafficking of caveolar domains in the cell: caveolae as stable, cargo-triggered, vesicular transporters. J Cell Biol 2005 August 29;170(5):769-79. (122) Pelkmans L, Kartenbeck J, Helenius A. Caveolar endocytosis of simian virus 40 reveals a new two-step vesicular-transport pathway to the ER. Nat Cell Biol 2001 May;3(5):473-83. (123) Pelkmans L, Puntener D, Helenius A. Local actin polymerization and dynamin recruitment in SV40-induced internalization of caveolae. Science 2002 April 19;296(5567):535-9. (124) Pelkmans L, Fava E, Grabner H, Hannus M, Habermann B, Krausz E et al. Genome-wide analysis of human kinases in clathrin- and caveolae/raft- mediated endocytosis. Nature 2005 July 7;436(7047):78-86. (125) Pelkmans L, Helenius A. Endocytosis via caveolae. Traffic 2002 May;3(5):311-20.

(126) Hansen CG, Bright NA, Howard G, Nichols BJ. SDPR induces membrane curvature and functions in the formation of caveolae. Nat Cell Biol 2009 July;11(7):807-14. (127) Hayer A, Stoeber M, Bissig C, Helenius A. Biogenesis of caveolae: stepwise assembly of large caveolin and cavin complexes. Traffic 2010 March;11(3):361-82. (128) Hansen CG, Nichols BJ. Molecular mechanisms of clathrin-independent endocytosis. J Cell Sci 2009 June 1;122(Pt 11):1713-21. (129) Glebov OO, Bright NA, Nichols BJ. Flotillin-1 defines a clathrin- independent endocytic pathway in mammalian cells. Nat Cell Biol 2006 January;8(1):46-54. (130) Frick M, Bright NA, Riento K, Bray A, Merrified C, Nichols BJ. Coassembly of flotillins induces formation of membrane microdomains, membrane curvature, and vesicle budding. Curr Biol 2007 July 3;17(13):1151-6. (131) Riento K, Frick M, Schafer I, Nichols BJ. Endocytosis of flotillin-1 and flotillin-2 is regulated by Fyn kinase. J Cell Sci 2009 April 1;122(Pt 7):912-8. (132) Swanson JA. Shaping cups into phagosomes and macropinosomes. Nat Rev Mol Cell Biol 2008 August;9(8):639-49. (133) Mercer J, Helenius A. Virus entry by macropinocytosis. Nat Cell Biol 2009 May;11(5):510-20. (134) Mercer J, Helenius A. Vaccinia virus uses macropinocytosis and apoptotic mimicry to enter host cells. Science 2008 April 25;320(5875):531-5. (135) Grassart A, Dujeancourt A, Lazarow PB, Dautry-Varsat A, Sauvonnet N. Clathrin-independent endocytosis used by the IL-2 receptor is regulated by Rac1, Pak1 and Pak2. EMBO Rep 2008 April;9(4):356-62. (136) Lamaze C, Dujeancourt A, Baba T, Lo CG, Benmerah A, Dautry-Varsat A. Interleukin 2 receptors and detergent-resistant membrane domains define a clathrin-independent endocytic pathway. Mol Cell 2001 March;7(3):661-71. (137) Sauvonnet N, Dujeancourt A, Dautry-Varsat A. Cortactin and dynamin are required for the clathrin-independent endocytosis of gammac cytokine receptor. J Cell Biol 2005 January 3;168(1):155-63. (138) Gesbert F, Sauvonnet N, Dautry-Varsat A. Clathrin-lndependent endocytosis and signalling of interleukin 2 receptors IL-2R endocytosis and signalling. Curr Top Microbiol Immunol 2004;286:119-48. (139) Sabharanjak S, Sharma P, Parton RG, Mayor S. GPI-anchored proteins are delivered to recycling endosomes via a distinct cdc42-regulated, clathrin- independent pinocytic pathway. Dev Cell 2002 April;2(4):411-23. (140) Mayor S, Pagano RE. Pathways of clathrin-independent endocytosis. Nat Rev Mol Cell Biol 2007 August;8(8):603-12.

(141) Doherty GJ, Lundmark R. GRAF1-dependent endocytosis. Biochem Soc Trans 2009 October;37(Pt 5):1061-5. (142) Lundmark R, Doherty GJ, Howes MT, Cortese K, Vallis Y, Parton RG et al. The GTPase-activating protein GRAF1 regulates the CLIC/GEEC endocytic pathway. Curr Biol 2008 November 25;18(22):1802-8. (143) Ewers H, Romer W, Smith AE, Bacia K, Dmitrieff S, Chai W et al. GM1 structure determines SV40-induced membrane invagination and infection. Nat Cell Biol 2010 January;12(1):11-8. (144) Studer D., Humbel B., Chiquet M. Electron microscopy of high pressure frozen samples: bridging the gap between cellular ultrastructure and atomic resolution. Histochemistry and Cell Biology 2008 September 16;130(5). (145) Robinson C, Sali A., Baumeister W. The molecular sociology of the cell. Nature 2007 December 13;450. (146) Koning RI, Koster AJ. Cryo-electron tomography in biology and medicine. Ann Anat 2009 November;191(5):427-45. (147) Plitzko JM, Rigort A., Leis A. Correlative cryo-light microscopy and cryo- electron tomography: from cellular terrtories to molecular landscapes. Curr Opin Biotechnol 2009 April 1;20. (148) Keranen S, Kaariainen L. Functional defects of RNA-negative temperature- sensitive mutants of Sindbis and Semliki Forest viruses. J Virol 1979 October;32(1):19-29. (149) Lulla V, Merits A, Sarin P, Kaariainen L, Keranen S, Ahola T. Identification of mutations causing temperature-sensitive defects in Semliki Forest virus RNA synthesis. J Virol 2006 March;80(6):3108-11. (150) Bartelma G, Padmanabhan R. Expression, purification, and characterization of the RNA 5'-triphosphatase activity of dengue virus type 2 nonstructural protein 3. Virology 2002 July 20;299(1):122-32. (151) Benarroch D, Selisko B, Locatelli GA, Maga G, Romette JL, Canard B. The RNA helicase, nucleotide 5'-triphosphatase, and RNA 5'-triphosphatase activities of Dengue virus protein NS3 are Mg2+-dependent and require a functional Walker B motif in the helicase catalytic core. Virology 2004 October 25;328(2):208-18. (152) Ivanov KA, Thiel V, Dobbe JC, van der MY, Snijder EJ, Ziebuhr J. Multiple enzymatic activities associated with severe acute respiratory syndrome coronavirus helicase. J Virol 2004 June;78(11):5619-32. (153) Mertens PP, Diprose J. The bluetongue virus core: a nano-scale transcription machine. Virus Res 2004 April;101(1):29-43. (154) Grimes JM, Jakana J, Ghosh M, Basak AK, Roy P, Chiu W et al. An atomic model of the outer layer of the bluetongue virus core derived from X-ray crystallography and electron cryomicroscopy. Structure 1997 July 15;5(7):885-93.

(155) Grimes JM, Burroughs JN, Gouet P, Diprose JM, Malby R, Zientara S et al. The atomic structure of the bluetongue virus core. Nature 1998 October 1;395(6701):470-8. (156) Gouet P, Diprose JM, Grimes JM, Malby R, Burroughs JN, Zientara S et al. The highly ordered double-stranded RNA genome of bluetongue virus revealed by crystallography. Cell 1999 May 14;97(4):481-90. (157) Diprose JM, Burroughs JN, Sutton GC, Goldsmith A, Gouet P, Malby R et al. Translocation portals for the substrates and products of a viral transcription complex: the bluetongue virus core. EMBO J 2001 December 17;20(24):7229-39. (158) Bieniasz PD. The cell biology of HIV-1 virion genesis. Cell Host Microbe 2009 June 18;5(6):550-8. (159) Zechmann B, Muller M, Zellnig G. Membrane associated qualitative differences in cell ultrastructure of chemically and high pressure cryofixed plant cells. J Struct Biol 2007 June;158(3):370-7. (160) Zhou ZH. Towards atomic resolution structural determination by single- particle cryo-electron microscopy. Curr Opin Struct Biol 2008 April;18(2):218-28. (161) Yu X, Jin L, Zhou ZH. 3.88 A structure of cytoplasmic polyhedrosis virus by cryo-electron microscopy. Nature 2008 May 15;453(7193):415-9. (162) Mancini EJ, Clarke M, Gowen BE, Rutten T, Fuller SD. Cryo-electron microscopy reveals the functional organization of an enveloped virus, Semliki Forest virus. Mol Cell 2000 February;5(2):255-66. (163) Subramaniam S, Bartesaghi A, Liu J, Bennett AE, Sougrat R. Electron tomography of viruses. Curr Opin Struct Biol 2007 October;17(5):596-602. (164) Grunewald K, Cyrklaff M. Structure of complex viruses and virus-infected cells by electron cryo tomography. Curr Opin Microbiol 2006 August;9(4):437-42. (165) Zanetti G, Briggs JA, Grunewald K, Sattentau QJ, Fuller SD. Cryo-electron tomographic structure of an immunodeficiency virus envelope complex in situ. PLoS Pathog 2006 August;2(8):e83. (166) Briggs JA, Grunewald K, Glass B, Forster F, Krausslich HG, Fuller SD. The mechanism of HIV-1 core assembly: insights from three-dimensional reconstructions of authentic virions. Structure 2006 January;14(1):15-20. (167) Carlson LA, Briggs JA, Glass B, Riches JD, Simon MN, Johnson MC et al. Three-dimensional analysis of budding sites and released virus suggests a revised model for HIV-1 morphogenesis. Cell Host Microbe 2008 December 11;4(6):592-9. (168) Bartesaghi A, Subramaniam S. Membrane protein structure determination using cryo-electron tomography and 3D image averaging. Curr Opin Struct Biol 2009 August;19(4):402-7.

(169) Huiskonen JT, Hepojoki J, Laurinmaki P, Vaheri A, Lankinen H, Butcher SJ et al. Electron cryotomography of Tula hantavirus suggests a unique assembly paradigm for enveloped viruses. J Virol 2010 May;84(10):4889-97. (170) Beck M, Forster F, Ecke M, Plitzko JM, Melchior F, Gerisch G et al. Nuclear pore complex structure and dynamics revealed by cryoelectron tomography. Science 2004 November 19;306(5700):1387-90. (171) Beck M, Lucic V, Forster F, Baumeister W, Medalia O. Snapshots of nuclear pore complexes in action captured by cryo-electron tomography. Nature 2007 October 4;449(7162):611-5.

(172) Forster F, Medalia O, Zauberman N, Baumeister W, Fass D. Retrovirus envelope protein complex structure in situ studied by cryo-electron tomography. Proc Natl Acad Sci U S A 2005 March 29;102(13):4729-34. (173) Medalia O, Beck M, Ecke M, Weber I, Neujahr R, Baumeister W et al. Organization of actin networks in intact filopodia. Curr Biol 2007 January 9;17(1):79-84. (174) Garvalov BK, Zuber B, Bouchet-Marquis C, Kudryashev M, Gruska M, Beck M et al. Luminal particles within cellular microtubules. J Cell Biol 2006 September 11;174(6):759-65. (175) Koning RI, Zovko S, Barcena M, Oostergetel GT, Koerten HK, Galjart N et al. Cryo electron tomography of vitrified fibroblasts: microtubule plus ends in situ. J Struct Biol 2008 March;161(3):459-68. (176) Helfrich W. Elastic properties of lipid bilayers: theory and possible experiments. Z Naturforsch C 1973 November;28(11):693-703. (177) Helfrich W. Out-of-plane fluctuations of lipid bilayers. Z Naturforsch C 1975 November;30(6):841-2. (178) Shibata Y, Hu J, Kozlov MM, Rapoport TA. Mechanisms shaping the membranes of cellular organelles. Annu Rev Cell Dev Biol 2009;25:329-54. (179) Huttner WB, Zimmerberg J. Implications of lipid microdomains for membrane curvature, budding and fission. Curr Opin Cell Biol 2001 August;13(4):478-84. (180) van Meer G, Voelker DR, Feigenson GW. Membrane lipids: where they are and how they behave. Nat Rev Mol Cell Biol 2008 February;9(2):112-24. (181) Zimmerberg J, Kozlov MM. How proteins produce cellular membrane curvature. Nat Rev Mol Cell Biol 2006 January;7(1):9-19. (182) Ahlquist P. Parallels among positive-strand RNA viruses, reverse- transcribing viruses and double-stranded RNA viruses. Nat Rev Microbiol 2006 May;4(5):371-82. (183) Gottwein E, Jager S, Habermann A, Krausslich HG. Cumulative mutations of ubiquitin acceptor sites in human immunodeficiency virus type 1 gag cause a late budding defect. J Virol 2006 July;80(13):6267-75.

(184) Miller DJ, Schwartz MD, Dye BT, Ahlquist P. Engineered retargeting of viral RNA replication complexes to an alternative intracellular membrane. J Virol 2003 November;77(22):12193-202. (185) Teterina NL, Gorbalenya AE, Egger D, Bienz K, Ehrenfeld E. Poliovirus 2C protein determinants of membrane binding and rearrangements in mammalian cells. J Virol 1997 December;71(12):8962-72. (186) Egger D, Wolk B, Gosert R, Bianchi L, Blum HE, Moradpour D et al. Expression of hepatitis C virus proteins induces distinct membrane alterations including a candidate viral replication complex. J Virol 2002 June;76(12):5974-84. (187) Roy P., Noad R. Bluetongue virus assembly and morphogenesis. Curr Top Microbiol Immunol 2006;309. (188) Hanson PI, Roth R, Lin Y, Heuser JE. Plasma membrane deformation by circular arrays of ESCRT-III protein filaments. J Cell Biol 2008 January 28;180(2):389-402. (189) Hurley JH. ESCRT complexes and the biogenesis of multivesicular bodies. Curr Opin Cell Biol 2008 February;20(1):4-11. (190) Wollert T, Wunder C, Lippincott-Schwartz J, Hurley JH. Membrane scission by the ESCRT-III complex. Nature 2009 March 12;458(7235):172-7. (191) Wollert T, Hurley JH. Molecular mechanism of multivesicular body biogenesis by ESCRT complexes. Nature 2010 April 8;464(7290):864-9. (192) Kieffer C, Skalicky JJ, Morita E, De D, I, Ward DM, Kaplan J et al. Two distinct modes of ESCRT-III recognition are required for VPS4 functions in lysosomal protein targeting and HIV-1 budding. Dev Cell 2008 July;15(1):62-73. (193) Huiskonen JT, de Haas F, Bubeck D, Bamford DH, Fuller SD, Butcher SJ. Structure of the bacteriophage phi6 nucleocapsid suggests a mechanism for sequential RNA packaging. Structure 2006 June;14(6):1039-48. (194) Kainov DE, Lisal J, Bamford DH, Tuma R. Packaging motor from double- stranded RNA bacteriophage phi12 acts as an obligatory passive conduit during transcription. Nucleic Acids Res 2004;32(12):3515-21. (195) Murk JL, Humbel BM, Ziese U, Griffith JM, Posthuma G, Slot JW et al. Endosomal compartmentalization in three dimensions: implications for membrane fusion. Proc Natl Acad Sci U S A 2003 November 11;100(23):13332-7. (196) Shirako Y, Strauss JH. Regulation of Sindbis virus RNA replication: uncleaved P123 and nsP4 function in minus-strand RNA synthesis, whereas cleaved products from P123 are required for efficient plus-strand RNA synthesis. J Virol 1994 March;68(3):1874-85. (197) Golubtsov A. Mechanisms for alphavirus nonstructural polyprotein processing. Doctoral dissertation, ISBN:978-952-10-4492-2 . 30-1-2008.

University of Helsinki, Faculty of Biosciences, Department of Biological and Environmental Sciences, and Institute of Biotechnogy.

(198) Vasiljeva L, Merits A, Golubtsov A, Sizemskaja V, Kaariainen L, Ahola T. Regulation of the sequential processing of Semliki Forest virus replicase polyprotein. J Biol Chem 2003 October 24;278(43):41636-45. (199) Sawicki DL, Kaariainen L, Lambek C, Gomatos PJ. Mechanism for control of synthesis of Semliki Forest virus 26S and 42s RNA. J Virol 1978 January;25(1):19-27. (200) Luzio JP, Pryor PR, Bright NA. Lysosomes: fusion and function. Nature Reviews Molecular Cell Biology 2007 August;8(8):622-32. (201) D'Souza-Schorey C, Chavrier P. ARF proteins: roles in membrane traffic and beyond. Nat Rev Mol Cell Biol 2006 May;7(5):347-58. (202) Belov GA, Feng Q, Nikovics K, Jackson CL, Ehrenfeld E. A critical role of a cellular membrane traffic protein in poliovirus RNA replication. PLoS Pathog 2008 November;4(11):e1000216. (203) Lanke KH, van der Schaar HM, Belov GA, Feng Q, Duijsings D, Jackson CL et al. GBF1, a guanine nucleotide exchange factor for Arf, is crucial for coxsackievirus B3 RNA replication. J Virol 2009 November;83(22):11940-9. (204) Lippincott-Schwartz J, Yuan L, Tipper C, Amherdt M, Orci L, Klausner RD. Brefeldin A's effects on endosomes, lysosomes, and the TGN suggest a general mechanism for regulating organelle structure and membrane traffic. Cell 1991 November 1;67(3):601-16. (205) Pierini R, Cottam E, Roberts R, Wileman T. Modulation of membrane traffic between endoplasmic reticulum, ERGIC and Golgi to generate compartments for the replication of bacteria and viruses. Semin Cell Dev Biol 2009 September;20(7):828-33. (206) Schlegel A, Giddings TH, Jr., Ladinsky MS, Kirkegaard K. Cellular origin and ultrastructure of membranes induced during poliovirus infection. J Virol 1996 October;70(10):6576-88. (207) Hamamoto I, Nishimura Y, Okamoto T, Aizaki H, Liu M, Mori Y et al. Human VAP-B is involved in hepatitis C virus replication through interaction with NS5A and NS5B. J Virol 2005 November;79(21):13473-82. (208) Hagemeijer MC, Verheije MH, Ulasli M, Shaltiel IA, de Vries LA, Reggiori F et al. Dynamics of Coronavirus Replication-Transcription Complexes. J Virol 2009 December 9. (209) Egger D, Bienz K. Intracellular location and translocation of silent and active poliovirus replication complexes. J Gen Virol 2005 March;86(Pt 3):707-18. (210) Wölk B, Büchele B, Moradpour D, Rice CM. A dynamic view of hepatitis C virus replication complexes. J Virol 2008 November;82(21):10519-31.

(211) Doedens J, Maynell LA, Klymkowsky MW, Kirkegaard K. Secretory pathway function, but not cytoskeletal integrity, is required in poliovirus infection. Arch Virol Suppl 1994;9:159-72. (212) Maynell LA, Kirkegaard K, Klymkowsky MW. Inhibition of poliovirus RNA synthesis by brefeldin A. J Virol 1992 April;66(4):1985-94. (213) Stertz S, Reichelt M, Spiegel M, Kuri T, Martinez-Sobrido L, Garcia-Sastre A et al. The intracellular sites of early replication and budding of SARS- coronavirus. Virology 2007 May 10;361(2):304-15. (214) Lajoie P, Nabi IR. Regulation of raft-dependent endocytosis. J Cell Mol Med 2007 July;11(4):644-53. (215) Le PU, Guay G, Altschuler Y, Nabi IR. Caveolin-1 is a negative regulator of caveolae-mediated endocytosis to the endoplasmic reticulum. J Biol Chem 2002 February 1;277(5):3371-9. (216) Kirkham M, Fujita A, Chadda R, Nixon SJ, Kurzchalia TV, Sharma DK et al. Ultrastructural identification of uncoated caveolin-independent early endocytic vehicles. J Cell Biol 2005 January 31;168(3):465-76. (217) Kirkham M, Nixon SJ, Howes MT, Abi-Rached L, Wakeham DE, Hanzal- Bayer M et al. Evolutionary analysis and molecular dissection of caveola biogenesis. J Cell Sci 2008 June 15;121(Pt 12):2075-86. (218) Zhao YY, Liu Y, Stan RV, Fan L, Gu Y, Dalton N et al. Defects in caveolin-1 cause dilated cardiomyopathy and pulmonary hypertension in knockout mice. Proc Natl Acad Sci U S A 2002 August 20;99(17):11375-80. (219) Lajoie P, Kojic LD, Nim S, Li L, Dennis JW, Nabi IR. Caveolin-1 regulation of dynamin-dependent, raft-mediated endocytosis of cholera toxin-B subunit occurs independently of caveolae. J Cell Mol Med 2009 September;13(9B):3218-25. (220) Rodal SK, Skretting G, Garred O, Vilhardt F, van Deurs B, Sandvig K. Extraction of cholesterol with methyl-beta-cyclodextrin perturbs formation of clathrin-coated endocytic vesicles. Mol Biol Cell 1999 April;10(4):961-74. (221) Zidovetzki R, Levitan I. Use of cyclodextrins to manipulate plasma membrane cholesterol content: evidence, misconceptions and control strategies. Biochim Biophys Acta 2007 June;1768(6):1311-24. (222) Nichols B. Caveosomes and endocytosis of lipid rafts. J Cell Sci 2003 December 1;116(Pt 23):4707-14. (223) Damm EM, Pelkmans L, Kartenbeck J, Mezzacasa A, Kurzchalia T, Helenius A. Clathrin- and caveolin-1-independent endocytosis: entry of simian virus 40 into cells devoid of caveolae. J Cell Biol 2005 January 31;168(3):477-88. (224) Lindmo K, Stenmark H. Regulation of membrane traffic by phosphoinositide 3-kinases. J Cell Sci 2006 February 15;119(Pt 4):605-14.

(225) Raynaud FI, Eccles S, Clarke PA, Hayes A, Nutley B, Alix S et al. Pharmacologic characterization of a potent inhibitor of class I phosphatidylinositide 3-kinases. Cancer Res 2007 June 15;67(12):5840-50. (226) Vihinen H, Ahola T, Tuittila M, Merits A, Kaariainen L. Elimination of phosphorylation sites of Semliki Forest virus replicase protein nsP3. J Biol Chem 2001 February 23;276(8):5745-52. (227) Swann SA, Williams M, Story CM, Bobbitt KR, Fleis R, Collins KL. HIV-1 Nef blocks transport of MHC class I molecules to the cell surface via a PI 3- kinase-dependent pathway. Virology 2001 April 10;282(2):267-77.

(228) Roeth JF, Collins KL. Human immunodeficiency virus type 1 Nef: adapting to intracellular trafficking pathways. Microbiol Mol Biol Rev 2006 June;70(2):548-63. (229) Kiss AL, Botos E. Ocadaic Acid Retains Caveolae in Multicaveolar Clusters. Pathol Oncol Res 2008 December 10. (230) Kiss AL, Botos E, Turi A, Mullner N. Ocadaic acid treatment causes tyrosine phosphorylation of caveolin-2 and induces internalization of caveolae in rat peritoneal macrophages. Micron 2004;35(8):707-15. (231) Workman P, Clarke PA, Raynaud FI, van Montfort RL. Drugging the PI3 kinome: from chemical tools to drugs in the clinic. Cancer Res 2010 March 15;70(6):2146-57.