Z E N T R U M F Ü R I N F E K T I O N S B I O L O G I E
ZIB ZIB
Z E N T R U M F Ü R I N F E K T I O N S B I O L O G I E
The Role of Lipid Rafts in
Canine Distemper Virus Infection
A thesis submitted for the degree of Doctor of Natural Sciences (Dr.rer.nat.)
in the subject of Virology by
Dipl. Biol. Heidi Imhoff November 2006
International PhD program “Infection Biology”
Institute of Virology
University of Veterinary Medicine Hannover
Acknowledged by the MD/PhD committee and head of Hannover Medical School
President: Prof. Dr. Dieter Bitter-Suermann
Supervisor: Prof. Dr. Ludwig Haas
Cosupervisor: Prof. Dr. Georg Herrler
External expert: Prof. Dr. Hassan Y. Naim
Internal expert: Prof. Dr. Edgar Maiß
Day of final exam/public defense: 16 February 2007
PhD project funded by the Ministry for Science
and Culture of Lower Saxony through the Georg-
Christoph-Lichtenberg Scholarship scheme.
Meinen Eltern
Dat Eumel ist fertig;
Ruhe im Glas!
Contents
Abstract 19
Objectives 21
1 Introduction 23
1.1 Canine distemper virus(CDV) . . . 23
1.1.1 Taxonomy . . . 24
1.1.2 Virus structure and genome organisation . . . 26
1.1.3 Properties of the H, F, and M proteins . . . 27
1.1.4 Cellular receptors for CDV . . . 29
1.1.5 Replication cycle . . . 30
1.1.6 Pathogenesis . . . 31
1.2 Lipid rafts . . . 32
1.2.1 Membrane biology . . . 32
1.2.1.1 Membrane lipids . . . 33
1.2.2 Lipid raft model . . . 34
1.2.3 Interactions of viruses with lipid rafts . . . 38
1.2.3.1 Virus entry . . . 38
1.2.3.2 Assembly and budding . . . 40
1.3 Cell polarity . . . 42
1.3.1 Impact on virus replication . . . 42
1.3.1.1 Entry and release . . . 43
1.3.1.2 Polar transport of proteins . . . 44
1.3.2 Lipid rafts in polarised epithelial cells . . . 45
2 Material 47 2.1 Viruses . . . 47
2.2 Cell lines . . . 48
2.3 Plasmids . . . 49
3 Methods 51 3.1 Cell culture . . . 51
3.1.1 Cell maintenance and passage . . . 51
3.1.2 Isolation of primary canine epithelial cells of the respiratory tract . . . 51
3.1.3 Freezing and thawing of cells . . . 52
3.1.4 Determination of cell numbers . . . 52
3.2 Viruses . . . 52
3.2.1 Propagation . . . 52
3.2.2 Plaque test . . . 53
3.3 Lipid rafts . . . 54
3.3.1 Fixation of cells . . . 54
3.3.2 Immunofluorescence analysis . . . 54
3.3.3 Flow cytometer analysis . . . 54
3.3.4 Depletion of cellular cholesterol using methyl-β-cyclodextrin (MβCD) 55
3.3.5 Determination of cellular cholesterol . . . 55
3.3.6 Determination of cell viability after cholesterol depletion . . . 56
3.3.7 Infection efficiency after cell membrane cholesterol depletion . . . 56
3.3.8 Reverse transcriptase (RT) PCR for SLAM gene detection . . . 56
3.3.9 Infection efficiency of CDV-Onderstepoort and CDV-5804P on Vero or Vero-SLAM cells . . . 57
3.3.10 Infection efficiency after virus envelope cholesterol depletion . . . 57
3.3.11 Replenishment of viral cholesterol . . . 58
3.3.12 Isolation of Detergent Resistant Membranes (DRMs) . . . 58
3.3.13 Sodium dodecylsulphate polyacrylamide gel electrophoresis (SDS-PAGE) 58 3.3.14 Western Blot . . . 59
3.3.15 Syncytia formation after cholesterol depletion . . . 59
3.3.15.1 Quantification of syncytia . . . 60
3.4 Cell polarity . . . 60
3.4.1 Filter supports and transepithelial resistance . . . 60
3.4.2 Polar entry of CDV into different cell lines . . . 61
3.4.3 Polar entry of CDV after depletion of bivalent ions from cells . . . 62
3.5 Cell lines stably expressing an eGFP-tagged protein . . . 62
3.5.1 Preparing heat-shock competentE. coli . . . 62
3.5.2 Plasmid propagation . . . 62
3.5.3 Determination of DNA concentration . . . 63
3.5.4 Establishment of eGFP-tagged CDV proteins . . . 63
3.5.5 Hybridisation of genes or gene fragments by polymerase chain reaction
(PCR) . . . 64
3.5.6 Colony-PCR . . . 65
3.5.7 Sequencing . . . 65
3.5.8 Alignment and other bioinformatics programs . . . 65
3.5.9 Heat shock transformation ofE. coliXL 1 blue . . . 66
3.5.10 Freezing of bacteria . . . 66
3.5.11 Transfection of eukaryotic cells . . . 66
3.5.12 Establishment of cell lines stably expressing a protein . . . 67
3.5.13 Recloning of cells . . . 67
4 Results 69 4.1 Role of cholesterol for initiation of CDV infection . . . 69
4.1.1 Role of cell membrane cholesterol for CDV-Onderstepoort infectivity . 69 4.1.1.1 MβCD-treatment decreases Vero cell cholesterol content . . 69
4.1.1.2 MβCD treatment does not affect Vero cell viability . . . 70
4.1.1.3 CDV-Onderstepoort infection is independent of cell mem- brane cholesterol . . . 70
4.1.2 Role of CDV receptor usage for cholesterol independence . . . 76
4.1.2.1 Vero-SLAM but not Vero cells express SLAM mRNA . . . . 76
4.1.2.2 CDV-5804P uses almost exclusively SLAM as a receptor . . 77
4.1.2.3 CDV-5804P infects Vero-SLAM cells independently of cell membrane cholesterol . . . 78
4.1.3 Role of viral cholesterol for CDV-Onderstepoort infection . . . 80
4.1.3.1 Infection efficiency of cholesterol-depleted CDV-Onderstepoort is decreased . . . 80 4.1.3.2 Replenishment of virus envelope cholesterol restores CDV-
Onderstepoort infectivity . . . 82 4.2 Role of cholesterol for late steps in the CDV replication cycle . . . 84
4.2.1 CDV H and F but not M partially associate with DRMs in infected Vero cells . . . 84 4.2.2 Role of cellular cholesterol for CDV-induced syncytium formation . . . 84
4.2.2.1 Cholesterol starvation of Vero cells for two days does not affect cell viability . . . 85 4.2.2.2 Cholesterol starvation of cells during CDV infection decreases
syncytium formation . . . 87 4.2.2.3 Quantification of syncytia . . . 89 4.2.2.4 Cellular cholesterol starvation decreases syncytium forma-
tion in CDV H and F transfected cells . . . 90 4.3 Entry of CDV into polar cells . . . 90 4.3.1 Entry of CDV into several polar cell lines is inefficient . . . 91 4.3.2 Entry of CDV into MDCK II after cellular depletion of bivalent ions is
inefficient . . . 91 4.3.3 Apical entry of CDV in primary canine epithelial cells of the respiratory
tract . . . 93
5 Discussion 95
5.1 Role of cholesterol for initiation of CDV infection . . . 95 5.2 Role of virus envelope cholesterol for cell entry . . . 97 5.3 Role of cholesterol for late steps in the CDV replication cycle . . . 100
5.4 Entry of CDV into polar cells . . . 100
5.5 Outlook . . . 102
References 103 A Appendix: Tools for investigations of CDV proteins i A.1 Establishment of eGFP-tagged CDV proteins . . . i
A.2 Verification of correct subcellular localisation of CDV fusion proteins . . . i
A.3 Establishment of Calu-3 cell lines stably expressing CDV fusion proteins . . . iv
B Appendix: Media, buffers, solutions and reagents v B.1 Cell culture . . . v
B.1.1 Media . . . v
B.1.1.1 Eagles Minimum Essential Medium (EMEM) . . . v
B.1.1.2 EMEM, modified according to Dulbecco (Edulb) . . . vi
B.1.1.3 Incubation-medium . . . vi
B.1.1.4 Digestion-medium . . . vi
B.1.1.5 Methylcellulose . . . vi
B.1.1.6 Cell freezing medium . . . vii
B.1.1.7 Sera . . . vii
B.1.2 Buffers . . . vii
B.1.2.1 Phosphate buffered saline (PBS) . . . vii
B.1.2.2 Hepes buffer (20x) . . . viii
B.1.2.3 Lysis buffer . . . viii
B.1.3 Solutions . . . viii
B.1.3.1 Versen-Trypsin solution (0,125 %) . . . viii
B.1.3.2 Ethylendiaminetetraacetate (EDTA)-solution . . . ix
B.1.3.3 Triton-X-100-solution . . . ix
B.1.3.4 Paraformaldehyde (PFA)-solution . . . ix
B.1.3.5 Mowiol mounting solution . . . ix
B.1.3.6 Solutions for flow cytometry . . . x
B.1.4 Reagents . . . x
B.1.4.1 Antibodies/-sera and secondary reagents . . . x
B.1.4.2 Other reagents . . . xi
B.2 Molecular biology work . . . xii
B.2.1 Media . . . xii
B.2.2 Buffers for competent bacteria . . . xii
B.2.2.1 TFB 1 . . . xii
B.2.2.2 TFB 2 . . . xii
B.2.3 Buffers for agarose gel electrophoresis . . . xiii
B.2.3.1 TBE-buffer, 10 x . . . xiii
B.2.3.2 TAE-buffer, 10 x . . . xiii
B.2.3.3 Loading buffer, 5 x . . . xiii
B.2.3.4 Gel staining bath . . . xiii
B.2.4 Kits . . . xiv
B.2.5 PCR-reagents and polymerases . . . xiv
B.2.6 Enzymes . . . xv
B.2.7 Other reagents . . . xv
B.3 Biochemical work . . . xv
B.3.1 Buffers and solutions for isolation of detergent resistant membranes (DRM) . . . xv
B.3.1.1 Sucrose-solutions . . . xv
B.3.1.2 MB buffer and MB++ buffer . . . xvi
B.3.2 Buffers and solutions for SDS-PAGE . . . xvi
B.3.2.1 Sample loading buffer, reducing, 2 x . . . xvi
B.3.2.2 Ammoniumpersulphate-solution (APS) . . . xvii
B.3.2.3 Collecting gel . . . xvii
B.3.2.4 Separating gel . . . xvii
B.3.2.5 Electrode buffer, 10 x . . . xviii
B.3.3 Buffers and solutions for Western Blot . . . xviii
B.3.3.1 Anode-buffer I, pH 9.0 . . . xviii
B.3.3.2 Anode-buffer II, pH 7.4 . . . xviii
B.3.3.3 Kathode-buffer, pH 9.0 . . . xviii
B.3.3.4 PBSM-Tween . . . xix
B.3.3.5 Blocking buffer . . . xix
B.3.3.6 Antibodies and reagents . . . xix
C Appendix: Protocols for molecular biology work xx C.1 Digestion assays . . . xx
C.1.1 Digestion with Kpn I and Apa I . . . xx
C.1.2 Digestion with BamH I and Pme I . . . xxi
C.2 Dephosphorylation . . . xxi
C.3 Ligation . . . xxi
C.4 Protocol for cycle ligation . . . xxii
C.5 Primers . . . xxii
C.5.1 Primers for SLAM and GAPDH amplification . . . xxii
C.5.2 Primers for sequencing . . . xxiii
C.5.3 Primers for hybridisation . . . xxiv
C.5.3.1 Hybridisation of CDV-F with eGFP at its C-terminus . . . . xxiv
C.5.3.2 Hybridisation of CDV-M with eGFP at its C-terminus . . . . xxv
C.5.3.3 Hybridisation of CDV-M with eGFP at its N-terminus . . . . xxv
C.5.3.4 Hybridisation of CDV-H with eGFP at its C-terminus . . . . xxvi
C.5.4 Hybridisation of CDV-H with eGFP following the transmembrane domainxxvi C.6 PCR-Master-Mixes . . . xxvii
C.6.1 Reverse transcription of SLAM and GAPDH mRNA . . . xxvii
C.6.2 Amplification of SLAM and GAPDH cDNA . . . xxvii
C.6.3 Amplification of CDV proteins and hybridisation products . . . xxviii
C.6.4 Colony-PCR . . . xxviii
C.7 PCR-protocols . . . xxix
C.7.1 Reverse transcription of cellular mRNA . . . xxix
C.7.2 Amplification of SLAM and GAPDH mRNA . . . xxix
C.7.3 Amplification of CDV-H, -F, -M, eGFP and hybridisation-products . . . xxx
C.7.4 Hybridisation of CDV-H, -F or -M with eGFP . . . xxx
C.7.5 Colony-PCR . . . xxxi
D Appendix: Material xxxii
D.1 General lab ware . . . xxxii D.2 Technical Equipment . . . xxxiii
List of Figures xxxv
List of Tables xliii
Publications and Presentations xliv
Curriculum Vitae xlv
Declaration xlvii
Acknowledgement xlix
List of abbreviations
BSA Bovine serum albumin
CD Cluster of differenciation
CDV Canine distemper virus
CIAP Calf intestine alkaline phosphatase
DEPC Diethylpyrocarbonate
dpi Days after infection (days post infection) DRM Detergent resistant membranes
D/R-water DNase/RNase free water
dsDNA Double stranded deoxyribonucleic acid
EB Elution buffer
E. coli Escherichia coli
EDTA Ethylenediaminetetraacetate
ER Endoplasmic reticulum
FRET Fluorescence resonsance energy transfer hpi Hours after infection (hours post infection)
HRP Horse radish peroxidase
IBV Infectious bronchitis virus
LB Luria Bertani
MβCD Methyl-β-cyclodextrin
MOI Multiplicity of infection
MV Measles virus
NDV Newcastle disease virus
o/n Overnight
PCR Polymerase chain reaction
17
18 List of abbreviations
PFA Paraformaldehyde
PI Propidium iodide
RdRp RNA-dependent RNA-polymerase
RNP Ribonucleoprotein
rpm Rounds per minute
RT Room temperature
SDS-PAGE Sodium dodecylsulphate polyacrylamid gel electrophoresis SLAM Signaling lymphocytic activating molecule
ssDNA Single stranded deoxyribonucleic acid SSPE Subacute sclerosing panencephelitis
TAE Tris acetate EDTA
TBE Tris borate EDTA
TER Transepithelial resistance VSV Vesicular stomatitis virus
Abstract
Cholesterol is known to play an important role in stabilising particular cellular membrane struc- tures, so-called lipid rafts. For several viruses a dependence on cholesterol for virus entry and/or morphogenesis has been shown. Using flow cytometry and fluorescence microscopy we demonstrate thatCanine distemper virus(CDV) infection of cells was not impaired after cellu- lar cholesterol had been depleted by the drug methyl-β-cyclodextrin (MβCD). This effect was independent of the multiplicity of infection (MOI) and the cellular receptor used for infection.
However, cholesterol depletion of the viral envelope significantly reduced CDV infectivity. Re- plenishment by addition of exogenous cholesterol restored infectivity up to 80 percent. Thus, we conclude that CDV entry is dependent on cholesterol in the viral envelope. Furthermore, a reduced syncytium formation was observed when the cells were cholesterol depleted during the course of the infection. This may be related to the observation that the CDV envelope proteins H and F partitioned into cellular detergent resistant membranes (DRMs). Therefore, a role for lipid rafts during virus assembly and release as well is suggested.
19
Objectives
Canine distemper virus(CDV) belongs to the familyParamyxoviridaewithin the orderMonone- gavirales and is classified within the genusMorbillivirus. This genus includes several closely related viruses, among themMeasles virus(MV). Two virus-encoded glycoproteins, the haemag- glutinin (H) and the fusion (F) protein, are integrated into a lipid envelope which surrounds the ribonucleoprotein core. The tetrameric H protein mediates adsorption to the cell by interaction with the cellular receptor. Upon receptor binding, a conformational change in the F protein ini- tiates the fusion of the viral and plasma membranes. CDV is transmitted by aerosoles. It infects epithelial cells at the late stage of infection for virus shedding with excretions and it may also infect epithelial cells during primary contact in the respiratory tract for host invasion.
The raft hypothesis claims that lipids and proteins are not distributed randomly in the plasma membrane but that sphingolipids and cholesterol form microdomains with high structural order.
By compartmentalising membranes these so-called lipid rafts can exert different functions, e.g.
concentrating proteins which is important during signal transduction cascades, targeted protein transport, or endocytosis. Cholesterol is known to be a critical component of rafts and choles- terol depletion has been shown to perturb many raft functions. For several viruses an essential role of cholesterol has been demonstrated at different stages of their replication cycle. Virus entry may be one critical step requiring cholesterol. In the case of enveloped viruses choles- terol may be required in the cellular membrane, in the virus envelope, in both or entry may be cholesterol independent. For many enveloped viruses an incorporation of several virus proteins into lipid rafts has been shown.
This work was aimed at investigating a possible role of lipid rafts for the CDV replication cycle. There were no published data describing a function of cellular lipid rafts for paramyxo- virus entry. An involvement of lipid rafts for paramyxovirus assembly has been shown for
21
22 Objectives
some paramyxoviruses other than CDV. Functional analyses should elucidate the importance of cholesterol in the cellular membrane as well as in the virus envelope for CDV entry and syncytium formation. Furthermore, the incorporation of CDV envelope proteins into lipid rafts should be investigated. In order to clarify the infection route of CDV, virus entry into polarised epithelial cells should be analysed.
1. Introduction
The family ofParamyxoviridaecomprises the etiological agents of many biologically and eco- nomically important diseases of humans and animals. Canine distemper is a long known disease in dogs and its causative agent, theCanine distemper virus (CDV) is clinically and immuno- logically similar toMeasles virus(MV). But in contrast to MV, which infects only humans and other primates, CDV has a broad host range infecting many species within the orderCarnivora.
However, the severity of CDV infection varies significantly among the different species. While CDV only has a low pathogenicity for domestic cats (Ikeda et al., 2001), infection of ferrets is almost always lethal (von Messling et al., 2003). CDV infection of dogs causes high morbidity, with mortality rates of approximately 50 % (Appel, 1969). Because of the frequent persistent infection in recovered dogs, resulting in a demyelinating disease, CDV infection is considered an animal model for multiple sclerosis in humans (Vandevelde and Zurbriggen, 2005). Since CDV disease symptoms in ferrets strongly resemble MV symptoms in humans, like rash, lym- phopenia, and immunosuppression, CDV infection of ferrets has recently been established as a model system for MV infection (von Messling et al., 2003).
1.1 Canine distemper virus (CDV)
Infections by CDV have caused several epidemics among different species within the last decades. In 1987/88 there was a devastating outbreak among freshwater seals (Phoca sibirica) in Lake Baikal in Russia and an estimated 10 000 Caspian seals (Phoca caspica) succumbed to a CDV infection at the shores of the Caspian Sea in Kazakhastan in April-May 2000 (Kennedy et al., 2000). In 1992, several species of large cats died in American zoos and in 1994, lions in Serengeti National Park, Tanzania, died in large numbers and in both cases CDV crossing
23
24 1. Introduction
the species barrier was identified as the cause of infection (Harder et al., 1995, 1996). These outbreaks were unusual in that CDV infection inFelidaewas believed not to cause clinical dis- ease. Sporadic outbreaks in dogs occured worldwide, among them 1989 in an animal shelter in Northern Germany that killed most of the dogs and three outbreaks in Finland in 1990, 1994 and 1995 that killed at least 5 000 animals (Ek-Kommonen et al., 1997).
While CDV outbreaks in domestic dogs can be minimised by vaccination, immunisation of other carnivores, many of which are kept in captivity, is still problematic. Live attenuated vaccines are not safe for all animal species and the effectiveness as well as the duration of immunity of recombinant vaccines based on avipox, canarypox, or vaccinia virus expressing the CDV H and F genes have been reported to be limited (Welter et al., 2000; Barrett, 1999).
CDV, like other morbilliviruses, does not persist in an infectious form and infection results in life-long immunity of the recovered host animal. Therefore, morbilliviruses rely on a large susceptible population for maintenance, which has been estimated to be at least 300 000 individ- uals for MV (Barrett, 1999). While MV is targeted by an eradication attempt (Griffin and Moss, 2006), there is little hope that a similar programme might be successful for CDV. Eradication of this virus is hampered by the wide host range, the numerous occurences of its hosts, and their worldwide distribution. This makes it difficult, if not impossible, to vaccinate enough animals to keep the number of susceptible animals below the critical number required for maintenance of the virus.
1.1.1 Taxonomy
CDV belongs to the order Mononegavirales. This designation implicates that these viruses contain a genome of unsegmented, single stranded RNA (ssRNA) in negative orientation. Fam- ilies within this order are theRhabdoviridae, Filoviridae, BornaviridaeandParamyxoviridae.
The classification within the familiy Paramyxoviridae is based on morphologic criteria, the organisation of the genome, the biologic activities of the proteins and the sequence relation- ship of the encoded proteins. Subdivision into two subfamilies, Paramyxovirinae and Pneu- movirinaeis based on differences in the second envelope glycoprotein and in the number of encoded genes. CDV belongs to the subfamily Paramyxovirinae, with a genome consisting of six genes and an attachment protein possessing haemagglutination, but not neuraminidase,
1.1. Canine distemper virus(CDV) 25
activity (Lamb and Kolakofsky, 2001). Within the subfamily Paramyxovirinaethere are five genera, Respirovirus, Rubulavirus, Morbillivirus and the newly established generaAvulavirus and Henipavirus (Lwamba et al., 2005). Viruses within the genera are differentiated by the size and shape of their nucleocapsid, antigenic cross-reactivity, the presence or absence of neu- raminidase activity and the differential coding potential of the P gene. CDV is most closely related toPhocine distemper virus;Dolphin morbillivirusis closely related toPorpoise morbil- livirus. Together with MV, which is most closely related to theRinderpest virus, andPeste-des- petits-ruminant virusthey are all classified as the genusMorbillivirus(Barrett, 1999). Table 1.1 gives an overview of the familyParamyxoviridae.
Table 1.1:Taxonomy ofParamyxoviridae(adapted from Lwamba et al. (2005)).
Subfamily Genus Virus, example
Paramyxovirinae Avulavirus Newcastle disease virus Avian paramyxoviruses Rubulavirus Mumps virus
Human parainfluenza virus type 2 Morbillivirus Measles virus
Canine distemper virus Phocine distemper virus Dolphin morbillivirus Porpoise morbillivirus Rinderpest virus
Peste-des-petits-ruminants-virus Henipavirus Hendravirus
Nipahvirus Respirovirus Sendai virus
Human parainfluenza virus type 1andtype 3 Bovine parainfluenza virus type 3
Pneumovirinae Pneumovirus Human respiratory syncytial virus Bovine respiratory syncytial virus Metapneumovirus human Metapneumoviruses
Avian metapneumovirus
26 1. Introduction
1.1.2 Virus structure and genome organisation
CDV is an enveloped virus. The overall virus structure is spherical with a size of approximately 150 - 350 nm. The nucleocapsid core inside of the envelope is helical and has a length of 1µm and a diameter of 18 nm (Lamb and Kolakofsky, 2001). Figure 1.1 shows a schematical drawing of the virion structure.
Haemagglutinin protein (H) Fusion protein (F)
Matrix protein (M) Large protein (L) Phosphoprotein (P)
Nucleocapsid protein (N) Single stranded RNA
Figure 1.1:Schematical drawing of aCanine distempervirion.
The CDV nucleocapsid core consists of ssRNA in antisense (negative) orientation encapsidated by the nucleocapsid (N) protein, which protects the RNA from degradation. One N protein en- closes six nucleotides, resulting in incomplete encapsidation of genomes with non-hexameric lengths which probably for that reason are replicated only inefficiently (“rule of six”) (Kolakof- sky et al., 2005). Associated with the N protein is the phosphoprotein (P), which acts as a cofactor for the L (large) protein, the RNA-dependent RNA-polymerase (RdRp), which tran- scribes the viral genes and adds the cap structure and the poly A tail to the viral mRNAs and also replicates the genome. The ribonucleoprotein (RNP), made up of the RNA encapsidated by the N, P and L proteins, never disassembles during the infectious cycle (Lamb and Kolakofsky, 2001).
Two virally encoded proteins are integrated into the envelope membrane. These proteins are the haemagglutinin (H) protein, which mediates attachment, and the fusion (F) protein, which mediates membrane fusion (Lamb et al., 2006). The matrix (M) protein is located at the inner
1.1. Canine distemper virus(CDV) 27
surface of the envelope and interacts with the cytoplasmic tails of the envelope glycoproteins as well as with the N protein, thus being a critical factor regulating assembly (Takimoto and Portner, 2004) (for more details to H, F and M see 1.1.3). The nonstructural proteins C and V interact with the host innate immune system thereby facilitating the establishment of CDV infection (von Messling et al., 2006).
The genome (CDV strain Onderstepoort) has a length of 15.69 kb (NCBI AF378705) and con- sists of a leader sequence at its 3’-end, a trailer sequence at its 5’-end and of six genes in between. The genes are separated by a conserved end sequence, a highly conserved intergenic region and a conserved start sequence, which are necessary for transcription. CDV shows the typical morbillivirus gene order, which is 3’-N-P-M-F-H-L-5’. In the genusMorbillivirus, the P gene contains two additional open reading frames, which give rise to the nonstructural proteins C and V. As a result of transcriptional regulation (described in 1.1.5), the vicinity of a gene to the 3’-end of the genome determines the efficiency of gene transcription and thereby the amount of protein expressed.
1.1.3 Properties of the H, F, and M proteins
The haemagglutinin protein (H) mediates virus attachment to the cell surface by binding to cel- lular receptor(s). H is a type II integral membrane protein, with a short hydrophobic domain near the N-terminus, which acts as a combined signal and anchorage domain (Lamb and Kolakofsky, 2001). The apparent molecular mass of the monomeric H protein is between 76 and 85 kDa, de- pending on different numbers of N-linked oligosaccharide side chains in different CDV strains (Rima, 1983; Iwatsuki et al., 1997; Harder et al., 1996; Cherpillod et al., 1999). The glyco- sylation pattern may also contribute to differences in antigenicity (Iwatsuki et al., 1997). The H protein forms disulfide-linked homodimers, two of which form a tetramer by noncovalent binding. A three-dimensional model based on the crystal structure of the haemagglutinin of Newcastle disease virus (NDV) suggests a globular ectodomain for the CDV H protein (von Messling et al., 2005).
The H protein is a major determinant of CDV tropism (Stern et al., 1995; von Messling et al., 2001). Two clusters have been found which are responsible for receptor recognition (von Messling et al., 2005) (for more details to receptor usage see 1.1.4). The H protein also affects fusion efficiency, and thus contributes to the cytopathogenicity (von Messling et al., 2001).
28 1. Introduction
The fusion protein (F) mediates membrane fusion. F is a type I integral membrane protein, with the transmembrane domain located near the C-terminus, spanning the membrane once. It contains a cleavable signal sequence at the N-terminus, which is preceeded by an unsusually long sequence (Cherpillod et al., 1999), termed Pre peptide (von Messling and Cattaneo, 2002).
The efficiency of removal of the signal peptide depends on the length of the translated Pre pep- tide and has been linked to a regulatory function determining F protein activity (von Messling and Cattaneo, 2002). The protein is synthesised as an inactive precursor (F0), which has to be cleaved posttranslationally into an N-terminal, smaller fragment, F2, and a C-terminal, larger fragment, F1. Cleavage of the F0 protein by the cellular protease furin exposes a hydrophobic region at the F1N-terminus, which is designated the fusion peptide (figure 1.2). The active F is a trimer of disulfide-linked F1and F2 heterodimers. The F protein is a typical type I fusion pro- tein (Lamb et al., 2006). The monomeric F1/F2-heterodimer has an apparent molecular mass of 55 kDa while it is 61 kDa for the F0-precursor (Rima, 1983). A three-dimensional model of the NDV F protein suggests the trimeric form to have a stalk, a neck and a globular head domain.
The F protein is the major antigen against which neutralising antibodies are directed (Norrby et al., 1996). Variability among F proteins from different CDV strains is less pronounced than among the H proteins (Iwatsuki et al., 1998; Liermann et al., 1998), probably because of their conserved function for protein folding, transport and fusion activity (von Messling and Catta- neo, 2003).
TM FP
SP
Pre F2 F1
F0
Furin cleavage SP
cleavage
NH2 COOH
Figure 1.2: Schematic drawing of the F protein translation product. SP: Signal peptide, FP:
Fusion peptide, TM: Transmembrane domain. Stars represent N-linked glycosylation sites (dark grey: used; white: potential (von Messling and Cattaneo, 2003)).
The matrix protein (M) is considered to be the central organiser of viral morphogenesis, being able to interact with the nucleocapsid core via the N protein as well as with the envelope gly- coproteins via their cytoplasmic tails (for a review see Takimoto and Portner (2004)). The M protein is quite hydrophobic, but there are no domains of sufficient length to span the mem-
1.1. Canine distemper virus(CDV) 29
brane. Nevertheless, it has been found that M associates with membranes. The monomeric M has an apparent molecular mass of 34 kDa (Rima, 1983).
Many investigations trying to elucidate the role of the M protein have been performed for MV in order to understand virus persistence associated with subacute sclerosing panencephalitis (SSPE), a fatal degenerative neurological disease. It has been found that virus particle produc- tion in M-less MV is defective but that M deficient MV spreads more efficiently by cell-to-cell fusion (Cathomen et al., 1998a). Interactions of M with the cytoplasmic tails of the envelope glycoproteins appear to downregulate fusion efficiency but to upregulate virus envelope gly- coprotein incorporation into virions during assembly (Cathomen et al., 1998b). In polarised epithelial cells, these effects could be attributed to a retargeting of the envelope proteins from the basolateral side, where these proteins mediate cell-to-cell fusion, to the apical side, where virus budding occurs (Maisner et al., 1998; Naim et al., 2000; Moll et al., 2004) (the role for the M protein in polarised cells is described in more detail in 1.3.1.2). It has also been shown that in SSPE-causing virus strains the M protein has lost its ability to efficiently bind to the N protein (Hirano et al., 1993), which may also be responsible for defective virus assembly and release as well as for increased fusion activity.
1.1.4 Cellular receptors for CDV
The only known CDV receptor is the signaling lymphocytic activating molecule (SLAM) (Tat- suo et al., 2001), which is mainly used by wild-type CDV strains. The vaccine strain Onder- stepoort has been adapted to cell lines not expressing SLAM. Therefore, molecules other than SLAM can act as CDV receptor. SLAM expression on activated T cells, immature thymocytes, memory T cells, a proportion of B cells, activated monocytes and dendritic cells (Cocks et al., 1995) explains the CDV lymphotropism (von Messling et al., 2004). This tropism has been at- tributed to the H protein rather than the F protein (Stern et al., 1995; von Messling et al., 2001, 2005). It has been shown that CD46, the receptor for attenuated MV (Naniche et al., 1993), is downregulated by MV andRinderpest virus, but not by CDV or Dolphin morbillivirus, sug- gesting that it cannot act as receptor for the latter viruses (Galbraith et al., 1998). CD9 has been discussed to be a receptor for CDV (L¨offler et al., 1997), but it has been shown later that only cell-to-cell fusion, not virus-to-cell fusion is inhibited by antibodies recognising CD9 (Schmid et al., 2000).
30 1. Introduction
1.1.5 Replication cycle
CDV binds to a cellular receptor via its attachment protein H. Successful fusion can not be mediated by the F protein alone; both proteins have to interact. Not only interaction between homologous H/F pairs but also between heterologous pairs from CDV and MV may lead to membrane fusion (Stern et al., 1995; von Messling et al., 2001). The molecular basis of the fusion process is not fully understood, but it is thought that a conformational change of the H protein upon receptor binding mediates a conformational change in the F protein, which results in the exposure of the fusion peptide (Lamb et al., 2006). Following virus-to-cell fusion, the nucleocapsid core is released into the cytoplasm. The mechanism of uncoating of the M protein is unclear.
All further steps of CDV replication take place in the cytoplasm. The RdRp transcribes the viral genes and replicates the genome. The paramyxovirus genome has only one promoter at its 3’- end, where initial RdRp binding occurs. The RdRp polyadenylates the mRNA by reiteratively copying some uridylylates, which are part of each genes’ end sequence. To terminate the newly produced mRNA, the RdRp has to terminate transcription. As reinitiation of the RdRp at the next gene is not perfect but binding of the RdRp can only occur at the genomes 3’-end, there is a gradient of mRNA abundance (Lamb and Kolakofsky, 2001).
Since the RdRp also replicates the genome, the switch from transcription of single genes to replication of the whole genome must be regulated. The determining factor for switching from transcription to replication is the concentration of free N protein. Accumulation of free N protein in the cytoplasm enables the RdRp to ignore the stop sequences at the gene’s ends, thereby switching from mRNA to antigenome production. Newly synthesised antigenomes are, like the genomes, immediatelly encapsidated by free N proteins, resulting in an equilibrium between transcription and replication.
Translation of the envelope proteins takes place at the rough endoplasmic reticulum (ER). They are translocated into the lumen of the ER and modified during transport via the Golgi apparatus to the plasma membrane. All other virus proteins are translated at ribosomes in the cytoplasm.
The N protein associates with newly synthesised viral RNA as well as with the P protein, which in turn associates with the L protein.
1.1. Canine distemper virus(CDV) 31
CDV assembly and budding take place at the cellular plasma membrane (Takimoto and Portner, 2004). The M protein appears to play a major role as assembly organiser. It interacts with the nucleocapsid core via the N protein and with the envelope protein via their cytoplasmic tails, thereby facilitating the recruitment of all virus components at the site of assembly. The role of lipid rafts as well as of cell polarity during virus assembly and budding are discussed in chapters 1.2 or 1.3, respectively.
1.1.6 Pathogenesis
CDV is highly contagious and is transmitted by aerosoles. Primary replication takes place in lymphoid tissues, that is in the bronchial lymph nodes and in the tonsils (Appel, 1969).
The role of the epithelial cells in the respiratory tract during primary infection is still unclear.
Two to three days post virus exposure, virus is found in the blood in mononuclear cells. This first viremia is accompanied by lymphopenia and unspecific symptoms, like fever or anorexia.
Replication at the first six days post-infection occurs only in the lymphatic system, explaining lymphopenia and immunosuppression. Approximately 50 % of infected dogs mount a success- ful immune response with high antibody titers nine days post infection, thereafter recovering in most cases. In dogs wich did not mount an appropriate immune response, infected macrophages and lymphocytes transport the virus to epithelial tissues especially of the respiratory, gastroin- testinal and urogenital tract six to nine days post virus exposure. After massive invasion of epithelial tissues, these dogs show a second viremia accompanied with fever and more severe symptoms, like diarrhea, conjunctivitis, and in some cases terminal convulsions. Many dogs succumb two to four weeks after infection. Some dogs mount a late immune response. Such dogs either succumb to a subacute disease, often encephalitis, or become persistently infected (Appel, 1969, 1987).
CDV infection of ferrets resembles that of dogs, but is almost always fatal and disease progres- sion is faster. Infected ferrets show a first viremia three to six days post-infection, accompanied with lymphopenia, transient fever and infection of all lymphatic tissues. Already within the next few days, epithelial cells are infected following a second viremia which marks the onset of the symptomatic phase. Symptoms in ferrets are similar to those seen in dogs and they frequently show a rash similar to measles. Gastrointestinal and respiratory symptoms may follow and sec- ondary infections may worsen the progress of the disease. Animals succumb two to three weeks post-infection (von Messling et al., 2003, 2004).
32 1. Introduction
1.2 Lipid rafts
Lipid rafts are microdomains within the cellular membrane which have a special lipid and pro- tein composition. These domains are dynamic in size as well as in their composition and they exert their functions by compartmentalisation. Their name originates from the ideas that inter- actions between lipids are the force keeping the domains together and that they “float” within the membrane just as a raft in the water. Actually the name is under discussion as it becomes more and more obvious that not only interactions between lipids but also those between proteins and between proteins and lipids regulate lipid rafts. Therefore, recently it has been proposed to rename these domains to “membrane rafts” (Pike, 2006). But as this name has not yet been adopted in the literature, membrane domains will be designated lipid rafts or just rafts.
1.2.1 Membrane biology
Singer and Nicolson proposed in the 1970’s (Singer and Nicolson, 1972) a new model for bio- logical membranes, which took advantage of Langmuir’s theories (Langmuir, 1917) concern- ing lipid-water interactions and of Gorter und Grendel’s theories (Gorter and Grendel, 1925) proposing membranes to be composed of lipid bilayers. According to the Singer-Nicolson model, lipids are arranged in a lipid bilayer with proteins being interspersed within the bilayer, either spanning the membrane or being associated with it peripherally (Singer and Nicolson, 1972). The model was also designated fluid-mosaic-model due to the realisation that lipids and proteins appeared to be able to freely float within the whole membrane (fluidity) and due to the mosaic-like, random distribution of the proteins within the membrane. Already a few years later, Jain and White (1977) showed in model membranes that the lateral mobility was not unre- stricted, favouring a domain model of the membrane. But only in the 1990’s the domain model became more and more accepted and it became evident that membrane domains have various cell functions (Simons and Ikonen, 1997) (for a review on the history of membran research see Edidin (2003a)).
1.2. Lipid rafts 33
1.2.1.1 Membrane lipids
Membrane lipids are amphiphatic lipids, with a lipophilic “tail” and a hydrophilic “head”. There are three different groups of membrane lipids, these are glycerophospholipids, sphingolipids and sterols (figure 1.3).
Membrane lipids
Fatty acid
Glycerin
Alcohol PO4
Fatty acid
Sphingosine
Fatty acid Alcohol
PO4 Sphingosine
Fatty acid Sugar Glycolipids
Steroids Glycerophospholipids Sphingolipids
Phospholipids
Figure 1.3: Schematical characterisation and classification of membrane lipids. Modified from Lehninger et al. (1998).
Glycerophospholipids are characterised by a glycerol backbone. One hydroxyl-group is es- terified with a phosphate and an alcohol, making up the hydrophilic “head”, while the other hydroxyl groups are esterified with fatty acids, making up the lipophilic “tails”. In most glycero- phospholipids, at least one acyl side chain is unsaturated (Lehninger et al., 1998).
Sphingolipids have a ceramide backbone, consisting of a sphingosin molecule connected to a fatty acid via an amide bond. Depending on the hydrophilic “head” which is connected to the ceramide they are classified as sphingomyeline, neutral glycolipids or gangliosides. The polar
“head” of sphingomyelines is a posphocholine or a phosphoethanolamine. Together with the glycerophospholipids they are also classified as phospholipids and they resemble each other in their properties and their three-dimensional structure. The free hydroxyl group of the ceramide
34 1. Introduction
molecule in neutral glycolipids is connected to one or more sugar molecules. Gangliosides have a complex hydrophilic “head” consisting of several sugar units. Terminal sugars are often N- acetylneuraminic acid (sialic acid), which is charged at pH 7. Due to the sugar residues in their
“head” neutral glycolipids and gangliosides are also classified as glycosphingolipids. For most sphingolipids both acyl side chains are saturated which therefore have a long structure. This is in contrast to the mostly unsaturated acyl chains of glycerophospholipids with their kinked structure (Lehninger et al., 1998).
Steroids are composed of four condensed rings of C-atoms, three of the rings consisting of six C-atoms and one of five. They form a planar and rigid structure, which does not have any ro- tational freedom. Cholesterol is the most important steroid of eukaryotic cells. A polar head group connected to the ring structure makes it an amphipathic molecule (Lehninger et al., 1998).
The composition of different cellular membranes is very inhomogeneous. Cholesterol, syn- thesised in the ER, and sphingolipids, synthesised in the Golgi apparatus, are only present at low concentrations in inner membranes but at high concentrations in the plasma mem- brane and endosomes. It has been found that even the lipid compositon of the inner and the outer leaflet of the plasma membranes is profoundly different. Most, if not all, sphingolipids are present in the outer leaflet, while some glycerophospholipids, like phosphatidylinositol, phosphatidylethanolamine and phosphatidylserine, are restricted to the inner leaflet (Bretscher, 1973). The distribution of cholesterol has been difficult to determine, as cholesterol readily switches between the two leaflets. It has been shown that cholesterol is abundantly present in the plasma membrane, making up 30 - 40 mol % of all lipids in eukaryotic plasma membranes.
Cholesterol is known to “thicken” the membrane by forcing the acyl side chains of neighbour- ing lipids into the straighttransconformation. This restriction in deformation of lipids results in lower permeability (for a review on the role of cholesterol within cells see Simons and Ikonen (2000)).
1.2.2 Lipid raft model
The raft hypothesis, as it was proposed by Simons and Ikonen (1997), claims that lipids and proteins are not distributed randomly within membranes, but that interactions between differ- ent lipids result in domain formation. Main constituents of raft domains are sphingolipids and cholesterol. The model proposes that sphingolipids associate laterally with one another and
1.2. Lipid rafts 35
are kept together by weak interactions between the carbohydrate heads of glycosphingolipids.
Cholesterol stabilises these interactions by filling the spaces within the hydrophobic plane of the membrane, which results from glycosphingolipid carbohydrate heads occupying a larger area in the plane than their mainly saturated fatty acids. These interactions result in an assembly with closely packed lipids, which “floats” in the membrane consisting of loosely packed glycero- phospholipids with mainly unsaturated, kinked fatty acids. The originally proposed function of lipid rafts was in the apical delivery of proteins in polarised epithelial cells. This function was thought to be fullfilled by specific inclusion or exclusion of proteins from these domains. Gly- cosylphosphatidylinositol (GPI)-anchored proteins, doubly acylated proteins of the Src-family as well as some transmembrane proteins were known to preferentially associate with lipid rafts.
Much of the raft hypothesis is based on studies with model membranes, which consist of lipid bilayers with known chemical composition. Depending on the lipid composition, membranes undergo phase transition at a specific temperature. At low temperatures the membrane is in a solid ordered (So), or gel phase, which is characterised by a highly ordered structure with acyl side chains in the straighttransconformation, tightly packed which results in a low lateral mobility of the lipids. As temperature increases, the ordered phase switches to a liquid dis- ordered (Ld) phase where acyl side chains are in the kinkedcis conformation and lipids have high lateral mobility. Sufficiently high cholesterol concentrations can promote a third phase, which is characterised by highly ordered acyl side chains resembling of the So phase, but also by a high lateral mobility of the lipids similar to that in the Ld phase. This phase is therefore designated liquid ordered (Lo) phase. While the existence of Ldphase and So phases depends on the temperature and both phases cannot co-exist, it has been shown for model membranes with appropriate mixtures of sphingomyelines, unsaturated phospholipids and cholesterol that Ldphases can co-exist with Lo phases. This enables domain formation driven by lipid-lipid in- teractions, as it is hypothesised for lipid rafts. Lipid rafts are postulated to be the cell membrane equivalent to the Lo phase of model membranes. But although the co-existence of different phases within model membranes is commonly accepted, it could not be shown unambiguously that this phase co-existence is also found in cell membranes (for reviews on model membranes refer to Simons and Vaz (2004); Edidin (2003b); London (2005)).
Another basis for the raft hypothesis was the finding that membranes, especially in the presence of cholesterol, are not completely solubilised by anionic detergents (mostly Triton-X-100) at 4◦C. Due to their high lipid proportion, these detergent resistant membranes (DRM) float to
36 1. Introduction
low densities in linear sucrose gradients. Proteins may also be associated with DRMs (Brown and Rose, 1992). Those detergent insoluble membranes have been found to be enriched in gly- cosphingolipids and cholesterol (therefore an alternative name for DRM is detergent-insoluble glycolipid-enriched complexes (DIG)) (Simons and Ikonen, 1997), and extraction of cholesterol prior to solubilisation resulted in solubilsation of otherwise DRM-associated proteins (Edidin, 2003b). The lipid composition and the dependence on cholesterol for detergent resistance led to the assumption that DRMs are isolated lipid rafts. This conclusion has been a matter of debate (Lichtenberg et al., 2005; Munro, 2003). Nevertheless, DRM association of proteins is still a commonly used method to show lipid raft association of proteins.
Based on the raft concept a lot of experiments have been made investigating the detergent resis- tance of many proteins and thus a potential association of proteins with rafts. Functional assays determined how cellular processes are perturbed upon cholesterol depletion. Many studies set out to investigate lipid phase behaviour in model membranes. From all these investigations, the original raft model has been modified to a more sophisticated model.
Lipid rafts are now proposed to be highly dynamic assemblies of tightly packed sphingolipids and cholesterol within the exoplasmic leaflet of the membrane and are variable in size (Rajen- dran and Simons, 2005). Small rafts can cluster upon stimuli, thereby forming larger domains, bringing proteins formerly separated in different rafts in close contact and excluding non-raft proteins. Thus, they exert their function by compartmentalising the membrane into different domains (Pike, 2006). Selective interaction of several proteins is facilitated by inclusion or exclusion of proteins (Rajendran and Simons, 2005). In contrast to the original hypothesis the emphasis in current models is no longer on lipid-lipid interactions being the main forces keep- ing together lipid rafts, but that lipid-protein or even protein-protein interactions may be equally important forces (Hancock, 2006; Edidin, 2003b; Munro, 2003). Additionally, it has been hy- pothesised recently that rafts do not pre-exist in a stable form but that they form spontaneously and degrade in a steady state (Hancock, 2006).
The probably most recent definition of rafts was made at the “Keystone Symposium on Lipid Rafts and Cell Function” and is the following: “Membrane rafts are small (10-200 nm), het- erogeneous, highly dynamic, sterol- and sphingolipid-enriched domains that compartmentalize cellular processes. Small rafts can sometimes be stabilized to form larger platforms through protein-protein and protein-lipid interactions.” (Pike, 2006).
1.2. Lipid rafts 37
The originally proposed function of rafts, the apical delivery of proteins, has been greatly ex- panded (Rajendran and Simons, 2005). Besides a role for apical transport of proteins (this is discussed in more detail in 1.3.2), general transport actions are mediated via lipid rafts. The most important function of rafts appears to be regulation within signal transduction. Rafts ap- pear to regulate signal transduction by clustering proteins necessary for a signal transduction cascade and by excluding inhibitors (Simons and Toomre, 2000). Many important complex sig- naling events, such as during T cell signalling, have a raft involvement (Kabouridis, 2006).
Endocytosis can also be mediated via lipid rafts. Besides poorly characterised lipid raft struc- tures which seem to be involved in endocytosis, there are well characterised structures spe- cialised in endocytosis designated caveolae. These structures show all biochemical character- istics of lipid rafts and are additionally enriched in the cholesterol binding protein caveolin-1.
Caveolin-1 reduces caveolae invagination at the plasma membrane, thereby regulating endocy- tosis. Invagination is an active process induced by signaling cascades. Endocytosis via caveolae is cholesterol and dynamin dependent (Nabi and Le, 2003; Smart et al., 1999; Pelkmans, 2005).
Many pathogens also exploit lipid raft functions (reviewed in Manes et al. (2003)). Interactions of viruses with lipid rafts will be described in more detailed in 1.2.3.
Most of the concepts of lipid rafts are based on studies investigating the detergent resistance of proteins, the effect of cholesterol depletion on cellular functions, or on results obtained from model membranes. All these methods are yielding results which cannot be interpretated unam- biguously with respect to the raft hypothesis. Additionally the composition of the inner leaflet of raft domains is still unclear, making adoption of results from model membranes even more critical. In cell membranes, the size or even the mere existence of rafts could still not be ambigu- ously proven, not even with more recent methods such as fluorescence resonance energy transfer (FRET), single particle trafficking, or atomic force microscopy (AFM). Therefore, the raft con- cept is still under debate (Munro, 2003). Nevertheless, the raft model remains a powerful tool to explain many cellular processes and Edidin (2003b) aptly described it: “Despite great reser- vations about the interpretation of “classical” operational definitions of lipid raft components and functions, we are left with cells stubbornly insisting that lipids, lipid-anchored proteins, and acylated cytoplasmic signaling proteins are selectively trafficked and associated, and left with viruses that selectively sample host lipids to enrich their envelopes with sphingolipids and cholesterol.”
38 1. Introduction
1.2.3 Interactions of viruses with lipid rafts
Viruses are dependent on the cellular machinery for many steps in their replication cycle. Sev- eral cellular mechanisms used by viruses either for entry, such as endocytosis or signal trans- duction pathways, or for morphogenesis, such as protein transport can involve lipid rafts. Not surprisingly it has been shown for several viruses that lipid rafts are crucial for their replication cycle (reviewed in Chazal and Gerlier (2003); Manes et al. (2003); Nayak and Barman (2002);
Ono and Freed (2005)).
1.2.3.1 Virus entry
Viruses have evolved different strategies to enter their host cells. All animal viruses that have extracellular phases initiate their replication by binding to one or more cellular receptor(s). In all cases viruses have to cross the cellular membrane to gain access to the cytoplasm. Non- enveloped viruses are usually internalised by endocytosis. Intracellular (endosomal, lysosomal or caveosomal) membranes may be overcome either by membrane lysis or by pore formation.
Enveloped viruses overcome the membrane barrier by fusion of their own lipid envelope with a cellular membrane. Viruses, which are internalised by endocytosis fuse with an internal mem- brane; others fuse at the plasma membrane (Flint et al., 2004).
For non-enveloped viruses a role for lipid rafts is restricted to the cellular membrane, like for coxsackieviruses (Triantafilou and Triantafilou, 2003, 2004). In the case of enveloped viruses, two membranes, cell membrane and viral envelope are involved in the fusion process. Lipid rafts may be required in either membrane, in both or entry may be lipid raft independent.
Murine leukemia virusis sensitive to cholesterol depletion from the cellular, but not from the viral membrane (Lu et al., 2002), whereas the reverse has been demonstrated for influenza virus (Sieczkarski and Whittaker, 2002; Sun and Whittaker, 2003; Takeda et al., 2003). Human im- munodeficiency virus (HIV) andHerpes simplex virus require cholesterol in both membranes (Liao et al., 2003, 2001; Manes et al., 2000; Graham et al., 2003; Lee et al., 2003; Bender et al., 2003). In contrastVesicular stomatitis virus(VSV) appears to be cholesterol independent (Thorp and Gallagher, 2004; Scheiffele et al., 1999; Brown and Lyles, 2003; Pessin and Glaser, 1980).
1.2. Lipid rafts 39
There are several different ways in which lipid rafts are exploited by viruses for their entry.
Perhaps the most direct way is the internalisation by a lipid raft dependent machinery such as via caveolae or by other, yet not well described, lipid raft mediated pathways (Pelkmans, 2005). The non-enveloped virusSimian virus 40(SV40) is probably the best known virus using caveolae as entry portals (Anderson et al., 1996; Stang et al., 1997; Chen and Norkin, 1999).
Some enveloped viruses, likeHuman coronavirus 229Ealso interact with caveolae for efficent internalisation (Nomura et al., 2004). A non-enveloped virus, Coxsackievirus B4, appears to exploit a lipid raft-dependent endocytic route other than caveolae to gain access to the cell (Triantafilou and Triantafilou, 2004).
Lipid rafts can also be used in more indirect ways to support virus entry. They can act as platform to concentrate proteins. Cholesterol, highly concentrated in lipid rafts, can support a special protein conformation. A prominent example is HIV which exploits lipid rafts in many ways. It has been proposed that cell membrane rafts are the regions where the HIV receptor CD4 and its co-receptors CXCR4 or CCR5 cluster upon binding of the HIV envelope protein, thereby meeting the requirement for three to six HIV envelope trimers to bind to several CD4 and four to six CCR5 molecules (Viard et al., 2002; Kuhmann et al., 2000; Manes et al., 2003).
Furthermore it has been shown that cholesterol is necessary for a correct conformation of the HIV co-creceptors CCR5 and CXCR4 (Nguyen and Taub, 2002b,a). For influenza virus which concentrates its envelope protein HA during assembly within lipid rafts it has similarly been discussed that virus envelope lipid rafts concentrate multiple HA trimers thereby facilitating successful fusion during entry (Takeda et al., 2003). The fusion process can also be supported by lipid rafts in a different way, as has been shown forSemliki forest virus, whose fusion peptide directly interacts with lipid rafts (Ahn et al., 2002).
No reports are available describing a role of cellular or envelope cholesterol for the entry process of morbilliviruses. For paramyxoviruses, there is only one recent publication reporting that although NDV assembles at lipid rafts and incorporates lipid rafts into the envelope, it does not require envelope cholesterol for cell entry (Laliberte et al., 2006). The requirement of cellular cholesterol for entry has not been investigated yet for any paramyxovirus.
40 1. Introduction
1.2.3.2 Assembly and budding
Assembly of viruses includes the association of all viral proteins and the genome in an ordered fashion. This process comprises transport and correct targeting of the proteins and concentra- tion of the proteins at the site of assembly. Enveloped viruses have to obtain their envelope membrane either by budding from an internal or from the plasma membrane. Interestingly, in many cases the composition of the virus membrane does not reflect the composition of the cellular membrane from which it is derived. This led to the proposal that virus budding often occurs at microdomains of the cell membrane (Pessin and Glaser, 1980).
Interactions of viral components with membranes during assembly are obvious for enveloped viruses, but not for non-enveloped viruses. Therefore, membrane interactions of virus proteins during assembly have been intensively investigated for enveloped viruses only (Ono and Freed, 2005). Most results concerning enveloped viruses have been obtained by investigating the in- teraction of viral wild type or mutant proteins with DRMs. Only more recent investigations include techniques such as FRET.
Lipid rafts can exert many functions during virus assembly. The interactions of viral proteins with lipid rafts have been best investigated for influenza virus and this virus exploits lipid rafts for a variety of purposes. Influenza viruses have three transmembrane proteins, the haemagglu- tinin (HA) protein, the neuraminidase (NA) protein and an ion-channel forming protein (M2), and one peripherally associated protein, the matrix (M1) protein. It has been shown that HA and NA partition, in a cholesterol dependent manner, into DRMs. Interactions of the HA and NA proteins with DRMs are necessary for a correct polar transport of these proteins to the apical side of epithelial cells (the role for lipid rafts in apical protein delivery is discussed in 1.3.2).
Finally, the high amount of cholesterol within the influenza virus envelope is dependent on the presence of intact HA and NA proteins as tailless mutants incorporate less cholesterol into the virion. The peripheral membrane protein M1 was found to be DRM associated in a timede- pendent manner and this association was also dependent on the expression of HA and NA with intact cytoplasmic tails. For M2, no DRM association was identified, but incorporation of this protein into virions is very low (reviewd in Ono and Freed (2005); Nayak and Barman (2002)).
These data suggest the following mechanism for influenza virus assembly: lipid rafts transport and target the HA and NA proteins to the apical side of epithelial cells. Additionally lipid rafts concentrate these proteins, thereby creating an assembly site. The M1 protein is dragged to
1.2. Lipid rafts 41
this site by interaction with the cytoplasmic tails of the glycoproteins and also directs the ri- bonucleocapsid via interactions to this site. The M2 protein, which is targeted to the apical side in a lipid raft independent manner, is probably mainly excluded from the lipid raft assembly site, making incorporation of this protein rare (reviewed in Ono and Freed (2005); Nayak and Barman (2002)).
Except for the functions exploited by influenza virus, lipid rafts can also be beneficial during the assembly process by facilitating multimerisation of viral proteins. In addition, specific inclusion or exclusion of cellular proteins can be advantageous. For example, HIV specifically includes the cellular decay accelerating factor (DAF or CD59), which protects cells from complement lysis, while CD45 is excluded from the viral particle in spite of its great abundance on the cell surface (Nguyen and Hildreth, 2000).
Several reports give strong evidence for an involvement of lipid rafts inparamyxovirusassem- bly. For MV, it has been shown that the H, F, M and N proteins partition into DRMs (Manie et al., 2000). The F protein has an intrinsic abilitiy to segregate into DRMs, while the H proteins becomes only detergent resistant if co-expressed with the F protein (Vincent et al., 2000). In contrast to influenza virus, the M (and N) proteins do no require the co-expression of the viral glycoproteins for partitioning into DRMs (Manie et al., 2000), but detergent resistance of the N protein requires the presence of the genome as well as of the M protein (Vincent et al., 2000).
This body of data suggests that lipid rafts are the assembly sites for MV. A model by Vincent et al. (2000) proposes that the F protein partitions into lipid rafts and additionally drags the H protein into these domains. The N protein, as RNP complex, is dragged to the assembly site by an interaction with the raft associated M protein. Association of the M protein with lipid rafts as well as interaction of the M protein with the cytoplasmic tail of the F protein, concentrates all viral components to a common assembly site within lipid rafts (Vincent et al., 2000). Indeed it has been shown that MV-components isolated from DRMs are infectious (Manie et al., 2000).
Sendai virusproteins also partition into DRMs, but in contrast to MV,Sendai virusF and HN both have intrinsic signals for a DRM association. However, mutational analyses as well as kinetic studies propose that the M protein has to interact with the cytoplasmic tail of HN or F, residing in DRMs, to also gain detergent resistance (Ali and Nayak, 2000).
Several reports also describe a role for lipid rafts in the assembly ofRespiratory syncytial virus.
Similar toSendai virusthe F proteins has intrinsic signals for association with DRMs, while the association of the M protein with DRMs depends on interaction with the F protein (Henderson et al., 2002; Fleming et al., 2006; Oomens et al., 2006; Marty et al., 2004).
42 1. Introduction
The F protein ofNewcastle disease virus(NDV) is also DRM resident (Dolganiuc et al., 2003) and very recently it has been published that also NDV HN and NP protein partition into DRMs and that virus assembly takes place at lipid rafts (Laliberte et al., 2006).
1.3 Cell polarity
Epithelial cells confine the body from the environment, thus facing a special problem. On the one hand, the side exposed to the environment has to withstand hostile conditions like digestive enzymes, dehydrating conditions or attack by pathogens. On the other hand, the side facing the underlying tissue has to be able to communicate with neighbouring cells and it has to provide blood supply for uptake of nutrients and molecular signals. Therefore, epithelial cells have evolved a special, polar organisation with functionally differentiated plasma membrane domains. The apical membrane is in contact with the environment and the basolateral domain faces the underlying tissue. These two membrane domains display unique protein and lipid compositions (Simons and Fuller, 1985). Tight junctions, located near the apical surface, not only connect neighbouring cells tightly, making passage of molecules between cells impossible, but also prevent random lipid and protein diffusion between the apical and basolateral side in the exoplasmic leaflet (van Meer and Simons, 1986). To sustain the different compositions of the membrane domains, newly synthesised lipids and proteins are targeted to the right domain by various mechanisms.
1.3.1 Impact on virus replication
Epithelial cells line all body cavities, making up the border between the environment and the internal milieu of the organism. The apical membrane domain has evolved a rigid structure to protect the body from invasion of pathogens. Most viruses have to overcome this barrier twice when interacting with a host: first during invasion of the host and secondly when spreading to a new host. Viruses have evolved different mechanisms how to overcome this barrier: some viruses circumvent infection of epithelial cells, for example by exploiting lesions within the cell layer or by using vectors like arthropods, while others have adapted to the specialised organisation of epithelial cells for sccessful infection.
1.3. Cell polarity 43
1.3.1.1 Entry and release
For several viruses, which have adapted to infect polarised epithelial cells, it has been shown that virus entry is restricted to one side and in many cases virus release is also restricted to one side. Polar distribution of the viral receptor is the most common cause for a restricted en- try side (Compans and Herrler, 2005). Some viruses, like VSV, orHerpes simplex virus, start infection predominantly from the basolateral membrane (Fuller et al., 1984; Schelhaas et al., 2003). Those viruses have evolved mechanisms to overcome the epithelial barrier during in- vasion of the host, for example by exploiting lesions within the epithelial cell layer, like VSV.
Other viruses, likeSendai virus, can infect epithelial cells from the apical side, making direct invasion of the host possible (Tashiro et al., 1990a).
For some paramyxoviruses, the interaction with polar cells has been investigated. For most of them it has been shown that infection of epithelial cells is more efficient from the apical side, e.g. forRespiratory syncytial virus(Zhang et al., 2002) andHuman parainfluenza virus type 3 (Zhang et al., 2005).
For MV there is some controversy about its polar entry. Blau and Compans (1995) showed that MV strain Edmonston enters the polar cell lines Vero C1008 and Caco-2 predominantly from the apical side and showed a mainly apical distribution of the MV vaccine strain receptor CD46.
However, Sinn et al. (2002) reported that MV strain Edmonston enters well-differentiated hu- man airway epithelial cells predominantly from the basolateral side although even in these cells CD46 was more abundantly expressed on the apical side.
Virus release from epithelial cells is often restricted to one side as well. For HIV and VSV it has been shown that they are released specifically from the basolateral side (Owens et al., 1991;
Rodriguez-Boulan and Sabatini, 1978; Fuller et al., 1984), while influenza virus is released from the apical side (Rodriguez-Boulan and Sabatini, 1978). For the paramyxovirusSendai virusit has been demonstrated that polarised virus release can be a virulence determinant. Infection of epithelial cells of the respiratory tract bySendai virustakes place at the apical side and the virus is predominantly released to this very side. Release from the apical side does not allow the virus to reach the blood stream. Therefore, apical release restricts virus dissemination to neighbouring cells, causing only local disease. For a mutant Sendai virus, F1-R, it has been shown that release is no longer restricted to the apical side, but that release from the basolateral side is almost equally efficient (Tashiro et al., 1990a,b). Release from the basolateral side may
44 1. Introduction
enable the virus to infect tissues underneath the epithelial cell layer and to reach the blood stream as well as the lymphatic vessels for efficient spread all over the body resulting in a systemic disease (Tashiro et al., 1992a).
Except for Sendai virus (Rodriguez-Boulan and Sabatini, 1978), an apical polar release has also been shown for other paramyxoviruses, namelyRespiratory syncytial virus(Roberts et al., 1995),Human parainfluenza virus type 3(Zhang et al., 2005), andSimian virus 5(Rodriguez- Boulan and Sabatini, 1978). MV has also been reported to be released from different cell lines as well as from well-differentiated human airway epithelial cells predominantly from the apical side (Blau and Compans, 1995; Maisner et al., 1998; Naim et al., 2000; Sinn et al., 2002).
1.3.1.2 Polar transport of proteins
For several paramyxoviruses, likeSendai virus,Respiratory syncytial virus, andHuman parain- fluenza virus type 3it has been shown that their glycoproteins are transported to the apical side of polarised epithelial cells, consistent with their preferred budding side (Rodriguez-Boulan and Pendergast, 1980; Roberts et al., 1995; Zhang et al., 2005).
However, for MV the polar distribution of the envelope glycoproteins is more complex. MV H and F are preferentially sorted to the basolateral side of epithelial cells, when expressed alone (Maisner et al., 1998) or in cells infected with a MV mutant that lacks the M protein (MV∆M) (Naim et al., 2000). This was ascribed to a tyrosine-based basolateral targeting signal within their cytoplasmic tails (Moll et al., 2001; Naim et al., 2000). In contrast, in wild type MV in- fected cells both proteins are targeted to the apical side at early times post infection (30 hours) (Naim et al., 2000). The VSV G protein, intrinsically targeted to the basolateral side, can be retargeted to the apical domain if its cytoplasmic tail is exchanged with the MV F cytoplasmic tail and is expressed together with the other MV proteins (Naim et al., 2000). It therefore has been proposed that the MV M protein redirects the intrinsically basolaterally transported H and F proteins to the apical side by interaction with their cytoplasmic tails (Naim et al., 2000). Ba- solateral expression of H and F mediates fusion with neighbouring cells, seen as syncytia (Moll et al., 2004). A basolateral distribution of H and F might therefore also mediate fusion with un- derlying cells, thus facilitating infection of underlying tissues without basolateral virus release.
Syncytium formation in Sendai virus infected cells has also only been seen for virus mutants which show a basolateral expression of the F protein (F1-R mutants) (Tashiro et al., 1992b).