Structure-function analysis of the Sendai virus Matrix protein

Thesis

Reference

Structure-function analysis of the Sendai virus Matrix protein

MIAZZA, Vincent

Abstract

Dans cette étude, notre attention s'est concentrée sur la protéine M du virus de Sendai ainsi que sur le processus de bourgeonnement. Dans la première partie de cette étude, nous avons tenté d'identifier une région minimale nécessaire et suffisante pour amener M à la membrane plasmique. Plusieurs mutants et délétions de M ont été effetués. Nous avons également étudié la relation entre les protéines M et C du virus de Sendai en utilisant les différents mutants de M obtenus, ainsi que plusieurs versions de C. Dans la deuxième partie de cette étude, nous avons voulu identifier de nouveaux partenaires cellulaires de la protéine M. Nous avons dès lors choisi d'étudier 2 protéines parmi les nombreux candidats identifiés:

"charged multivescular body protein 5 (CHMP5)" et "cytoplasmic actin 1 (β-actin)".

MIAZZA, Vincent. Structure-function analysis of the Sendai virus Matrix protein. Thèse de doctorat : Univ. Genève, 2009, no. Sc. 4063

URN : urn:nbn:ch:unige-22422

DOI : 10.13097/archive-ouverte/unige:2242

Available at:

http://archive-ouverte.unige.ch/unige:2242

Disclaimer: layout of this document may differ from the published version.

UNIVERSITE DE GENEVE

Département de Biologie Cellulaire FACULTE DES SCIENCES

Professeur Angela Krämer

Département de Microbiologie et Médecine FACULTE DE MEDECINE

Moléculaire Professeur Laurent Roux

Structure-function analysis of the Sendai virus Matrix protein

THESE

Présentée à la Faculté des Sciences de l’Université de Genève pour obtenir le grade de Docteur ès science, mention biologie

par

Vincent MIAZZA de

Chêne-Bourg (GE)

Thèse n° 4063 Genève

2009

Remerciements

Je voudrais tout d’abord remercier Laurent Roux pour m’avoir accueilli chaleureusement au sein de son laboratoire durant ces quelques années et de m’avoir ainsi permis d’accomplir ce travail.

Mes remerciements vont également aux différents collègues de paillasse que j’ai eu la chance de côtoyer durant ces années, tant pour les échanges scientifiques que pour les bons moments de complicité. Un grand merci à Geneviève Mottet-Osman, la fée du logis comme aime le préciser Laurent, sans qui bien peu des grandes réalisations auraient été possibles. Merci également à Samuel Cordey, l’ancien doctorant devenu grand, qui m’a accompagné durant mes premiers pas au CMU ainsi qu’aux « frenchies postdocs» Anne- Sophie Gosselin-Grenet et Carole Bampi pour quelques précieux conseils et beaucoup de bons moments passés ensemble. Enfin, merci aux travailleuses de l’ombre Alexandra Chassot et Anne-Lyse Kahr pour les nombreux services rendus.

Finalement, merci à toutes les personnes qui m’ont soutenu d’une manière ou d’une autre au cours de ce long périple.

TABLES OF CONTENTS

RESUME EN FRANÇAIS 5 ENGLISH SUMMARY 9

GENERAL INTRODUCTION 12

Classification 13

Sendai virus genome organization 15 Overview of a Paramyxoviridae replication cycle 16 Virion assembly and budding 18 The Nucleocapsid assembly 18

The Matrix protein 18

The Glycoproteins HN and F 24

The C protein 28

Protein-protein interactions 31 The M protein and the viral components 31 The M protein and the cellular partners 36 A platform of assembly: the lipid rafts 38 Hijacking of the multivesicular bodies machinery (MVB) 42 The use of the cellular cytoskeleton 46 Reverse genetic system: recombinant Sendai viruses and defective interfering particles 48 Defective interfering particles: 51

GOALS OF THE THESIS 52 MATERIAL AND METHODS 53

RESULTS 63

Structure-function analysis of SeV Matrix protein 63

Introduction 63

Methodology 65

GFP-HA-M fusion proteins: the proof of principle 67 GFP-HA-M mutant’s analysis by SeV mixed stocks infections 70 HA-M mutant analysis by SeV mixed stocks infections 73 HA-Mwt and HA-M30 localization by transient expression 74

HA-M mutant localization by transient expression 76 Relationship between C and M: characterization of the C protein 79 C1-23TomY and HA-M proteins co-expressions 81 C1-23Tom and HA-M proteins co-expressions 85 Φ5ATomY and HA-M proteins co-expressions 88 Localizations of GFP-C and M during SeV infection 91

Conclusions 92

Identification of new cellular partners of the SeV matrix protein 95

Introduction 95

Methodology 96

TAP-M proteins expression 97 Tandem Affinity Purification steps 100 Identification of new cellular partners 104 Yeast Two-hybrid screening 107 Relevance of Charged multivesicular body protein 5 (CHMP5) in Sendai virus infection 110

Conclusions 111

Remodeling of cytoplasmic actins during Paramyxoviruses infections: a process induced by the matrix protein to optimize virus production? 114

Introduction 114

Publication (submitted to Journal of Virology) 116

GENERAL DISCUSSION 141 REFERENCE LIST 150

Résumé en français

Le virus de Sendai appartient à la famille des Paramyxoviridae. Il est composé d’une nucléocapside à morphologie hélicoïdale qui contient un ARN simple brin de polarité négative de 15kb. Le génome code pour six protéines structurales N, P, M, F, HN et L. Le gène P a la particularité de coder également pour des protéines non-structurales, à savoir V et un groupe de protéines appelées C.

Les Paramyxovirus se répliquent entièrement dans le cytoplasme, puis les virions sont assemblés à la membrane plasmique d’où ils acquièrent leur envelope virale et bourgeonnent. Un mécanisme appelé en anglais « pinching-off » (pincement) permet le relâchement des particules virales. Le processus de bourgeonnement requiert une grande coordination des différents composants viraux impliqués, ainsi que probablement de quelques facteurs cellulaires additionnels. Or, cette coordination est en grande partie dirigée par la protéine de matrice (M). En effet, elle contacte tous les composants viraux, à savoir la nucléocapside ainsi que les deux gylcoprotéines F et HN.

Dans cette étude, notre attention c’est concentrée sur la protéine M du virus de Sendai ainsi que sur le processus de bourgeonnement. La protéine M comporte 348 acides aminés (~38kDa), elle est basique (pI 9.68) et modérément hydrophobique.

Selon les prédictions, peu de structures secondaires sont attendues, à savoir 11% d’hélice α, 19% de feuillet β et 70% de pelotes statistiques (Giuffre, Tovell et al., 1982). Malgré ces prédictions, la structure de M reste presque totalement inconnue. Une de ces principales caractéristiques est liée à sa capacité de multimérisation (Büechi & Bächi, 1982) et à son accumulation juste en dessous du double feuillet lipidique. Au niveau intracellulaire, M a été observée à de nombreuse reprise associée aux membranes cellulaires (Stricker, Mottet et al., 1994;Sanderson, McQueen et al., 1993).

Toutefois, M reste une protéine relativement mystérieuse dans la mesure où les mécanismes lui permettant de s’accumuler à la membrane plasmique et ceux lui permettant de dirigé le bourgeonnement restent pour l’heure encore largement inconnus.

Dans la première partie de cette étude, nous avons tenté d’identifier une région minimale nécessaire et suffisante pour amener M à la membrane plasmique. Plusieurs mutants et délétions de M ont été éffectués. Ces différentes protéines M ont été fusionnées soit à HA, soit à GFP afin de facilement les suivre. Ces différentes versions de M ont été exprimées dans un contexte d’infection avec le virus de Sendai ou par simple expression transitoire. Lors de cette étude nous avons clairement pu constater que même des changements minimes au niveau de sa structure primaire affectaient grandement sa bonne localisation, et spécialement sa capacité à atteindre la membrane plasmique.

Malgrés nos efforts, nous n’avons pas pu définir une région minimale et nécessaire responsable de son acheminement à la membrane plasmique. Toutefois, l’analyse des différents mutants nous a permi de conclure que si une telle région devait exister, elle se situerait vraisemblablement dans les 2/3 N-terminal de la protéine. Cette étude nous a également permis la découverte de deux mutants montrant des phénotypes partiels, à savoir HA-M^YPNV_4A et HA-M^IRKL_4A. Ces deux mutants, où deux motifs (YPNV et IRKL) situés dans la région C-terminale de la protéine avaient été remplacés par des alanines, gardent une certaine capacité à s’accumuler à la membrane plasmique.

Par la suite nous avons étudié la relation entre les protéines M et C en utilisant les différents mutants de M obtenus, ainsi que plusieurs versions de C. En effet, C pourrait détenir un rôle dans l’assemblage et le bourgeonnement du virus de Sendai comme il avait été proposé récemment (Hasan, Kato et al., 2000). Ceci nous avait fortement interpellés et nous avons voulu voir si une éventuelle coopération entre les deux protéines pouvait avoir lieu. Nous avons montré que M et C sont en mesure d’atteindre la membrane plamisque de manière indépendante et qu’elles co-localisent à ces endroits.

Cette co-localisation a également pu être vérifiée dans le contexte d’une infection, ce qui pourrait parler en faveur d’une coopération entre les deux proteines.

Dans la deuxième partie de cette étude, nous avons voulu identifier de nouveaux partenaires cellulaires de la protéine M. Deux stratégies ont été choisie afin de découvrir des protéines intéragissant avec M : la première approches, appelée « Tandem Affinity Purification », consiste en une purification de M dans des conditions natives, alors que la

seconde approche, dite du « double-hybride » dans la levure, permet l’identification d’interactions protéine-protéine.

La combinaison de ces deux approches nous a permis d’identifier plusieurs partenaires potentiels de la protéine M. Nous avons dès lors choisi d’étudier 2 protéines parmi les nombreux candidats : charged multivesicular body protein 5 (CHMP5) et cytoplasmic actin 1 (β-actin).

CHMP5 est une protéine appartenant à une machinerie cellulaire appelée MVBs (Multivesicular bodies). Grâce au « double-hybride » accompli par nos collaborateurs (Pierre-Olivier Vidalain et son équipe), une interaction entre CHMP5 et M sauvage avait été identifiée. Cette interaction était en revanche perdue avec le mutant M^YPNV_4A or, le motif YPNV avait été proposé comme étant potentiellement un « domaine tardif » responsable du recrutement d’un membre de la machinerie cellulaire MVBs (Schmitt, Leser et al., 2005). Afin de vérifier le niveau d’implication de CHMP5 dans la multiplication du virus de Sendai, nous avons supprimé cette protéine en utilisant des petits ARN d’interference. Malgrés une suppression atteignant 90% aucun effet n’a été constaté ni sur la réplication virale, ni sur le bourgeonnement. Nous en avons donc conclu que CHMP5 était biologiquement non pertinente pour la multiplication virale du virus de Sendai.

Dans la troisième partie de cette étude, nous avons vérifié l’importance pour la multiplication du virus de Sendai de la protéine β-actine, qui avait été co-purifiée lors de la purification dite « Tandem Affinity Purification ». La β-actin est une des deux isoformes d’actine (avec la γ-actine) exprimée dans tous les tissus mammifères et essentielle pour la survie cellulaire (Karabinos, Schmidt et al., 2001). Grâce à des anticorps spécifiques dirigés contre ces deux isoformes d’actine obtenus suite à une collaboration avec Christine Chaponnier, nous avons pu suivre les modifications du cytosquelette d’actine après infection par le virus de Sendai.

Nous avons remarqué suite à l’infection par le virus de Sendai dans des cellules MDCK polarisées, que les deux isoformes d’actines et particulièrement la β-actine, étaient complètement re-localisées du côté apical aux endroits où le virus bourgeonne.

Nous avons montré par la suite que ce type de re-localisation était dépendant d’une

infection lytique car inexistente lors d’une infection persitente, où par ailleurs, M est très instable et peu présente. Ce type de re-localisation n’a pas pu être observé lors d’infection avec le virus de stomatite vésiculaire (vesicular stomatitis virus, VSV) ainsi qu’avec le virus parainfluenza humain de type 5 (human parainfluenza virus 5, HPIV5). La suppression individuelle ou combinée des isoformes d’actines par des petits ARN d’interference, a démontré une réelle importance des actines pour le bourgeonnement du virus. Plusieurs évidences parlent en faveur d’un rôle de la protéine M favorisant ce remodelage du cytosquelette d’actin. D’une part, M co-localise de manière extrêmement claire avec les deux isoformes aux lieux de re-localisation. D’autre part, M sauvage affecte grandement la morphologie cellulaire lorsqu’exprimée transitoirement, alors que tel n’est pas le cas pour deux mutants de M (M30 et M^IRKL_4A), ce qui indique une possible action de la protéine sur le cytosquelette. Néanmoins, l’interprétation de ce dernier résultat s’est révélée difficile à cause du changement morphologique des cellules.

Nous concluons que M est certainement nécessaire pour ce remodelage, mais que des facteurs additionnels présents lors d’une infection doivent être indispensables.

English summary

Sendai virus (SeV) belongs to the Paramyxoviridae family. Its nucleocapsid has a helical morphology and contains a negative single stranded RNA genome of 15 kb tightly associated to the viral nucleoprotein. The genome codes for six structural proteins N, P, M, F, HN and L. Additionally, the P gene codes for several non structural proteins known as V and a set of C proteins.

Paramyxoviruses replicate in the cytoplasm, and assemble at the plasma membrane from which they bud and aquire the viral envelope. A mechanism called pinching-off allows the release of the viral particles. The budding process is believed to require a high coordination of the different viral components and possibly of some host cellular factors. This coordination is mainly directed by the viral matrix protein (M), which is believed to contact all the viral components, namely the viral nucleocapsid, and both glycoproteins F and HN.

In the present study, we focused our attention on M and on viral budding. Sendai virus M protein contains 348 amino acids (~38kDa), is a basic protein (pI 9.68), and is moderately hydrophobic. The rest of the protein is predicted to include little secondary structure, namely 11% α-helix, 19% β-pleated sheet and 70% random coil (Giuffre, Tovell et al., 1982), but very little has been studied about the structure of M to date. M is known to multimerize (Büechi & Bächi, 1982) and is found just beneath the viral lipid bilayer. It has been shown several times to associate with cellular membranes (Stricker, Mottet et al., 1994;Sanderson, McQueen et al., 1993).

However, little is known on how M is targeted to the plasma membrane and how M is driving viral assembly and by which mechanisms it is able to promote the formation of a bud. Thus, we wanted to clarify some of these aspects.

In the first part of this study, we tried to identify a minimal region necessary and sufficient to target M at the plasma membrane. For this purpose, several mutants and deleted M proteins were performed and fused to two different tags, namely HA or GFP.

These different versions of M were expressed either in an infection context, or by transient expression. Our studies clearly indicated that minimal changes in the primary structure of M, strongly affected its phenotype and especially its targeting to the plasma membrane. Therefore, we were not able to identify a minimal region necessary and sufficient to reach the plasma membrane, but after the mutant’s analysis we concluded that if such a region would exist, it would be within the 2/3^th N-terminal portion of the protein. Nevertheless, this approach allowed us to identify two interesting mutants within the C-terminal part of the protein, namely HA-M^YPNV_4A and HA-M^IRKL_4A, showing partial phenotypes. These mutants, where two motifs (YPNV and IRKL) were replaced by Alanines, retained a little capacity to accumulate at the plasma membrane, but not in the same extend as the wild-type protein.

We took advantage of the newly developed HA-M mutants and the availibility of three different versions of the C protein to study their relationship when transiently co- expressed. In fact, the non-structural C protein was also proposed to play a role for SeV assembly and budding (Hasan, Kato et al., 2000), and we were concerned about a possible cooperations of the two proteins for that purpose. In this context, we were able to show that wild-type M and C proteins could independently reach the PM, where they greatly co-localized. This co-localization was also verified during SeV infections, supporting a possible cooperation.

In the second part of this study, we wanted to identify cellular partners of M. In order to find M interacting proteins, we chose two different approaches: the first one consisted in the purification of M under native conditions, a technique called Tandem Affinity Purification (TAP), while the second one, consisted in the collaboration with a group (Vidalain and colleagues) working on a high throughput yeast two-hybrid system.

The combination of the two approaches allowed us to identify several potentially new partners. We chose to analyse among the potential partners two proteins: charged multivesicular body protein 5 (CHMP5) and cytoplasmic actin 1 (β-actin).

Therefore, we focused our attention on one member of the MVBs machinery, namely CHMP5, which was identified to interact with the wild-type M thanks to the two- hybrid screening. This interaction was lost with the M^YPNV_4A mutant. Interestingly, the

YPNV motif was proposed to be an “L-domain” responsible for interacting with a member of the MVBs machinery. In order to verify any important implication of CHMP5 during SeV multiplication, we successfully knocked-down the protein by the use of small interfering RNA. However, 90% suppression of CHMP5 had no deleterious effects neither on SeV replication nor on SeV budding. We therefore concluded that even if this interaction could be real, it had no biological significance for the virus.

In the third part of this study, we verified the importance for SeV multiplication of the β-actin, which was identified after the Tandem Affinity Purification. β-actin, together with γ-actin, is one of two isoforms that are expressed in all mammalian tissues and essential for cell survival (Karabinos, Schmidt et al., 2001). Thanks to a collaboration with Christine Chaponnier, we took advantage of two specific antibodies against β-actin and γ-actin isoforms in order to follow any modifications in the actin cytoskeleton.

Upon SeV infection, both isoforms, but especially β-actin, were completely apically re-localized during SeV infection in MDCK cells at places where the virus was supposed to bud. We were able to show that this change of pattern was dependent on a lytic infection where M was highly unstable and consequently poorly expressed. We also demonstrated that this actin cytoskeleton re-distribution was specific for Sendai virus, as it was never observed after vesicular stomatitis virus (VSV) or human parainfluenza type 5 (HPIV5) infections. Knock-down of both isoforms with small interfering RNA allowed us to definitely show that actin was involved in SeV budding. We searched for additional evidences that M was implicated in this remodeling. We evidenced clear co-localizations of M with both actin isoforms where β- and γ-actin re-localize. We checked if M could have been sufficient for promoting such a remodeling by simply transient expression of wild-type M and two mutants, namely M30 and M^IRKL-4A. It was difficult to interprete of these results, as Mwt completely affected the overall shape morphology of the cells, which was not the case with the mutants. Nevertheless, this was an indication of a possible cytoskeleton manipulation by the wild-type protein that would not be the case for the mutants anymore. However, M is probably not sufficient to promote this remodeling by itself. Nonetheless, we think that M is necessary for that purpose, but with the help of additional factors.

General introduction

The first isolation of Sendai virus (SeV) came from the lung of a newbornchild who died of fatal pneumonia in 1952 at Tohuku University Hospital, Sendai (Japan). The primary isolation was made in mice instead of eggs and caused influenza-like symptoms in mice lungs. Later, in 1954, it was proposed that the mice used for the isolation were contaminated with a new virus, which was Sendai virus (FUKUMI, NISHIKAWA et al., 1954). Since then, Sendai virus has been found all over the world in mouse colonies and also in rats, hamsters and guinea pigs (Parker & Reynolds, 1968). However, Skiadopoulos and co-workers reported recently that an egg-adapted strain of SeV (Z strain), which is moderately virulent for laboratory mice, could replicate as efficiently as human Parainfluenza 1 (HPIV1) in the monkey and chimpanzee models of human respiratory infection (Skiadopoulos, Surman et al., 2002). Moreover, it is known that some of the original patient isolates have not been passaged in mice (Ishida & Homma, 1978). Thus, according to these datas, it is possible that SeV could have been responsible for the severe outbreak in Sendai (Japan) in 1952.

Two types of epidemiological patterns of infections have been recognized: the epizootic and the enzootic type.

The epizootic type is generally acute and is accompanied by a very high mortality.

In contrast, the enzootic type is characterized by a latent state infection with a long persistence of the virus, ready to be activated under favorable conditions. Usually, this latter type of infection is the one commonly found in laboratory mice, rats and hamsters infected by SeV (Zenner & Regnault, 2000).

Classification

According to the ICTV (International Committee on Taxonomy of Viruses), Sendai virus (also known as murine parainfluenza virus type 1) is a member of the Respirovirus genus, which belongs to the Paramyxoviridae family. The Paramyxoviridae family (or Paramyxovirus family) is one of the four families forming the Mononegavirales order. The three others families are the Bornaviridae, the Filoviridae, and the Rhabdoviridae. This classification is based on four main criteria, which includes:

morphologic criteria, the organization of the genome, the biological activities of the proteins, and the sequence of the encoded proteins. The Paramyxoviridae are enveloped viruses with a pleomorphic, spherical or filamentous virion morphology. Their nucleocapsid has a helical morphology and contains a negative single stranded RNA genome of 13-18 kb tightly associated to the viral nucleoprotein. They all possess a virion-associated RNA-dependent RNA polymerase (RdRp) for the replication and the transcription of the genome and share a very similar genome organization: 3’untranslated region-core protein genes-envelope protein genes-a large polymerase gene-5’untranslated region. Importantly, this family is thought to have an exclusively cytoplasmic replicative cycle.

The Paramyxoviridae family has been reclassified in 2000 and has been subdivided in two subfamilies, the Paramyxovirinae containing the Respirovirus (SeV), the Avulavirus, the Hepinavirus, the Morbilivirus and the Rubulavirus genera, and the Pneumovirinea which contains the Pneumovirus and the Metapneumovirus genera (Table 1).

Table 1: Classification of the Mononegavirales order. Adapted from http://www.ictvonline.org/index.asp

The Paramyxoviridae virion structure

Most Paramyxoviridae are enveloped viruses with a spherical morphology of about 150-300 nm in diameter but they can also have filamentous or pleomorphic forms.

The lipid bilayer envelope is derived from the host cell plasma membrane, in which are inserted the viral glycoproteins sipkes. In the case of SeV, the two glycoproteins incorporated in the envelope are the fusion glycoprotein F (present as a trimer) and the hemagglutinin-neuraminidase HN (present as a tetramer). These integral membrane proteins are implicated in the entry of the virus (adsorption and fusion) and in the exit following multiplication (neuraminidase activity). Some Paramyxoviridae members (Rubulavirus and Pneumovirinae) encode for a third integral membrane protein called SH

(Small Hydrophobic protein). In the case of PIV5 (Rubulavirus) and possibly also RSV (Pneumovirus), SH is believed to block virus-induced apoptosis through the inhibition of the TNF-α pathway (Wilson, Fuentes et al., 2006;Fuentes, Tran et al., 2007). Tightly associated to the inner leaflet of the viral envelope, stands the most abundant protein of the virion, the Matrix protein (M). This protein is thought to interact with itself forming multimers, with the lipid bilayer, with the cytoplasmic tails of the glycoproteins HN and F, and with the nucleopcapsid. Thus, M contacts all the viral components and is considered to be the main organizer of the viral morphogenesis. Finally, in the core of the virion, the nucleoproteins (N) and genome RNA together form the helical nucleocapsid, to which the phosphoproteins (P) and the large proteins (L) are attached. One L protein and four P proteins form together the viral polymerase complex that is necessary for both transcription and replication activities.

Figure 1: Schematic representation of a Sendai virus particle. (Swiss Institute of Bioinformatics, http://www.expasy.ch/viralzone)

Sendai virus genome organization

SeV was completely sequenced in 1986 and contains precisely 15’384 nucleotides (Shioda, Iwasaki et al., 1986). The organization of the nonsegmented single stranded

RNA genome of negative polarity was already mentioned in the “classification” section (see also Figure 2). The genome contains a 3’extracistronic region of 55nt (called leader) and a 5’extracistronic region of 57nt (called trailer), which flank the six genes in the order N, P, M, F, HN and L. These regions are very rich in A/U and are essential for transcription and replication. At the beginning and end of each gene, there are transcription regulatory sequences called respectively gene-start and gene-end, which are transcribed. Between each cistron, a small intergenic non-transcribed region of three nucleotides (GAA, except in one case GGG) is present. Each gene codes for one corresponding protein, except in the particular case of the P gene, which allows the production of the following polypeptides: P, V, W, C’, C, Y1, Y2 and X. The mechanisms involved for this multiple production includes RNA editing (P, V and W) and the use of alternative translation initiation codons (C’, C, Y1 and Y2) (Curran &

Kolakofsky, 1990;Lamb & Parks, 2007) (see also figure 4 in section “The C protein”).

Figure 2: Schematic representation of SeV genome organization. (Swiss Institute of Bioinformatics, http://www.expasy.ch/viralzone)

Overview of a Paramyxoviridae replication cycle

The virus adsorption and entry, as already mentioned, is mediated by respectively HN and F. HN attach to cellular receptors, which are sialic acid containing molecules in the case of SeV HN (Figure 3). After the attachment, the F protein mediates a pH- independent fusion between the viral membrane and the cellular plasma membrane. The fusion allows the release into the cytoplasm of the viral nucleocapsid from M by an unknown mechanism, where all the replication cycle can take place as far as it is known.

At this point, the viral polymerase complex transcribes each gene into the corresponding viral mRNA. Since at each stop/restart signal, the RNA polymerase has a certain

probability to release the template, the genes are transcribed with a gradual decrease from the 3’end to the 5’end of the genome. After translation of the primary transcripts and accumulation of the viral proteins, the replication of the antigenomes (5’-3’) begins, which in turn serves as templates for new copies of genomic RNA (3’-5’). Importantly, no genomic or antigenomic RNA are supposed to be “naked” in the cytoplasm. In fact, newly synthesized N proteins immediately associate with nascent genomic or antigenomic RNA to form the helical nucleocapsid structure. The nucleocapsid represents the active template for the viral polymerase complex, which is unable to transcribe or to replicate from a “naked” RNA. Viral glycoproteins are synthesized in the endoplasmic reticulum and go through the secretory pathway to the plasma membrane. In contrast, M is synthesized in the cytoplasm and accumulates at the inner leaflet of the host plasma membrane where it is believed to act as the central organizer of the viral assembly concentrating all the viral components. Finally, the virions are assembled at the plasma membrane by a process called budding.

Figure 3: Paramyxoviridae replication cycle (Takimoto & Portner, 2004). For details see the text.

Virion assembly and budding

As previously mentioned, Paramyxoviridae assemble at the plasma membrane from which they bud. A mechanism called pinching-off allows the release of the viral particles. The budding process is believed to require a high coordination of the different viral components and possibly of some host cellular factors. This coordination is mainly directed by the viral matrix protein. In the next sections we will focus on the different viral components namely the nucleocapsid, the matrix protein, the two glycoproteins, the C protein, and their roles and dynamics in the process of assembly and budding.

The Nucleocapsid assembly

The SeV nucleocapsid is composed of precisely 2564 N proteins directly bound to the viral RNA and approximately 300 P and 50 L proteins forming a left-handed helical structure of 1 μm length and 18 nm in diameter (Lamb, Mahy et al., 1976;Compans &

Choppin, 1967;Finch & Gibbs, 1970). Importantly, each SeV N protein binds exactly to six nucleotides (Egelman, Wu et al., 1989), which is not always the case for other Mononegavirales, like i.e. for VSV whose N protein binds to 9 nucleotides (Thomas, Newcomb et al., 1985;Albertini, Wernimont et al., 2006;Green, Zhang et al., 2006).

The SeV nucleocapsids assemble in the cytoplasm in two steps (Kingsbury, Hsu et al., 1978). First, the N proteins bind the genome RNA to form the left handed helical structure and secondly, the polymerase complex composed of L and P binds to the nucleocapsid.

The Matrix protein

The Paramyxoviridae matrix protein (M) is believed to play a key role in viral budding, and is perhaps therefore the most abundant protein in the virion (Lamb & Parks, 2007). SeV M protein can self-associate forming a crystalline array in infected cells

(Büechi & Bächi, 1982) or after puirification (Heggeness, Smith et al., 1982;Hewitt, 1977;Hewitt & Nermut, 1977), and is found just beneath the viral lipid bilayer as shown by electron microscopy (EM)(Lamb & Parks, 2007). Interestingly, the M protein of vesicular stomatitis virus (VSV), a Rhabdoviridae, is known to polymerize in a two- dimensional fashion, indicating that M-M interaction is not due to aggregation (Gaudin, Barge et al., 1995;Gaudin, Sturgis et al., 1997). M of Paramyxoviridae has been shown several times to be associated with cellular membranes as for Newcastle disease virus (NDV) (Faaberg & Peeples, 1988;Nagai, Ogura et al., 1976), or for SeV (Stricker, Mottet et al., 1994;Sanderson, McQueen et al., 1993). As already mentioned, SeV M has also been shown to be associated with the viral nucleocapsid (Stricker, Mottet et al., 1994) and with the glycoproteins (Sanderson, McQueen et al., 1993;Sanderson, Wu et al., 1994), but the multiple interactions of M will be refered better in the section “Protein- protein interactions”.

In the case of Sendai virus, M contains 348 amino acids (~38kDa), is a basic protein (pI 9.68), and is moderately hydrophobic. The rest of the protein is predicted to include little secondary structure, namely 11% α-helix, 19% β-pleated sheet and 70%

random coil (Giuffre, Tovell et al., 1982).

The SeV M protein can be partially phosphorylated on Ser⁷⁰, but depending on the cell type, the phosphorylated form may not be found in the virion (Lamb & Choppin, 1977). Moreover, a virus mutant, where the M phosphorylation site was eliminated with a substitution (S70A) was successfully rescued, indicating that the phosphorylated M is not essential for virus multiplication either in cell cultures or in mice (Sakaguchi, Kiyotani et al., 1997).

In contrast to this particular M mutant, most mutations on matrix protein usually have deleterious or catastrophic effects on its budding function. This is not surprising in the sense that SeV matrix protein is essential for viral budding. In fact, efficient suppression by siRNA of M provokes about a 100-fold diminution of SeV particles production (Mottet-Osman, Iseni et al., 2007) and deletion of M in a recombinant SeV abort almost completely viral particles formation (Inoue, Tokusumi et al., 2003). This is also true for another Paramyxoviridae, namely Measles virus (MeV), where the absence of M reduces the viral titer of about 250 fold (Cathomen, Mrkic et al., 1998).

Interestingly, both viruses deleted for M shows extensive syncytia formation indicating a large propagation by cell-to-cell spreading. This particular feature was linked to a rare complication of MeV infection called subacute sclerosing panencephalitis (SSPE) where M-defective MeV genomes expand clonally in the CNS (Baczko, Lampe et al., 1993), strongly suggesting a cell-to-cell spreading. This type of persistent infection of the brain is a progressive and fatal neurological disease observed one in 300’000 victims of measles (Acheson N.H., 2007) and defects of M are very common in SSPE, which correlates with strong viral particles production diminution (Cattaneo, Schmid et al., 1988).

It is also possible to promote persistent infections with Sendai virus using defective interfering particles (DI). Like with MeV, M appears to be highly implicated in the change from a lytic to a persistent infection. In fact, BHK cells persistently infected with Sendai virus showed that the viral M protein was greatly reduced in amount or absent in these cells (Roux & Waldvogel, 1982).

Another strong indication of the central role played by M in viral assembly and budding came from a temperature-sensitive mutant of SeV (Cl.151), in which viral particles production was significantly impaired. This mutant containing three amino acids changes in its M protein was highly unstable at non-permissive temperature (38°C) and therefore was unable to accumulate sufficient M protein in infected cells to promote efficient viral budding (Yoshida, Nagai et al., 1979;Yoshida, Hamaguchi et al., 1982).

The viral particles production could be restored by supplementation of not only the M wild type protein, but also of the Cl.151 M protein (Kondo, Yoshida et al., 1993).

A motif very well conserved among Paramyxoviridae (Table 1) was studied in the context of SeV and MeV infection. This motif, composed of the amino acids residues VRRT, was mutated and analyzed in the context of a SeV infection (Mottet, Muller et al., 1999). The mutant, called HA-M³⁰ is an M protein carrying two substitutions, namely Thr¹¹²Met and Val¹¹³Glu, and tagged with an influenza virus haemagglutinin epitope (HA). It was expressed by its insertion in a DI genome, which allows a strong expression of the inserted gene under conditions of infection and co-expression with M^wt. This mutant was shown to bind to membrane fractions and to accumulate to perinuclear regions. It was not able to co-precipitate M^wt as an HA-Mwt did in the same context.

Finally, HA-M³⁰ was also shown to interfere with the binding of nucleocapsids to membranes suggesting again the critical role of M in the viral assembly and budding.

This mutant was also inserted in a recombinant SeV instead of the Mwt protein, but the recovery of this virus by reverse genetic has never been possible (Mottet & al., unpublished), indicating that essential M functions are disrupted in the mutant protein.

Later, a group working on MeV also decided to mutate this conserved motif with a Val¹⁰¹Ala substitution in the MeV matrix protein and to study the phenotype by expressing it in a recombinant virus containing the mutated M protein instead of the wild- type one (Runkler, Pohl et al., 2007). This recombinant virus had a low titer and virus release into the supernatant was reduced by 10 to 100-fold. Syncytia formation in cells infected by this virus was increased indicating the incapacity of M to downregulate the MeV glycoprotein-mediated cell-to-cell fusion. This mutated M protein had a rapid turnover either when expressed by infection or by transient expression and was unable to accumulate at the cell surface (as for SeV) not allowing the nucleocapsid transport at the plasma membrane. In contrast to SeV HA-M30, the MeV M Val¹⁰¹Ala protein retained apparently its ability to bind to other matrix and nucleoproteins and had a reduced capacity to bind to intracellular membranes, but these experiments were only done with transient protein expressions and not in the context of a viral infection.

Taken together, both studies highlighted again the central role of matrix’s Paramyxoviruses proteins just by analyzing a protein with one or two point mutations.

Finally, this very well conserved motif could either be responsible for an interaction with a cellular partner or could contribute to the correct overall conformation of the protein.

Table 1: Partial M protein sequence comparison among Paramyxoviridae. The M protein alignment in the region encompassing the putative α−helix (shaded box; residues 104–119 in SeV) emphasizes the conservation of the tetrapeptide VRRT (in bold; residues 113–116 in SeV) at an almost identical position in all the M proteins, as indicated by the residue numbering at the right. HPIV, Human parainfluenza virus ; BPIV, bovine parainfluenza virus ; RPV, rinderpest virus ; DDV, dolphin distemper virus ; CDV, canine distemper virus ; NDV, Newcastle disease virus ; SV41, simian parainfluenza virus 41; SV5, simian parainfluenza virus 5. Sequences were obtained from GenBank (Mottet, Muller et al., 1999).

Recently, another way has been adopted by researchers to study the matrix proteins of different Paramyxoviruses without doing any infection. Many of them have indeed chosen to study these proteins by transient expression and to look at formation and release of virus-like particles (VLPs) very likely because most mutations are “lethal” for the viruses.

When transiently expressed, several M proteins are found in the supernatants not only for Paramyxoviruses like in the case of SeV and hPIV1 (human parainfluenza virus 1) (Coronel, Murti et al., 1999) but also for other viruses like VSV (vesicular stomatitis virus) (Harty, Paragas et al., 1999), Ebola and Marburg viruses (Harty, Brown et al., 2000), and Influenza virus (Gomez-Puertas, Albo et al., 2000) among several. These studies led to indentification of some conserved motifs, called “L-domains” (late- domains), responsible for the recruitment of components of the VPS (vacuolar protein sorting) cellular machinery. This particular pathway and its implication for the viral budding will be discussed in more details in a further section of this introduction (see

“Hijacking of the multivesicular bodies (MVB) machinery”).

Expression of SeV M from transfected cells result in the formation of VLPs of size between 50-150 nm in diameter, which is slightly smaller than complete virus particles of about 200 nm in diameter (Takimoto, Murti et al., 2001b). Using this approach, Takimoto & al., demonstrated that a mutant deleted for the last five C-terminal amino acids of SeV M, namely KIRKL (amino acids 344-348), was no more competent for VLPs formation. Because this motif is similar to an actin-binding domain (KLKK identified in thymosin β4 protein (Van, Dewitte et al., 1996)) and because actin was found in the virus-like particles of M, it was proposed that the KIRKL sequence was responsible for a direct interaction of M with cellular actin (see for more details the section “The M proteins and the cellular partners”).

Another motif was identified by this kind of approach, YLDL (amino acids 50- 53), which was proposed to contact a member of the VPS machinery, namely Alix (see for more details the section “The M proteins and the cellular partners” and “Hijacking of the multivesicular bodies (MVB) machinery”). This motif, turned out to be essential for the production of VLPs.

Not surprisingly, also the matrix protein of hPIV1 (which is very close to SeV), was shown to be sufficient to induce its own vesiculation (Coronel, Murti et al., 1999).

Furthermore, the co-expression of M and NP resulted in the production of vesicles containing nucleocapsid-like structures.

The matrix of NDV (Newcastle disease virus), another Paramyxoviridae (Avulavirus genus) was also demonstrated to be sufficient to promote the release of VLPs (Pantua, McGinnes et al., 2006). Furthermore, it was absolutely necessary to allow the release in the supernatant of the other viral proteins, namely NP, HN and F. When expressed all together, the four proteins were found in the same VLPs with a density very close to that of true virus particles.

Interestingly, unlike SeV, hPIV1 and NDV, PIV5 matrix protein transiently expressed is not sufficient to induce virus-like particles production. It was found that M protein in the form of VLPs was only secreted from cells, when M protein was coexpressed with one of the two glycoproteins, HN or F, together with the nucleocapsid (NP) protein (Schmitt, Leser et al., 2002). These VLPs appeared similar morphologically to authentic virions by electron microscopy. Later, to confirm the central role of M in this context a motif composed of 4 amino acids, FPIV, was identified within the PIV5 M protein that linked the virus to the MVB machinery. When mutated, the M protein was no more able to generate VLPs and a recombinant virus containing the mutations failed to multiply normally (Schmitt, Leser et al., 2005).

Thus, it is interesting to notice that very close Paramyxoviruses as SeV or PIV5 that have common overall assembly and budding strategies seem to differ in the role of the M protein, according to results obtained from VLPs studies.

The Glycoproteins HN and F

All Paramyxoviridae encode at least two integral membrane proteins. Some of them (Rubulaviruses and Pneumoviruses) possess also a third integral membrane protein (SH). All integral glycoproteins are inserted to the membrane by an anchor segment of their polypeptide and project most of their structure to the outside, forming the so-called viral spikes. After synthesis, the viral glycoproteins undergo conformational maturation assisted by folding enzymes and molecular chaperones (Lamb & Kolakofsky, 2001). The corrected folded proteins are targeted to the endoplasmic reticulum and to the Golgi apparatus where they acquire the necessary modifications to become biologically active.

These modifications include gylcosylation and disulfide bond formation in order to promote their oligomerization. As already mentioned, one of them is responsible for the cell attachment (HN, H or G) and the second one (F) mediates the fusion between the viral membrane and the plasma membrane of the host cell in a pH-independent fashion.

In this section, we will focus only on F and HN proteins.

The attachment proteins of Respiroviruses have both hemagglutination (capacity to agglutinate erythrocytes) and neuraminidase (capacity to destroy the receptor) activity, and have been therefore called HN proteins. This glycoprotein is known to be the major antigenic determinant of the Paramyxoviridae. The Respiroviruses HN binds to sialic acid-containing cell surface molecules and mediates what is called adsorption. HN has also enzymatic activity and cleaves sialic acid from the infected cell surface and from the virion surface. This activity is believed to be necessary to prevent self-aggregation of viral particles during the release from infected cells. The HN protein is a type II integral membrane protein that spans the membrane once and contains an N-terminal cytoplasmic tail, a single N-terminal transmembrane (TM) domain, a membrane-proximal stalk domain, and a large C-terminal globular head domain (Lamb & Parks, 2007)(Figure 4).

This last domain is responsible for all the biological activities, which are the adsorption and the enzymatic activity mentioned above. The stalk domain is essential for the homotetramer formation of HN, which is a non-covalent association of two covalently bound dimers (S-S)(Lamb & Parks, 2007). Importantly, HN is also known to enhance the F fusion activity and this particular feature is called “fusion promotion activity” as it has

been describe for NDV (Morrison, McQuain et al., 1991) and hPIV3 (Moscona & Peluso, 1991).

The fusion protein of Paramyxoviridae is a typical type I glycoprotein with a cleaved N-terminal hydrophobic signal peptide and a C-proximal membrane anchor (Figure 4). This protein is responsible not only for the fusion between the viral and the plasma membranes, but later in the infection it can also mediates the formation of syncytia by fusion of neighbouring cells. F forms homotrimers and is synthesized as an inactive precursor, F0, which is converted into the active form (the fusogenic form) after cleavage by host cell proteases at the cleavage activation site. The cleavage leads to the formation of two subunits, F2 and F1 which remain linked by a disulfide bond. The N- terminus of F1 is a hydrophobic region called “fusion peptide” that is thought to insert in the host plasma membrane to initiate the fusion process (Lamb & Parks, 2007).

Figure 4: Paramyxovirdae HN and F gylcoproteins. a) Hemagglutinin-neuraminidase attachement protein (based on the predicted sequence of PIV5 HN gene). b) Fusion protein (based on the predicted sequence of PIV5 HN gene). Adapted from(Lamb & Parks, 2007).

Both glycoproteins are known to play critical roles in virion assembly contacting by their cytoplasmic tail the matrix protein. However, each other importance varies among several members of the Paramyxoviridae family.

It has been shown in SSPE that not only M was a major determinant of this kind of MeV persistent infection, but that also the cytoplasmic tail of the F protein is altered (Schmid, Spielhofer et al., 1992), strongly suggesting an interference at the level of virus assembly and release. This kind of phenotype with strong cell-to-cell fusion activity has been confirmed with recombinant MeV that had truncation in their F cytoplasmic tails (Cathomen, Naim et al., 1998;Moll, Klenk et al., 2002).

It has been known for a long time that in the case of SeV, HN is completely dispensable for efficient viral production. The first evidences came from a temperature sensitive mutant, called ts271, which was shown to assemble into virus particles at the non-permissive temperature in the absence of any HN glycoprotein, thereby indicating that HN is dispensable for viral particles release (Portner, Marx et al., 1974;Portner, Scroggs et al., 1975;Markwell, Portner et al., 1985;Tuffereau, Portner et al., 1985;Stricker & Roux, 1991). A motif, SYWST (amino acids 10-14), was described to be necessary for the uptake of HN into the viral particles. First, Takimoto & al., were able to show that this motif was necessary for the incorporation of transiently expressed HN proteins into virions formed from infected cells (Takimoto, Bousse et al., 1998). Later, Fouillot-Coriou & al., confirmed this observation with a recombinant virus where the wild-type HN protein was replaced by mutants of HN (Fouillot-Coriou & Roux, 2000).

Interestingly, in virus like particles studies, it was demonstrated that HN could be found into VLPs when transiently co-expressed together with the matrix protein, which is not the case when expressed alone (Takimoto, Murti et al., 2001a). But surprisingly, in this study both proteins were found in different vesicles, suggesting that they do not interact.

In contrast to SeV HN glycoprotein, SeV F has been shown to be critical for efficient viral particles production. Truncations or mutations in the cytoplasmic tail clearly affect viral budding as demonstrated with recombinant SeV (Fouillot-Coriou &

Roux, 2000). In this context, a minimal region comprised between amino acids 538 and 550, was identified to be critical for virion production. Further studies identified more

precisely a motif within this region necessary for the release of F VLPs, namely TYTLE (amino acids 542-546) (Takimoto, Murti et al., 2001b). Interestingly, the co-expression of M enhances the released of F by a factor of four. But like in the case of HN, biochemical investigations pointed out that both proteins were found in distinct vesicles. However, in the same context, Sugahara et al. found that both glycoproteins and M were found in the same density fractions (Sugahara, Uchiyama et al., 2004). The discrepancies between these data are not well understood and further investigations will be needed to address this question.

The story concerning the roles for virus assembly and budding attributed to PIV5 HN and F glycoproteins differ a little bit from the situation observed in SeV.

Using recombinant viruses and deleting progressively the HN cytoplasmic tail, its presence was shown to be necessary for concentrating efficiently the other viral components into patches at the infected cell surface. Furthermore, the release of virus particles was very inefficient in the absence of the HN cytoplasmic tail (Schmitt, He et al., 1999). The same approach has been used to determine the role of the F cytoplasmic tail and it was demonstrated that it is dispensable for normal virus replication and budding (Waning, Schmitt et al., 2002). Consistent with this, the subcellular localization of F and HN were not altered. Thus, this situation seems to be quite different from the one observed with SeV, where HN is dispensable and F necessary. Subsequent analysis of recombinant PIV5 viruses containing truncations in both HN and F cytoplasmic tails, have shown a more deleterious effect for budding as compare to the recombinant virus lacking only the HN cytoplasmic tail (Waning, Schmitt et al., 2002). This suggested a possible redundant role between the two cytoplasmic tails of F and HN.

A confirmation of this possible redundancy came from the studies done with the VLPs approach. As already mentioned in the previous section, the formation of PIV5 VLPs takes place only when several viral components are transiently co-expressed, namely NP, M, and one of the two glycoproteins HN or F (Schmitt, Leser et al., 2002). Furthermore, the same truncation in the HN cytoplasmic tail as the one that was inserted in the recombinant virus described above inhibited the formation of VLPs when expressed either alone with NP and M or even when expressed with NP, M and F.

All these data coming from several Paramyxoviridae clearly indicate that F and/or HN play important roles for the virus assembly and budding. As suspected the cytoplasmic tails of both glycoproteins were shown to be a key portion of the macromolecules for that purpose. In fact, they are the only potential regions to be able to contact any of the other viral components. The interaction aspects between the viral components will be described in a next section of this introduction (see “Protein-protein interactions”).

The C protein

The C protein is an abundant protein that is known to interact with both the nucleocapsid and the L protein (Karron & Collins, 2007). It is found in small traces in the virus particles. The SeV C proteins are encoded by the P gene and comprise a set of four carboxy-coterminal proteins, namely C, C’, Y1 and Y2. These proteins are translated from alternative start codons, with the C protein ORF being in the +1 reading frame relative to the P ORF, and codes for polypeptides from 175 amino acids residues (Y2) to 215 amino acids residues (C’) (Figure 5). All four proteins share the same C-terminal residues (translation is terminated at the same Stop codon) and the C protein is the most abundant of the four (Karron & Collins, 2007;Lamb & Parks, 2007).

Figure 5: Representation of SeV P mRNA to illustrate the mechanisms of producing P,V, and C proteins (Lamb & Parks, 2007).

The SeV C proteins are multifunctional proteins (Karron & Collins, 2007;Lamb & Parks, 2007). They are well known inhibitors of innate immune response. By interfering with activation of interferon regulatory factor 3 (IRF-3) they inhibit interferon α and β (INFα/β) induction by the host cell. Another important function of C is to bind transducer and activator of transcription 1 (Stat1) to prevent the induction of not only IFNα/β but also INFγ. Additional roles that have been attributed to C proteins include inhibition of cellular apoptosis, regulation of viral RNA synthesis (Cadd, Garcin et al., 1996;Tapparel, Hausmann et al., 1997) and inhibition of cellular apoptosis (Karron &

Collins, 2007;Lamb & Parks, 2007).

Recently, the C SeV protein has been proposed to play an important role in virus assembly and budding. This is of particular interest because it was the first time that a non structural protein was suspected to play any role in this step of the viral multiplication. VLP release studies have shown that the C protein, which is not able to induce vesiculation by itself, could act as an enhancer of VLPs formation (Sugahara, Uchiyama et al., 2004). In fact, when co-expressed with M, NP, F and HN it was able to enhance VLPs formation by a factor comprised between two and three. However, some contradictories data appeared in this study as compared to others already mentioned in previous sections of this introduction. First, HN was shown to have an inhibitory effect in VLPs formation, in contrast to the observations made by Takimoto & al. showing no effect for M-driven vesicle release (Takimoto, Murti et al., 2001b). This was also contradictory with the notion that HN is dispensable for SeV budding as mentioned before. Second, in the study proposed by Sugahara & al. the viral proteins were found in the same density fractions, this was not the case for Takimoto & al. where HN, F and M were found in different density fractions.

Despite these contradictories data, other evidences support a role for C in promoting viral release. A recombinant SeV, which expressed none of the four C proteins (C’, C, Y1 and Y2) was not only highly attenuated for replication, but had also a reduced viral yield (Hasan, Kato et al., 2000). Nevertheless, the interpretation of this result is quite hard because of the multiple roles of the C protein and all their implications in a correct viral multiplication cycle.

Another evidence and possible explanation for the SeV C protein budding enhancer’s role is the potential link to the VPS machinery through its interaction to Alix (also called AIP1) (Sakaguchi, Kato et al., 2005), a member of this pathway. This study used transiently expressed proteins and measured virus budding efficiency by the use of VLPs. The results suggested that Alix played a role in efficient VLPs budding and that SeV C protein facilitated this budding through its interaction with Alix. The interaction was shown to take place most likely between the amino terminus of Alix and the carboxyl terminus of C protein, but no “L-domain” motif has been identified so far. The mechanism by which this interaction mediates an enhancement of VLPs formation is not well understood, but it was later shown that it is dependent on the ability of C protein to recruit Alix at the plasma membrane (Irie, Nagata et al., 2008c). Hence, it seems that any role attributable to C protein as an enhancer’s budding could be dependent on its ability to be targeted at the plasma membrane. Consistent with this, the inhibitory role on Stat1 was already shown to be dependent on its aptitude to localize at the plasma membrane (Marq, Brini et al., 2007). The target and the binding at the plasma membrane were shown to be dependent on a peptide-only membrane anchor at the N-terminus of the C protein (Marq, Brini et al., 2007). It is interesting to note that two SeV proteins seem to contact Alix, as M (see previous section “The matrix protein”) does it as well in an independent way of C protein (Irie, Shimazu et al., 2007).

However this interesting story suffers because of a recent publication demonstrating that in the context of a SeV infection (and not in a VLPs background) budding is not sensitive to neither small interfering RNA (siRNA) against Alix, nor to a dominant negative (VPS4A) of the VPS machinery (Gosselin-Grenet, Marq et al., 2007).

Thus, the question whether C protein is important or not for SeV budding is still open and it is likely that more than one exit way could be taken during an infection.

Protein-protein interactions

The M protein and the viral components

In order to play its attributed roles in virus assembly and budding, M has to contact several viral members.

The current model for correct assembly and budding of Paramyxoviruses stipulates that all the structural components have first to be transported to the site of assembly. There, the virion members have to interact with each other to initiate the budding process. Obviously, the interactions could potentially take place not only during the time of assembly, but also during the transport of some of the components.

Thus, it is not surprising if M-nucleocapsid and M-glycoproteins interactions are supposed to play a very important role in the proper assembly and budding.

M and the nucleocapsid

For HPIV1 (the human counterpart of SeV) it has been shown that M and the nucleoprotein interact in a very specific fashion. As already mentioned, when transiently expressed, HPIV1 M protein can induce the formation and release of vesicles called VLPs. On the other hand, NP protein of this virus is not capable to do the same as M, but when transiently expressed it produces intracellular nucleocapsid-like structures (NC-like structures) (Coronel, Murti et al., 1999). These NC-like structures, when co-expressed with M, were found in the VLPs induced by the matrix protein. Furthermore, the vesicles were similar in size to those of native virus particles as shown by electronic microscopy (EM). All together, these results suggested a possible interaction between the two proteins.

These intracellular NC-like structures were also observed for SeV when NP was transiently expressed (Coronel, Takimoto et al., 2001). In the same study, Coronel & al.

co-expressed HPIV1 NP and SeV NP and demonstrated that the NC-like structures

formed contained both proteins. This finding suggests that the domains required for NP- NP interaction, and thus nucleocapsid formation, are very well conserved in the NPs of both viruses. This was all consistent with the previous notion that the strongly conserved N-terminal portion of the Paramyxoviridae NP proteins was responsible, among other functions, for the NP-NP association (Lamb & Parks, 2007). Very interestingly, in cells infected with SeV and transfected at the same time with HPIV1 NP, it was noticed that pure nucleocapsids composed of only SeV NP or HPIV1 NP were detected with the following ratios: 50% of mixed nucleocapsids, 25% of homologous SeV, and 25% of homologous HPIV1. This suggested that other viral components were responsible for the formation of nucleocapsids composed of homologous NPs. Unexpectedly this ratio was not found anymore within the SeV virions, where the nucleocapsids containing HPIV1 nucleoproteins were highly excluded. This suggested very specific interactions that allow the uptake of the nucleocapsid into the virion. Supporting this notion, co-expression of HPIV1 NP and HPIV1 M in SeV infected cells increased significantly the uptake of nucleocapsids containing HPIV1 NP into the SeV virions. This strongly indicated a direct interaction between M and NP that allowed the specific uptake of nucleopcapsids into the virions. By the use of chimera SeV-HPIV1 NP, it was shown that the C-terminal region between amino acids 421 to 466 of NP would have been responsible for the interaction between M and NP (Coronel, Takimoto et al., 2001). This kind of interaction was previously shown to take place in SeV infected cells by cross-linking experiments (Markwell & Fox, 1980).

Sugahara & al. found SeV M and N in the same VLPs (Sugahara, Uchiyama et al., 2004) and during SeV infection it was proposed that M could be recruited by the viral nucleocapsid in the cytoplasm in perinuclear regions to participate in viral assembly and budding (Stricker, Mottet et al., 1994). Together, these results suggest an interaction between the two proteins.

For MeV, as already mentioned, it was shown that the proper localization of the nucleocapsid is dependent on the stability and the quantity of the matrix protein (Runkler, Pohl et al., 2007). Furthermore, M and N were co-immunoprecipitated when co- expressed. Taken together, these results suggest a possible direct interaction between M and N proteins for this virus.

NDV (Newcastle disease virus) NP protein was only found in VLPs when co- expressed with at least M, and immunoprecipitations from Triton X-100 treated VLPs could show a direct interaction between N and M (Pantua, McGinnes et al., 2006).

M and the glycoproteins

The interaction between M and the glycoproteins has been under investigation for a long time, but definitive prove of this interaction is still lacking for most of the Paramyxoviridae. Nevertheless, the sum of several evidences tends to argue for direct interactions between the matrix and HN or F. Some evidences came from studies done in the context of virus infections, but most of them came again by the use of VLPs systems already described in this introduction.

SeV M protein was shown by confocal microscopy to partially co-localize with both glycoproteins in infected BHK cells (baby hamster kidney) (Ali & Nayak, 2000).

This co-localization was observed either at the cell periphery or at the perinuclear region (possibly mid-Golgi) and this especially when cells were treated with monensin, known to block the exocytic transport of the glycoproteins F and HN. Experiments on membrane-associated M protein also suggested interactions with the two glycoproteins (Ali & Nayak, 2000). In fact, M expressed from recombinant vaccinia virus (recVV) was shown to be associated with cellular membranes, but M was completely solubilized after treatment with 0.03% TX-100 (triton X-100) in contrast to the glycoproteins F and HN that were predominantly detergent-resistant. When M was co-expressed with F and/or HN (by co-infections of recVV encoding for M, F or HN), the membrane-bound M fraction became mostly detergent-resistant, indicating that M could be recruited specifically at sites of detergent-resistant membranes (DRMs) by the glycoproteins.

Furtheremore, chimeras SeV F containing the transmembrane domain or the cytoplasmic tail of influenza H protein suggested that both domains could help recruiting M at these DRMs. However, the pertinence of the DRMs for viral assembly and viral budding is still controversial for SeV, and it will be discussed in a further section of this introduction (“A platform of assembly: the lipid rafts”).

A previous study on SeV also suggested a possible interaction of M with the glycoproteins at the Golgi membranes during SeV infection. By the use of low- temperature incubations or the ionophore monensin, the authors were able to block the viral glycoproteins transport to the Golgi membranes, where M was also seen to accumulate and reflected the F protein distribution (Sanderson, McQueen et al., 1993).

Importantly, M was redistributed when the temperature was turned again at 37°C.

Thus, several data suggest an interaction between SeV M and both glycoproteins.

Concerning F, it seems that both the cytoplasmic tail and the transmembrane domain of the glycoprotein could interact with M. Concerning SeV HN, as already mentioned, its cytoplasmic tail and more precisely the motif SYWST could be responsible for the interaction with the matrix protein. In fact chimera NDV HN that contained the SeV cytoplasmic domain fused to the transmembrane and external domains of the NDV was incorporated in SeV virions when added in trans, which was not the case for wild-type HN NDV (Takimoto, Bousse et al., 1998). Further characterization of this phenotype demonstrated that the five amino acids SYWST were required for this incorporation in the virions. Hence, the SeV HN cytoplasmic tail seems to possess a signal that gives specificity for incorporation in the virion, and as in the case of the HPIV1 nucleocpasid (Coronel, Takimoto et al., 2001) described in the previous section, it is tempting to speculate that the interaction with M is responsible for this selectivity. The importance of the motif SYWST was then confirmed in an infection context with recombinant SeV lacking portion of the HN cytoplasmic tail (Coronel, Takimoto et al., 2001;Fouillot- Coriou & Roux, 2000).

Curiously, the absence of glycoprotein cytoplasmic tails of several viruses seems to be deleterious for the specificity of assembly and protein incorporation into the virions. In fact, cellular proteins were incorporated in a larger amount into virions when the glycoprotein cytoplasmic tails of MeV or PIV5 were either mutated or truncated in recombinant viruses (Cathomen, Naim et al., 1998;Schmitt, He et al., 1999). This indicated that glycoprotein cytoplasmic tail-truncated viruses had defect in the exclusion of cellular host proteins from progeny virions.

In cells infected by recombinant PIV5 viruses (recPIV5) with truncated HN cytoplasmic tails, it was observed that the characteristic patches distribution on the cell

surface of the wild-type HN (wtHN) was lost as seen by confocal microscopy.

Interestingly, the M protein was also observed in patches on the cell surface in a distribution very similar to that observed for wtHN after infection with recPIV5 containing wtHN. However, in cells infected by the cytoplasmic-tail-truncated recPIV5, M protein was found to be distributed randomly throughout the cytoplasm (Schmitt, He et al., 1999). Unfortunately, the authors were not able to perform double antibody labeling in order to confirm that HN and M were present at the same patches close to the cell surface. Nevertheless, it is reasonable to imagine that they would have co-localized, and that this would have strongly suggested an interaction between the two proteins.

Investigations support evidences of co-localizations between M and one or both of the glycoproteins for MeV. This was shown in an infection context with recombinant MeV lacking the matrix proteins (MeV-ΔM). In a wild-type MeV infection, the glycoproteins and M co-localized in patches inside the cell and at the plasma membrane, but in MeV-ΔM infected cells there was in contrast an overall homogeneous distribution of both glycoproteins (Cathomen, Mrkic et al., 1998). This strongly supports interactions between the matrix protein and the glycoproteins. Further studies with recombinant MeV containing tail-less glycoproteins confirmed these co-localizations and suggested that the regions responsible for M-H or M-F interactions could be also elsewhere than in the cytoplasmic tails (Cathomen, Naim et al., 1998). Nevertheless, the cytoplasmic tail of F appeared to be required for accurate virus assembly, because tail-less F were less abundant in virus particles. This may be due to a direct interaction between M and F cytoplasmic tail, since it was shown that a hybrid VSV (vesicular stomatitis virus, a Rhabdoviridae) glycoprotein containing the MeV F cytoplasmic tail directed the incorporation of the MeV M protein in the envelope of the MeV-VSV chimera whereas a standard VSV glycoprotein did not (Spielhofer, Bachi et al., 1998).

Another Paramyxoviridae, NDV, was also studied in order to determine whether M interacted with the glycoproteins or not. This was done in a VLPs system were the M protein was shown to be necessary for any vesicle production containing either HN or F with M. In fact, expression of HN, F or NP in all possible combinations but without M, totally failed to produce VLPs (Pantua, McGinnes et al., 2006). Electronic microscopy (EM) and biochemical analysis strongly suggested that all proteins M, HN, F and N,