J. gen. Virol. (1985), 66, 395-408. Printedin Great Britain 395 Key words: SFV/pathogenicity/mouse model

REVIEW ARTICLE

Semliki Forest Infection of Mice: A Model for Genetic and Molecular Analysis of Viral Pathogenicity

By GREGORY J. ATKINS, 1. BRIAN J. SHEAHAN 2 AND NIGEL J. DIMMOCK 3 1Department of Microbiology, Moyne Institute, Trinity College, Dublin 2, 'Department of Veterinary Pathology and Microbiology, Veterinary College of Ireland, Ballsbridge, Dublin 4, Ireland and 3Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, West Midlands, U.K.

INTRODUCTION During the past two decades there has been a great increase in our knowledge of the mechanisms of virus multiplication in cultured cells. The cells utilized, however, have usually been undifferentiated standard cell lines, and little attention has been given to the interactions of with differentiated cells, either in tissue culture or in the intact animal. Until recently, studies of viral pathogenicity have tended to be largely descriptive but a start has now been made in the analysis of the genetic and molecular control of virus pathogenicity for a number of model systems. Here we focus on one such system, Semliki Forest virus (SFV) infection of the laboratory mouse. This system has the initial advantages that the molecular biology of SFV and the closely related has been extensively studied using standard tissue culture cell lines, that SFV is neurotropic, that strains of extremes of virulence for mice are available, and that inbred and immune-deficient mouse strains can be utilized. While infections are primarily acute, realization that SFV can also cause persistent infections in the brain either by the addition of exogenous neutralizing antibody (Seamer et al., 1971) or defective interfering virus (T. Atkinson, A. D. T. Barrett & N. J. Dimmock, unpublished data) has greatly extended the potential of the model. SFV is a member of the genus of the family Togaviridae (Porterfield et al., 1978). It is a small, enveloped virus containing RNA of positive polarity. The RNA codes for four structural and four non-structural proteins which are formed by separate post-translational cleavage pathways. As well as the genomic 42S RNA, a subgenomic 26S RNA species is formed in infected ceils. This codes for the four virion proteins only (K~iri/iinen & S6derlund, 1978). Therefore, because of its relative simplicity, SFV is highly amenable to genetic and molecular manipulation. In the natural state, SFV is a -transmitted virus; the first isolation was by Smithburn & Haddow (1944) from mosquitoes in Uganda. This was followed by three more early isolations (summarized by Bradish et al., 1971) from which most laboratory strains are derived. Our intention is to review the genetic and molecular characteristics of the virus which influence its pathogenicity in the laboratory mouse. For a broader review of alphavirus pathogenicity the reader is referred to Schlesinger (1980).

Analysis of virulence Using naturally occurring virulent and avirulent strains Following their initial isolation of SFV, Smithburn & Haddow (1944) showed that the virus causes a lethal in adult mice inoculated intracerebrally. Even at the outset, mice died rapidly in about 6 days but after 40 passages mice inoculated with a high dose of virus usually died 2 days after infection. Subsequently, Seamer et al. (1967) studied the course of SFV infection in mice using the original isolate of Smithburn & Haddow (1944). They showed that,

0000-6311 © 1985 SGM 396 G. J. ATKINS, B. J. SHEAHAN AND N. J. DIMMOCK

following peripheral inoculation of mice of various ages, the virus multiplies in the spleen, liver, lymph nodes and central nervous system (CNS) but that only the CNS shows histological lesions. Although virus infection is invariably fatal, the mean time of death increases with age and was least for 3-day-old mice and greatest for 50-day-old mice. Murphy et al. (1970) later showed that multiplication of virus in the muscle of neonatal mice produces a viraemia which leads to invasion of the C NS. This was confirmed for weanling mice by Pusztai et al. (1971 ). The mechanism of entry of SFV into the brain from the blood was studied by Pathak & Webb (1974). Electron microscopy suggested two mechanisms. Firstly, coated vesicles formed at the surface of endothelial cells lining brain capillaries transport virus to the basement membrane and release it into extracellular spaces around brain cells. Secondly, actively migrating leukocytes may also transport virus across the blood-brain barrier. Alternatively, invasion of the brain may follow multiplication of the virus in vascular endothelial cells. Virulent and avirulent strains of SFV have been shown to differ in this property (Pusztai et al., 1971). However, at this stage it is not possible to be more precise about how virus crosses the blood-brain barrier. Bradish et aL (1971) studied the virulence of Smithburn & Haddow's original strain and three other isolates which had been passaged in animals or tissue culture to different extents. Although these strains were indistinguishable serologically, they differed markedly in their virulence for mice, guinea-pigs and rabbits when administered by the intraperitoneal (i.p.), intracerebral (i.c.) or respiratory routes. For mice, neonatal animals were susceptible to all strains, but adult mice injected i.p. were susceptible only to virulent strains. There was sharp transition to resistance to the avirulent A7 strain at 16 to 17 days of age. Infection with avirulent strains then led to benign replication of the virus and protection of the host against challenge with a virulent strain (Bradish et al., 1972). They suggested that there is a dynamic interaction in the host between distinct lethal and protective responses, but the mechanisms of virulence/~nd protection remained unclear. Following this initial description of the pathogenic properties of virulent and avirulent strains, several studies utilized strains of defined virulence in an attempt to relate virulence to properties of the virus. Pusztai et al. (197l) compared the multiplication of a virulent strain (V13) to that of an avirulent strain (A7) in weanling mice. They showed that, following i.p. inoculation, these strains multiply similarly in the early stages of infection. A viraemia of similar titre occurred, which was followed by invasion of the brain. However, V13 appeared in the brain 12 h before A7 and increased in titre until the death of the mouse, whereas A7 reached a peak in the brain and then declined. That the early pattern of infection is indistinguishable for virulent and avirulent strains was confirmed by Bradish & Allner (1972). Studies of the multiplication of virulent and avirulent strains in mouse brain organ cultures have produced conflicting results. Pattyn et al. (1975) and Woodward et al. (1978) found that the ability of SFV to multiply in mouse brain organ culture was dependent on the age of the mouse from which the culture was derived. Cultures from mice less than 15 days old supported the multiplication of both virulent (V13) and avirulent (A7) strains, whereas the avirulent strain multiplied poorly in cultures of adult mouse brain compared to virulent strains. This suggested that avirulence is due to a slower rate of multiplication in brain tissue rather than to any interaction with the immune system. Fleming (1977), however, found no difference in the ability of adult brain organ cultures to replicate virulent (L10) or avirulent (A774) strains. The disagreement with Woodward et al. (1978) may lie with the strains of virus used since both groups used Porton mice. Cultures from young mice could be infected with smaller quantities of virus than cultures from older mice. In infected animals, the avirulent strain multiplied more slowly in the brain than virulent strains following i.p. injection. It was suggested that this slower multiplication allows immune inter- vention before clinical signs become apparent. However, in this study the dose of the avirulent strain given was 1.8 x 104 p.f.u, compared to 4.2 × 105 p.f.u, for the virulent strain. It is not clear what influence this difference in dose had on the outcome of the infection. Zlotnik & Harris (1970) carried out electron microscopic studies of the brains of SFV-infected mice. Although the strain of SFV used is not specified, their description of clinical signs suggests that it was a virulent strain. Replicating virus could be detected in brain cells 48 h after infection of weanling mice, and was largely associated with neurons. Pathak et al. (1976) confirmed the Review: SFV injection of mice 397

_L 5v- o, . " ' .

Fig. 1. Neuronal cytoplasm with developmentalstages of the virus and budding of virus particles from the plasma membrane into the extracellular space. Weanling BALB/c mouse, 5 days post-infection with wild-type SFV. Bar marker represents 300 nm. destruction of neurons by a virulent strain and the association of replicating virus with these cells. Oligodendrocytes are also infected by virulent SFV (Pathak & Webb, 1983a). Particles of the avirulent A7 virus could be detected in the brains of infected neonatal mice but not in infected weanling mice. However, there were other signs of infection in weanling mice including possible nucleocapsids and cytoplasmic vesicles, along with inflammation and astrocytic hypertrophy (Pathak & Webh, 1978).

Using avirulent mutants induced J~'om a virulent strain Barrett et al. (1980) approached the analysis of virulence by isolating neurovirulence mutants in order to correlate alterations in virulence with alterations in properties of the virus. A cloned stock of the virulent L10 strain was subjected to mutagenesis in a manner known to generate temperature-sensitive mutants of the closely related Sindbis virus (Atkins et al., 1974). Selection was for survival of BALB/c mice injected with low doses of mutagenized clones. From 200 clones tested, four neurovirulence mutants, designated M4, M9, M103 and M136, were obtained. The parental L10 strain is lethal for both neonatal and adult mice. Survivors, which had been given doses of virus around the LDs0, were not protected against lethal challenge with the same strain. However, the majority of adult mice infected with the mutants survived the infection and were protected against lethal challenge; neonatal mice were equally susceptible to infection by mutants and wild-type. Virulent virus could be isolated from a minority of adult mice which died from infection with M4, M9 and M136, but not M103, suggesting that the former may be single site mutants. Analysis of the multiplication of the mutants in adult mice infected i.p. showed that M4 and M103 caused a viraemia but failed to enter the brain, M136 entered the brain only transiently, whereas M9 showed sustained multiplication in the brain (Barrett et al., 1980; Atkins & Sheahan, 1982). Unlike the wild-type or the other mutants, M9 produced clinical signs involving paralysis in 12~ of infected mice (Atkins & Sheahan, 1982). Electron microscopy of the brains of L10-infected mice showed replicating virus in neurons as well as glial cells (Barrett et al., 1980). Damage to neurons was a prominent feature (Fig. 1). 398 G. J. ATKINS, B. J. SHEAHAN AND N. J. DIMMOCK

Fig. 2. Vacuolated oligodendrocyte cytoplasm with developmental stages of virus. Weanling BALB/c mouse, 5 days post-infection with M9 mutant of SFV. Bar marker represents 700 nm.

However, in mice infected with mutants M9 and M136, virus was present in degenerating glial cells especially oligodendrocytes, but not in neurons that remained intact (Sheahan et al., 198 I, 1983; Atkins & Sheahan, 1982; Fig. 2). Compared to the wild-type L10 strain, mutants M4 and M103 have therefore lost the ability to enter the brain, whereas M9 and M136 have lost the ability to damage neurons. Both these defects lead to loss of neurovirulence. In order to ascertain the nature of the mutations leadi.ng to attenuation, multiplication of wild-type, mutant SFV strains (M4, M9, M 103, M 136) and the avirulent strain A7 was studied in tissue culture (Atkins & Sheahan, 1982; Atkins, 1983). Three cell lines were used: the hamster BHK-21 cell line, C 1300 mouse neuroblastoma cells and G26-24 mouse oligodendroglioma cells (Table 1). More detailed analysis showed that M4 was defective in nucleocapsid assembly, M136 in 26S RNA synthesis and M9 in total RNA synthesis. The nature of the defect in M103 is unclear, but there is less core protein synthesis in C1300 cells (Atkins & Sheahan, 1982). Reduction in A7 virus- specified RNA synthesis resulted in a decreased yield of infectious virus in C1300 cells only. Woodward & Smith (1979) showed that the multiplication of the avirulent A7 strain of SFV is temperature-sensitive compared to a virulent strain, but the molecular basis of this defect is not known. Following the analysis of neurovirulence mutants, the avirulent A7 strain was studied in the same way and compared to the virulent L10 strain (Atkins, 1983; Table 1). Like the neurovirulence mutant M9, A7 synthesized less RNA in BHK-21, G26-24 and C1300 cells. None of these avirulent viruses produced a c.p.e, in the C 1300, neuron-derived cell line. Thus, it seems likely that the lethal event in virulent SFV infection is the destruction of neurons. Limited damage to glial cells is probably tolerated by the mouse, and allows immune intervention before a lethal amount of damage can occur. Additional insight into the effect of SFV infection on CNS function has been gained from examining those neurochemical and neurophysiological changes which accompany SFV infection. Virulent virus caused a profound increase in dopamine metabolism which followed the peak of infectivity, and a smaller but still significant increase in serotonin breakdown. No changes were detected in the synthesis of acetylcholine or ~-amino butyric acid (A. D. T. Review: SFV infection of mice 399 Table 1. Properties of the naturally occurring avirulent A 7 and induced neurovirulence mutants of SFV (M4, M9, M103 and M136) in three cell lines BHK-21 C1300 G26-24 Property (neuroblastoma) (oligodendroglioma) Multiplication-defective All* All All C.p.e. All None All Inhibition of host All None None protein synthesis * Except A7; all categories are relative to the virulent wild-type L10 SFV.

Barrett, A. J. Cross, T. J. Crow, J. A. Johnson & N. J. Dimmock, unpublished data). These data suggest that although virulent SFV has a tropism for neurons it is more selective than was previously realized. Peripheral inoculation with the avirulent strain of SFV produces patchy demyelination within the CNS including the optic nerves (Illavia et al., 1982; Fleming et al., 1983) and studies have shown correlated physiological deficits in the visual system (Tremain & Ikeda, 1983) and an increase in the axonal transport of protein (Pessoa & Ikeda, 1984). Changes in behaviour have also been recorded (Webb et al., 1979).

The role of the immune system in recovery from SFV infection B- and T-lymphocytes Mice infected with avirulent strains of SFV become immune to challenge with virulent strains (Bradish et al., 1971). The mechanism of immunity has been analysed by measurement of the type of immunity developing, and by the use of immune-deficient and immune-modified mice. Allner et al. (1974) showed for the avirulent A7 strain of SFV that neutralizing antibody is detectable at 3 days following i.p. infection, and peaks at 7 days. Since the viraemia peaks at 2 days and disappears at 3 days, the development of neutralizing antibody coincides with the disappearance of virus from the blood but not from the brain. The importance of antibody in recovery from SFV infection is shown by the effect of cyclophosphamide, a known inhibitor of B-lymphocyte function. Treatment of mice with cyclophosphamide increases the duration of viraemia, depresses neutralizing antibody synthesis and enhances mortality (Bradish et al., 1975a). This has since been confirmed with the neurovirulence mutant M9 (Gates et al., 1984). The role of T-lymphocytes in immunity to SFV infection has been studied using nude (nu/nu; T-lymphocyte-deficient) mice, and by analysing the activity of T-lymphocytes from infected mice. Jagelman et al., (1978) showed that lesions in brains of nude mice infected with avirulent SFV are less severe than in phenotypically normal (nu/+) controls, and that clearance of virus from the brain is incomplete. Using the same system, Bradish et al. (1979) showed that the viraemia in nude mice is cleared normally but that virus persists in the brain. A peak of neutralizing antibody appeared at 5 to 9 days, only to disappear by 14 days after infection. Efficient clearance of virus and neutralizing antibody synthesis could be restored to nude mice by the transfer of spleen cells from immune-competent but uninfected animals. However, spleen cells from normal immune animals protect nude mice from infection with the virulent strain in the absence of detectable neutralizing antibody synthesis (Bradish et al., 1980). Infection of nude mice by the neurovirulence mutant M9 also leads to only marginally increased mortality (Gates et al., 1984). Although the situation is far from clear, complete recovery from infection appears to be assisted by both B- and T-lymphocytes. However, since a substantial number of nude mice survive infection with functions of both sets of cells impaired, it seems that other more important factors are at work.

Other components of the immune system Bradish & Titmuss (1981) studied the effects of administration of interferon and double- stranded RNA, an interferon inducer. Both these agents reduce the efficiency of infection with 400 G.J. ATKINS, B. J. SHEAHAN AND N. J. DIMMOCK SFV, but mortality is unaffected. Finter (1966), however, was able to protect mice against virus challenge using exogenously administered interferon, and Fauconnier (1971) showed that administration of anti-interferon antiserum exacerbates SFV infection in mice. Bradish & Allner (1972) and Fleming (1977) reported that little interferon is induced in the brain early in infection by either virulent or avirulent SFV, and others have found high levels only where there are high virus titres (T. Atkinson, A. D. T. Barrett & N. J. Dimmock, unpublished data). Blackman & Morris (1984) have shown that both interferon-c~fl and interferon-), are produced by spleen cells late in infection with SFV, but that both will stimulate cell-mediated cytotoxicity against SFV-infected target ceils. While most of these studies indicate that interferon is produced too late in infection to be protective, it seems that when infection hangs in the balance interferon can be an effective modulator. Activity of isolated natural killer cells and macrophages against SFV targets has also been reported (Macfarlan et al., 1977). In natural killer cell-deficient (beige) mice, the severity of early lesions is increased for the M9 mutant, but mortality is not affected (Gates et al., 1984). The role of macrophages in immunity to SFV infection was studied by Allner et al. (1974), who used a number of inhibitors of macrophage function. With the exception of sodium aurothiomalate (SATM or 'Myocrisin'), none of these influences the outcome of infection with an avirulent strain. As discussed below, SATM may have other effects on the host besides an effect on macrophages. Clofazimine, another inhibitor of macrophage function (Conalty & Jina, 1971), has no effect on the outcome of infection with the M9 mutant of SFV (M. C. Gates, unpublished results). Further studies showed that with both low and high doses of avirulent SFV, SATM increased mortality, its effect being to convert a normally avirulent infection into a virulent one (Bradish et al., 1975 b). The action of SATM was not associated with depression of antibody response, and it was suggested that it inhibited macrophages (Allner et al., 1974). However, Lipsky & Ziff(1976) showed that SATM inhibits lymphocyte proliferation, and Gates et al. (1984) have shown that SATM treatment depresses neutralizing antibody synthesis following SFV infection of BALB/c mice but not of C57BL mice. The effect of SATM on macrophages was studied by Oaten et al. (1980). It had been shown by van der Groen et al. (1976) that SFV is able to multiply in mouse peritoneal macrophages. Oaten et al. (1980) showed that the yield of SFV is enhanced in macrophages taken from SATM-treated mice. However, SATM also appears to affect the susceptibility of neurons in the brains of adult mice. Allner et al. (1974) showed that neuronal damage could be detected in the brains of mice infected with avirulent SFV only if the animals were pretreated with SATM. Subsequently, Pathak & Webb (1983b) showed that treatment with SATM allows the detection of mature virus particles by electron microscopy in the brains of mice infected with the avirulent A7 strain. In particular, multiplication of virus could be detected in neurons, which was not the case for untreated, infected controls. It was shown that SATM produces a membrane proliferating effect in the brain cells of uninfected, SATM-treated mice and the authors suggested that this enhances the replication of the virus. Another immunomodulator, 1-asparaginase which inhibits proliferation of B- and T-lymphocytes, enhances protection against low doses of avirulent virus but with high doses mortality is increased (Bradish et al., 1975b). These data underline the difficulty of interpreting what are likely to be multiple effects of inhibitors of biological systems. Recovery from SFV infections, in addition to the minor role played by T- and B-cells postulated above, is also modulated by interferon (Fauconnier, 1971) though whether this acts as an antiviral or an immune modulator is unknown. SATM may inhibit macrophages as well as have effects on intracellular membrane proliferation. If so, macrophages are likely to be the element responsible for the resistance of nude mice to SFV. The effects of anti-interferon and macrophage inhibitors in nude mice might finally resolve these uncertainties. Conversely, the addition of lymphokines (including interferon ),) would show if missing T cell products are responsible for the diminished ability of nude mice to combat SFV infection. Role of the immune system in the prevention of infection with SFV Neutralizing and non-neutralizing antibodies can protect mice from lethal challenge with a Review: SFV infection of mice 401 virulent strain. Boere et al. (1983) showed that neutralizing and a non-neutralizing monoclonal antibody to the E2 envelope protein both protect mice from a lethal challenge, but the neutralizing antibody was 100-fold more effective. In earlier work Seamer et al. (1971) similarly obtained protection with neutralizing antibody but in follow-up studies 15~ of surviving mice developed a delayed onset encephalitis, sometimes with virus in brain up to 70 days later. Most failed to develop serum-neutralizing antibody which indicated that they were to some degree immunosuppressed. Kraaijeveld et al. (1979a, b) have shown that mice may be protected against lethal challenge by intracutaneous immunization with purified inactivated SFV which results in cellular immunity without detectable antibody synthesis. The animals were protected against subcutaneous challenge, which spreads slowly, but not against i.p. challenge, which spreads rapidly, unless the i.p. cell population is stimulated with thioglycolate. Protection could be transferred to uninfected mice with peripheral lymph node cells, but not with spleen cells. Evidence that there is cell-mediated immunity to SFV came from cross-protection studies against SFV in mice immunized against Sindbis virus (Peck et al., 1975). Latif et al. (1979) showed that both nude and immune-competent mice immunized with Sindbis virus are equally protected against challenge with a virulent strain of the same virus. However, nude mice are less protected against heterologous challenge with SFV than immune-competent mice. Smith- Owirodu et al. (1980) showed that mice immunized with Sindbis virus are cross-protected against SFV: anti-Sindbis virus antibodies were present in high titre soon after infection, but no antibodies to SFV could be detected. In addition, it was found that cross-protection in Sindbis virus-immunized animals involves a temporary suppression of antibody synthesis to SFV following challenge with this virus.

Central nervous system demyelination Demyelination of CNS axons is a feature of some important virus diseases of man and animals. In particular, it has been proposed that the most common demyelinating disease of man, multiple sclerosis, may have a viral aetiology (McKhann, 1982). In order to study the mechanism of virus-induced demyelination, a number of laboratory models have been used and, from these, two theories have emerged. Firstly, demyelination could involve the triggering of an autoimmune response by the virus which results in myelin stripping. Secondly, demyelination could result from destruction of the myelin-producing cell, which in the CNS is the oligodendrocyte (Dal Canto & Rabinowitz, 1982). It is clear that these two theories are not necessarily mutually exclusive. CNS demyelination as a feature of SFV infection was first described by Smithburn & Haddow (1944) using their original isolate. Demyelination was induced by two or three i.p. inoculations of mice with avirulent SFV by Chew-Lira et al. (1976) although most subsequent studies have used a single inoculation (Suckling et aL, 1978). Infection and demyelination by avirulent SFV are accompanied by an inflammatory response (Chew-Lim, 1975; MacKenzie et al., 1978) and also an increase in glycosidase activity (Suckling et al., 1976, 1977) which may be a consequence of the inflammation. Leukocyte infiltration has been shown to accompany demyelination both for avirulent SFV (Illavia et al., 1982; Kelly et al., 1982) and for the demyelinating neurovirulence mutants M136 (Sheahan et al., 1981) and M9 (Sheahan et al., 1983). In addition, both immunoglobulin G levels and leukocyte counts have been shown to be elevated in the cerebrospinal fluid of the mouse following avirulent SFV infection (Parsons & Webb, 1982a, b). Several studies have utilized immune-modified and immune-deficient mice to analyse the contribution of the immune response to the demyelination process. In an initial study, Chew- Lim et al. (1977) studied the effect of irradiation of the host on demyelination by avirulent SFV. This increased rather than inhibited the severity of demyelination and also produced more extensive virus multiplication. Hence, it was concluded that demyelination is due to direct viral damage to the CNS rather than to an immunological reaction to CNS components. This was confirmed by Chew-Lira (1979) who found that demyelination was more extensive in T- lymphocyte-deficient (nude) mice than in immune-competent mice. Several other studies 402 G.J. ATKINS, B. J. SHEAHAN AND N. J. DIMMOCK disagreed. Jagelman et al. (1978) and Fazakerley et al. (1983) have found that nude mice show only mild microcystic lesions with no demyelination, whereas Berger (1980) reported that, although demyelination occurs in nude mice, it is decreased in severity compared to immune- competent mice. Using the M9 neurovirulence mutant, Gates et aL (1984) have shown that demyelinating lesions are less severe in nude mice and remyelination occurs more rapidly in these mice than in immune-competent mice. Berger (1980) has carried out reconstitution experiments following immune suppression with cyclophosphamide and has shown that other components of the immune system, besides T-lymphocytes, are important in demyelination. In immune-suppressed animals, immune spleen cells depleted of T-tymphocytes could reconstitute lesions, as could spleen cell populations depleted of B-lymphocytes. When both T- and B- lymphocytes were depleted, no lesions occurred, but lesions were most severe when both T- and B-lymphocytes were present. Lesions could also be reconstituted with immune serum plus non- immune bone marrow cells. This suggests that T-lymphocytes and B-lymphocytes contribute to the lesions, and that bone marrow-derived cells, possibly macrophages, also contribute. The consensus is that the immune system plays an important role in demyelination but how this is triggered by the virus is not known. One experimental model of CNS demyelination, experimental allergic encephalomyelitis (EAE), does not involve virus infection. Immune-mediated demyelination is brought about by injection of myelin or its components, together with adjuvants (Lando et al., 1980; Wisniewski et al., 1982). One of the factors that may exacerbate EAE is a prior infection with SFV (Suckling et a1.,1982). However, release of myelin components is not a normal event, and probably only follows CNS damage. Release of myelin antigens could be brought about by virus infection of oligodendrocytes either by damage to the cell or by incorporation into virus particles. It has also been suggested that incorporation of host cell glycolipid into the viral envelope may stimulate demyelination by the induction of a host autoimmune response (Webb et al., 1984; Webb & Fazakerley, 1984). For the M136 and M9 mutants of SFV, early infection of oligodendrocytes and damage to these cells precedes maximum demyelination which is then followed by remyelination (Sheahan et aL, 1981, 1983). Early electron microscopic studies of mice infected with the A7 strain of SFV did not detect replicating or mature virus in the brains of adult mice (Pathak et al., 1976; Pathak & Webb, 1978). It was concluded that demyelination is immune-mediated, but the early events which produced this result were unclear (Jagelman et al., 1978). G. J. Atkins et al. (unpublished results) have made a detailed comparison of infection of BALB/c mice with the same dose of the neurovirulence mutant M9, the avirulent A7 strain and the virulent L10 strain. Following i.p. infection, M9 multiplies in the brain to higher titre than A7, but lower than L10. Also, M9 produces clinical signs in a proportion of infected mice (Atkins & Sheahan, 1982). Although the severity of demyelination induced by A7 is dependent on the mouse strain employed (Suckling et al., 1980), it is generally less severe than that produced by M9, being detectable by light microscopy in a maximum of 25~o of A7-infected mice (Jagelman et al., 1978) and in 94~ of infected mice for M9 (Atkins & Sheahan, 1982). Following i.c. injection, however, A7 multiplies to a higher titre than following i.p. infection, and in this case replicating virus can be detected in oligodendrocytes but not neurons, and maximum demyelination occurs later (M. C. Gates, B. J. Sheahan & G. J. Atkins, unpublished results). Thus, it can be concluded that both M9 and A7 have defects (probably in RNA synthesis) which result in restricted cytopathogenicity both in tissue culture and in the whole animal. Multiplication of virus results in damage to oligodendrocytes which is not lethal. For this and other systems, it has been proposed that demyelination results from a 'bystander' effect, i.e. that the immune reaction to the virus somehow triggers an autoimmune reaction to myelin (Jagelman et al., 1978; Dal Canto & Rabinowitz, 1982). It may be that, for SFV, the damage to oligodendrocytes is incidental to immune-mediated demyelination. However, it is also possible that the release of normally sequestered myelin components, following damage to oligodendro- cytes, triggers immune-mediated demyelination in a manner similar to EAE. The initial phase of oligodendrocyte infection may be difficult to detect in some systems due to the insensitivity of electron microscopy. Since one oligodendrocyte makes and sustains the myelin sheaths of many Review: SFV infection of mice 403 internodal segments (Bunge, 1968; Peters & Proskauer, 1969), damage to a small minority of oligodendrocytes could result in significant release of myelin components.

Modulation of SFV pathogenieity by defective interfering (DI) particles DI particles are readily generated by undiluted passage or by growth of virus at low multiplicity (Bruton & Kennedy, 1976; Barrett et al., 1981). DI viruses are deletion mutants that have lost a large part, often the majority, of their . They synthesize no virus or novel proteins (Barrett & Dimmock, 1985) and are replicated only in cells co-inoculated with infectious SFV. It has been suggested that DI particles act to modulate virus infections (Huang, 1973). Woodward & Smith (1975) originally claimed that an avirulent strain of SFV generated DI particles in the brains of infected mice, in contrast to infection with a virulent strain but this claim was subsequently withdrawn (Woodward et al., 1978). Atkins (1983) reported that there is no significant difference between a virulent and avirulent strain in ability to generate DI particles on continuous passage. However, administration of a DI particle preparation simultaneously with the standard virus can prevent a lethal infection with SFV inoculated intranasaUy (Dimmock & Kennedy, 1978) or intraperitoneally (Barrett & Dimmock, 1984e). Intranasal administration of DI particles also prevents the appearance of histological lesions characteristic of SFV infection (Crouch et al., 1982). Thus, although DI virus converts an acute into a subclinical infection, it does not convert the pathology into that seen during infection with the classical 'avirulent' strain of virus as described above. Initially, it seemed that DI SFV RNA evolved through a number of stages to a molecule of constant size and sequence as judged by T1 oligonucleotide mapping (Stark & Kennedy, 1978). However, later work showed that there was extensive heterogeneity in sequence, size and cap structure of DI RNAs (Lehtovaara et al., 1981; K/i/iri/finen et al., 1981; Pettersson, 1981; S6derlund et al., 1981) and there was evidence of continuous change during passage. Barrett & Dimmock (1984 a) found that there is also considerable heterogeneity in biological properties of DI SFV. For instance not all DI virus preparations, which were standardized for interference by an in vitro assay, had the ability to modulate infection in mice. Thus, it was argued that interference by the DI virus alone is not sufficient to protect the mouse and it was suggested that protecting DI viruses also stimulates some part of the host defence system. There was further evidence of biological heterogeneity from the effects of DI virus on virus protein synthesis in vitro, on virus RNA synthesis, on the production of progeny infectious virus particles (Barrett et al., 1984a) and on the extent of interference with different strains of SFV or Sindbis virus (Barrett & Dimmock, 1984b). These DI particles were also physically heterogeneous as shown by centrifugation analysis: What was even more surprising was the way in which properties changed on passage as measured in vitro and in vivo (Barrett & Dimmock, 1984a, b, c, 1985). In the mouse, DI viruses varied in the way in which they induced protection. Always the amount of infectious virus in the brain was greatly decreased (> 99~), often only traces being present (Barrett et al., 1984b), and always these mice were symptomless. Protection was thought to be due to the combination of DI virus-mediated interference together with host responses stimulated by the DI virus which prevented the lethal progress of the virulent infection. However, despite the still extensive multiplication of virus in the CNS and systemically, mice protected by one of the DI viruses failed to mount an immune response capable of resisting challenge by a second dose of the virulent SFV, whereas mice protected to an identical extent by a second DI virus were immune. The possibility that under the latter conditions mice were producing immunizing amounts of non-infectious antigen was ruled out by immunochemical investigation of sections of brain (Crouch et al., 1982; Barrett et al., 1984b). Thus, it seems that some, as yet unknown, immunosuppression had occurred. One unexplained aspect has been the failure to demonstrate the presence of DI virus in the CNS of mice whose infection is clearly modulated by the DI virus (Barrett et al., 1984b). This may simply be a problem of detection as an amplification step enabled DI virus to be isolated (Dimmock & Kennedy, 1978). Modulation of the virulent infection by DI SFV has provided evidence of persistence in vivo as virus could be directly isolated from brain up to 3 months post-infection (T. Atkinson, A. D. T. Barrett & N. J. Dimmock, unpublished data). Up to 16~ of mice are persistently infected 404 G. J. ATKINS, B. J. SHEAHAN AND N. J. DIMMOCK following co-infection by DI virus and SFV and some of the isolates are of undiminished virulence. All but one of these infections are entirely subclinical. Thus, there is no doubt that DI SFV has the potential to modulate infection and its heterogeneity suggests that it can achieve this in a variety of ways. However, there is as yet no understanding of the stucture : function relationship of interfering DI RNAs.

Effect of SFV infection on the developing embryo A common feature of many togavirus infections is a teratogenic or lethal effect on the developing embryo. However, in contrast to infection of the CNS, this aspect of SFV pathogenicity has received little attention. Atkins et al. (1982) showed that the A7 strain of SFV, although avirulent for adult mice, is lethal for developing mouse embryos. Millner & Marshall (1984) later showed that this strain of virus infects the placenta before infecting the foetus. Infection of pregnant mice with A7 resulted in enhanced cell-mediated immunity, but this may have resulted from increased antigenic stimulation due to replication of virus in placental and embryonic tissue (Millner et al., 1984). The neurovirulence mutant M 103, however, does not kill developing embryos but does induce post-natal immunity to challenge with the virulent L10 strain. Caesarean section and fostering experiments have shown that this immunity could be transferred across the placenta, or by suckling to an immune mother. Thus, A7 and M103 represent the extremes of responses by the mother to infection. M 103 is lethal for neonatal mice, but is avirulent for embryos because it cannot traverse the placenta from the maternal blood. This is associated with inability of M 103 to infect the trophoblast cells surrounding the embryo, although A7 is able to do this efficiently (Atkins et al., 1982). The virulent L10 strain is also unable to traverse the placenta efficiently and infect the embryo, although it is able to infect the brain and kill the mother (P. Brennan & G. J. Atkins, unpublished results). Thus, SFV strains vary in their ability to traverse the placenta or to enter the brain and replicate efficiently.

CONCLUSIONS This review has summarized the advantages of SFV infection of the laboratory mouse as a modcl system for the genetic and molecular analysis of viral pathogenicity. Prominent among these is the ease with which infectionof thc brain or embryo can be accomplished by peripheral inoculation of relativcly low doses, the availability of virus strains of varying virulence, thc universal susceptibilityof inbred linesof mice and the wcll understood molecular biology of the virus. As the sequence of the gcnomc becomes fullyknown [the 26S RNA of one strain has been complctely sequcnccd (Garoff et al., 1980)] it will become possible to relate pathogenic properties of the virus to base sequence. Current understanding of pathogenicity suggests the following. I. Thc lethalityof virulentSFV strainsin adult mice resultsfrom damage to neurons, or more likcly particular subsets of neurons. 2. Avirulent strains (whether induced or arising naturally) infect oligodendrocytes,causing demyelination possible via an autoimmune reaction. 3. The major mechanism of rccovcry from infection is probably not mediated by B- or T- lymphocytes, although thesc play an auxiliary role. 4. Although DI virus particles exert their effect at the cellular levcl, they can influence pathogenic events in the intact animal. DI particles show great variation in their gcnomic sequences and biological properties including their abilityto affect SFV pathogenicity. They clearly have enormous potential for modulating both lethal and demyelinating infections. 5. In most studiesof SFV pathogenicityonly acute disease has been seen. Although Zlotnik et a/. (1972) showed that infection of mice with an avirulent strain of SFV could produce demyelinating lesions2 years latcr,this type of chronic disease has not been studied. Clearly, if SFV can be modified to producc a chronic, rather than acute disease, it would be a more appropriate model for chronic human diseases of possible viral actiology such as multiple sclerosis.Production of chronic diseasc may be accomplished by manipulating the virus or the host to induce virus persistcnceor the virus could be uscd to triggera chronic disease similar to chronic relapsing EAE. The former has been achievcd by thc addition of neutralizingantibody Review ." SFV infection of mice 405 or the intervention of DI SFV. The attractive prospect of the SFV-mouse system as a model for chronic human neurological or neuropsychiatric disease awaits experimentation. 6. Analysis of SFV infection of mouse embryos is currently in its early stages. However, since extreme responses to maternal infection can be produced using different strains of virus, it should be possible to analyse embryonic infection through the isolation and characterization of variants of A7, in the same way as it was possible to analyse the neurovirulence of L10.

We thank Sondra Schlesinger for her criticism of this manuscript. Work on analysis of SFV pathogenicity and demyelination is supported in our laboratories by the Medical Research Council of Ireland, the Multiple Sclerosis Society of Ireland (G.J.A. and B.J.S.) and the Medical Research Council (N.J.D.). This review was written during the tenure of a Fogarty International Fellowship from the National Institutes of Health to G.J.A.

REFERENCES ALLNER, K., BRADISH,C.J., FITZGEORGE, R. & NATHANSON,N. (1974). Modification by sodium auro-thio-malate of the expression of virulence in mice by defined strains of Semliki Forest virus. Journal of General Virology 24, 221- 228. ATKINS, G.J. (1983). The avirulent A7 strain of Semliki Forest virus has reduced cytopathogenicity for neuroblastoma cells compared to the virulent LI0 strain. Journal of General Virology 64, 1401-1404. ATKINS, G.J. & SHEAHAN, B. J. (1982). Semliki Forest virus neurovirulence mutants have altered cytopathogenicity for central nervous system cells. Infection and Immunity 36, 333-341. ATKINS, G.J., SAMUELS. J. & KENNEDY, S.I.T. (1974). Isolation and preliminary characterization of temperature- sensitive mutants of Sindbis virus strain AR339. Journal of General Virology 25, 371-380. ATKINS, G.J., CARTER, J. & SHEAHAN, B.J. (1982). Effect of alphavirus infection on mouse embryos. Infection and Immunity 38, 1285 1290. BARRETT, A.D.T. & DIMMOCK, N.J. (1984a). Modulation of Semliki Forest virus-induced infection of mice by defective interfering virus. Journal of lnfectious Diseases 150, 98-104. BARRETT, A.D.T. & DIMMOCK, N.J. (1984b). Variation in homotypic and heterotypic interference by defective interfering viruses derived from different strains of Semliki Forest virus and from Sindbis virus. Journal of General Virology 65, 1119-I 122. BARRETT, A.D.T. & DIMMOCK, N.J. (1984c). Modulation of a systemic Semliki Forest virus infection in mice by defective interfering virus. Journal of General Virology 65, 1827-1831. BARRETT,A.D.T. & DIMMOCK,N.J. (1985). Differential effects of defective interfering Semliki Forest virus on cellular and virus polypeptide synthesis. Virology (in press). BARRETT, A.D.T., CROUCH, C.F. & DIMMOCK, N.J. (1981). Assay of defective-interfering Semliki Forest virus by the inhibition of synthesis of virus-specified RNAs. Journal of General Virology 54, 273 280. BARRETT, A.D.T., CROUCH,C.F, & DIMMOCK,N.J. (1984a). Defective interfering Semliki Forest virus populations are biologically and physically heterogeneous. Journal of General Virology 65, 1273-1283. BARRETT,A.D.T., GUEST, A.R., MACKENZIE,A. & DIMMOCK,N.J. (1984b). Protection of mice infected with a lethal dose of Semliki Forest virus by defective interfering virus: modulation of virus multiplication. Journal of General Virology 65, 1909-1920. BARRETT, P.N., SHEAHAN, B.J. & ATKINS, G.J. (1980). Isolation and preliminary characterization of Semliki Forest virus mutants with altered virulence. Journal of General Virology 49, 141-147. BURGER, M.L. (1980). Humoral and cell-mediated immune mechanisms in the production of pathology in avirulent Semliki Forest virus encephalitis. Injection and Immunity 30, 244 253. BLACKMAN, M.J. & MORRIS, A.G. (1984). Gamma interferon production and cytotoxicity of spleen cells from mice infected with Semliki Forest virus Journal oJ General Virology 65, 955-961. BOERE, W.A.M., BENAISSA-TROUW,B.J., HARMSEN, M., KRAA IJEVELD, C.A. & SNIPPE, H. (1983). Neutralizing and non- neutralizing monoclonal antibodies to the E, glycoprotein of Semliki Forest virus can protect mice from lethal encephalitis. Journal of General Virology 64, 1405-1408. BRADISH, C.L & ALLNER, K. (1972). The early responses of mice to respiratory and intraperitonea/ infection by defined virulent and avirulent strains of Semliki Forest virus. Journal of General Virology 15, 205-218. BRADISH, C.J. & TITMUSS, D. (1981). The effects of interferon and double-stranded RNA upon the virus-host interaction: studies with togavirus strains in mice. Journal of General Virology 53, 21-30. BRADISH, C.L, ALLNER,K. & MABER,H.B. (1971). The virulence of original and derived strains of Semliki Forest virus for mice, guinea-pigs and rabbits. Journal of General Virology 12, 141-160. BRADISH, C.J., ALLNER, K. & MABER, H.B. (1972). Infection, interaction and the expression of virulence by defined strains of Semliki Forest virus. Journal oJ General Virology 16, 359-372. BRADISH, C.J., ALLNER. K. & FITZGEORGE, R. (1975a). Immunomodification and the expression of virulence in mice by defined strains of Semliki Forest virus: the effects of cyclophosphamide. Journal of General Virology 28, 225 237. BRADISH, C.J., ALLNER, K. & FITZGEORGE, R. (1975b). Immunomodification and the expression of virulence in mice by defined strains of Semliki Forest virus: the effect of myocrisin and l-asparaginase. Journal of General Virology 28, 239-250. 406 G. J, ATKINS, B. J. SHEAHAN AND N. J. DIMMOCK

BRADISH, C.J., FITZGEORGE, R., TITMUSS, D. & BASKERVILLE, A. (1979). The responses of nude-athymic mice to nominally avirulent togavirus infections. Journal of General Virology 42, 555-566. BRADISH, C.J., FITZGEORGE, R. & TITMUSS, D. (1980). The responses of normal and athymic mice to infections by togaviruses: strain differentiation in active and adoptive immunization. Journal of General Virology 46, 255 265. BRUTON, C.J, & KENNEDY, S.I.T. (1976). Defective-interfering particles of Semliki Forest virus: structural differences between standard virus and defective-interfering particles. Journal of General Virology 31, 383 395. BUNGE, R.P. (1968). Glial cells and the central myelin sheath. Physiological Reviews 48, 197-251. CHEW-LIM, M. (1975). Mouse encephalitis induced by avirulent Semliki Forest virus. Veterinary Pathology 12 387- 393 CHEW-LIM, M. (1979). Brain viral persistence and myelin damage in nude mice. Canadian Journal of Comparative Medicine 43, 39 43. CHEW-LIM, M., SUCKLING,A.J. & WEBB, H.E. (1976). Demyelination in mice after two or three inoculations of Semliki Forest virus. Veterinary Pathology 14, 67-72. CHEW-L1M, M., WEBB, H.E. & JAGELMAN, S. (1977). The effect of irradiation on demyelination induced by avirulent Semliki Forest virus. British Journal of Experimental Pathology 58, 459-464. CONALTY, M.L. & JINA, A.G. (1971). The antileprosy agent clofazamine (B. 663) in macrophages: light, electron microscope and function studies. In The Reticuloendothelial System and Immune Phenomena, pp. 323-331. Edited by N. R. D. Di Luzio. New York: Plenum Press. CROUCH, C.F., MACKENZIE,A. & DIMMOCK,N.J. (1982). The effect of defective-interfering Semliki Forest virus on the histopathology of infection with virulent Semliki Forest virus in mice. Journal oflnfectious Diseases 146, 411- 416, DAL CANTO, M.C. & RABINOWITZ, S.G. (1982). Experimental models of virus-induced demyelination of the central nervous system. Annals of Neurology 11, 109-127. DIMMOCK, N.J. & KENNEDY, S.I.T. (1978). Prevention of death in Semliki Forest virus-infected mice by administration of defective-interfering Semliki Forest virus. Journal of General Virology 39, 231-242. FAUCONNIER, B. (1971). Effet du s&um anti-interferon sur la pathog~nicit6 virale experimentale 'in vivo'. Pathologie et biologie 19, 575-578. FAZAKERLEY, J.K., AMOR, S. & WEBB, H.E. (1983). Reconstitution of Semliki Forest virus-infected mice induces immune-mediated pathological changes in the CNS. Clinical and Experimental Immunology 53, 115 130. FINTER, N.B. (1966). Interferon as an antiviral agent in t,it,o: quantitative and temporal aspects of the protection of mice against Semliki Forest virus. British Journal oJ Experirnental Pathology 47, 361-371. FLEMING, P. (1977). Age-dependent and strain-related differences of virulence of Semliki Forest virus in mice. Journal o/General Virology 37, 93 105. FLEMING, S., IKEDA, H., LANDON, D.N., PESSOA, V.F. & TANSEY, E.M. (1983). Fibre diameter spectra in Semliki Forest virus-induced demyelination of optic nerves of mice. Journal of Physiology 346, 24P. GAROFF, H., FRISCHAUF, A.M., SIMONS, K., LEHRBACH,H. & DELIUS, H. (1980). Nucleotide sequence of cDNA coding for Semliki Forest virus membrane glycoproteins. Nature, London 288, 236 241. GATES, M.C., SHEAHAN, B.J. & ATKINS, G.J. (1984). The pathogenicity of the M9 mutant of Semliki Forest virus in immune-compromised mice. Journal o/General Virology 65, 73-80. HUANG, A.S. (1973). Defective interfering viruses. Annual Review of Microbiologv 27 101 117. ILLAVIA, S.J., WEBB, H.E. & PATHAK,S. (1982). Demyelination induced in mice by avirulent Semliki Forest virus. 1. Virology and effects on optic nerve. Neuropathology and Applied Neurobiology 8, 35-42. JAGELMAN, S., StJCKLING, A.I., WEBB, lq.e. & 8OWEN, E.T.W. (1978). The pathogenesis of avirulent Semliki Forest virus infections in athymic nude mice. Journal q/'General Virology 41, 599 607. K~RIA1NEN, L. & SODERLUND,/-I. (1978). Structure and replication of ~-viruses. Current Topics in Microbiology and Immunology 82, 15-69. K~tRI'itINEN, L., PETTERSSON, R.F., KER,~NEN, S., LEHTOVAARA,P., SODERLUND, H. & UKKONEN,P. (1981). Multiple structurally related defective-interfering RNAs formed during undiluted passages of Semliki Forest virus. Virology 113, 686--697. KELLY, W.R., BLAKEMORE, W.F., JAGELMAN,S. & WEBB, H.E. (1982). Demyelination induced in mice by avirulent Semliki Forest virus. II. An ultrastructural study of focal demyelination in the brain. Neuropathology and Applied Neurobiology 8, 43 53. KRAAIJEVELD, C.A., HARMSEN, M. & KHADER BOUTAHAR-TROUW,B. (1979a). Cellular immunity against Semliki Forest virus in mice Injection and Immunity 23, 213-218. KRAAI/EVELD, C.A., HARMSEN, M. & KHADER BOUTAHAR-TROUW,B. (1979b). Delayed-type hypersensitivity against Semliki Forest virus in mice. Injection and Immunity 23, 219 223. LANDO, Z., TEITELBAUM,D. & ARNON, R. ( 1980)~ Induction of experimental allergic encephalomyelitis in genetically resistant strains of mice. Nature, London 287, 551 552. LATIF, Z., GATES, D., WUST, C.J. & BROWN, A. (1979). Cross protection among togaviruses in nude mice and litter mates. Journal of General Virology 45, 89-98. LEHTOVAARA,P., SODERLUND, H., KER.~NEN, S., PETTERSSON, R.F. & K~RIAINEN, L. (1981). 18S defective interfering RNA of Semliki Forest virus contains a triplicated linear repeat. Proceedings of the National Academy of Sciences, U.S.A. 78, 5353-5357. Review: SFV injection of mice 407

LtPSKV, P.E. & ZlFF, M. (1976). Inhibition of antigen and mitogen-induced human lymphocyte proliferation by gold sodium thiomalate. Arthritis and Rheumatism 91, 808 812. MACFARLAN, R.I., BURNS, W.H. & WHITE, D.O. (1977). Two cytotoxic cells in peritoneal cavity of virus-infected mice: antibody-dependent macrophages and non-specific killer cells. Journal oflmmunology 119, 1569 1574. MACKENZIE. k., SUCKLING, h.J., JAGELMAN, S. & WILSON, A.M. (1978). Histopathological and enzyme histochemical changes in experimental Semliki Forest virus infection in mice and their relevance to scrapie. Journal of Comparath,e Pathology 88, 335-344. McKHANN, G.M. (1982). Multiple sclerosis. Annual Review ~2[Neuroscience 5, 219-240. MILLNER, A.R. & MARSHALL, I.D. (1984). Pathogenesis of in utero infections with abortigenic and non-abortigenic in mice. Journal of General Virology 50, 66 72. MILLNER, A.R. & MARSHALL, I.D. & MULLBACUER, A. (1984). Effect of pregnancy on stimulation of alphavirus immunity in mice. Journal of General Virology :50, 73 76. MURPHY, V,A., HARRISON, A.K. & COLLINS, W.K. (1970). The role of extraneural infection in the pathogenesis of encephalitis. Laboratory lm'estigation 22, 318--328. OATEN, S.W., JAGELMAN, S. & WEBB, H.E. (1980). Further studies of macrophages in relationship to avirulent Semliki Forest virus infections. British Journal oJ Experimental Pathology 61, 150-155. PARSONS, L.M. & WEBB, H.E. (1982a). Virus titres and persistently raised white cell counts in cerebrospinal fluid in mice after peripheral infection with demyelinating Semliki Forest virus. Neuropathologr and Applied Neurobiology 8, 395-40 I. PARSONS, L.M. & WEBB, H.E. (1982b). Blood brain barrier disturbance and immunoglobulin G levels in the cerebrospinal fluid of the mouse following peripheral infection with the demyelinating strain of Semliki Forest virus. Journal of the Neurologieal Sciences 57, 307 318. PATHAK, S. & WE~B, H.E. (1974). Possible mechanisms for the transport of Semliki Forest virus into and within mouse brain. An electron-microscopic study Journal o/Neurological Sciences 23, 175-184. PATnAK, S. & WEBB, H.E. (1978). An electron-microscopic study of avirulent and virulent Semliki Forest virus in the brains of different ages of mice. Journal o/the Neurological Seiences 39, 199 211. PATHAK, S. & WEBB, H.E. (1983a). Semliki Forest virus multiplication in mouse brain with reference to demyelination. Journal oJ Physiology 339, 17P. PA'rHAK, S. & WEBB, H.E. (1983b). Effect of myocrisin on the morphogenesis of avirulent Semliki Forest virus in mouse brain: an electron microscopical study, Neuropathology and Applied Neurobioh~gy 9, 313 328. PATHAK, S., WEBB, H.E., OATEN, S.W. & BAXEMAN, S. (1976). An electron microscopic study of the development of virulent and avirulent strains of Semliki Forest virus in mouse brain. Journal of the Neurologieal Sciences 28, 289 295. PATTYN, S.R., DE VLEESSCHAUWER, L. & VAN DER GROEN, G. (1975). Replication of in mouse organ cultures. II. Multiplication of virulent and avirulent Semliki Forest viruses and western equine encephalitis viruses in mouse organ cultures. Archives of Virology 49, 33.38. PECK, R.D., BROWN, h. & WUST, C'.J. (1975). Preliminary evidence for cell-mediated immunity in cross-protection among group A arboviruses. Journal o/' Immunology 114, 518-584. PESSOA, V.E. & IKEDA, H. (1984). Increase in axonal transport in demyelinating optic nerve fibers in mice infected with Semliki Forest virus. Brain 107, 433-446. PETERS, A. & PROSKAUER, C.C. (1969). The ratio between myelin segments and oligodendrocytes in the optic nerve of the adult rat, Anatomical Record 163, 243. PETTERSSON, R. (1981). Y-terminal nucleotide sequences of Semliki Forest virus 18S defective interfering RNA is heterogeneous and different from the genomic 42S RNA. Proceedings of the National Academy of Sciences, U.S.A. 78, 115 119. PORTERFIELD, J.S., CASALS, J., CHUMAKOV, M.P., GAIDAMOVICH,S.Y., HANNOUN, C., HOLMES, I.H.. HORZINEK, M.C., MUSSGAY, M., OKER-BLOM, N., RUSSELL, P.K. & TRENT, D,W. (1978). Togaviridae. IntertirohJgv 9, 129 148. PUSZTAI, R., GOULD, E.A. & SMITH, H. (1971 ). Infection pattern in mice of an avirulent and virulent strain of Semliki Forest virus. British Journal o/Experimental Pathology 52, 669 677. SCHLESINGER, R.W. (1980). Virus-host interactions in natural and experimental infections with alphaviruses and flaviviruses. In The Togaviruses: Biology, Structure, Replication, pp. 83-106. Edited by R.W. Schlesinger. New York: Academic Press. SEAMER, J., RANDLES, W.J. & FITZGEORGE, R. (1967). The course of Semliki Forest virus infection in mice. British Journal oJ Experimental Pathology 48, 395--402. SEAMER, J.H., BOULTER, E.A. & ZLOTNIK, I. (1971). Delayed onset of encephalitis in mice passively immunized against Semliki Forest virus. British Journal of Experimental Pathology 52, 408-414. SHEAHAN, B.J., BARRETT, P.N. & ATKINS,G.J. ( 1981). Demyelination in mice from infection with a mutant of Semliki Forest virus. Acta neuropathologica 53, 129 136 SHEAHAN, B.J., GATES, M.C., CAFFREY, J.F. & ATKINS, G.J. (1983). Oligodendrocyte infection and demyelination produced in mice by the M9 mutant of Semliki Forest virus. Acta neuropathologica 60, 257 265. SMITHBURN, K.C. & HADDOW, A.J. (1944). Semliki Forest virus. I. Isolation and pathogenic properties. Journal of Immunology 49, 141 - 157. SMITH-OWIRODU, A., WUST, C..L, GATES, D. & BROWN, A. (1980). Cross-protection correlates with delayed antibody formation to challenge virus after immunization with Sindbis virus. Journal of General Virology 51, 351- 357. 408 G.J. ATKINS, B. J. SHEAHAN AND N. J. DIMMOCK

SODERLUND, H., KER.~NEN, S., LEHTOVAARA, P., PALVA, t., PETTERSSON, R. & K~RIA1NEN, L. (1981). Structural complexity of defective-interfering RNAs of Semliki Forest virus as revealed by analysis of complementary DNA. Nucleic Acids Research 9, 3403-3417. STARK, C. & KENNEDY, S.I.T. (1978). The generation and propagation of defective-interfering particles of Semliki Forest virus in different cell types. Virology 89, 285 299. SUCKLING, A.J., WEBB, H.E., CHEW-LIM, M. & OATEN, S.W. (1976). Effect of inapparent on the levels of lysosomal glycosidases in mouse brain. Journal of the Neurological Sciences 29, 109-116. SUCKLING, A.J., JAGELMAN,S. & WEBB, H.E. (1977). Brain lysosomal glycosidase activity in immunosuppressed mice infected with avirulent Semliki Forest virus. Injection and Immunity 15, 386-391. SUCKLING, A.J., PATHAK, S,, JAGELMAN, S. & WEBB, H.E. (1978). Virus-associated demyelination. A model using avirulent Semliki Forest virus infection of mice. Journal of the Neurological Sciences 39, 147-154. SUCKLING, A.I., JAGELMAN, S., ILLAVIA, S.L & WEBB, H.E. (1980). The effect of mouse strain on the pathogenesis of the encephalitis and demyelination induced by avirulent Semliki Forest virus infections. British Journal of Experimental Pathology 61, 281-284. SUCKLING, A.J., WILSON, N.R. & RUMSBY, M.G. (1982). Experimental allergic encephalomyelitis: modification of optic nerve pathology by antecedent virus infection. Acta neuropathologica 58, 101 106. TREMAIN, K.E. & IKEDA, H. (1983). Physiological deficits in the visual system of mice infected with Semliki Forest virus and their correlation with those seen in patients with demyelinating disease. Brain 106, 879 895. VAN DER GROEN, G., VANDEN BERGHE, D.A.R. & PATTYN, S.R. (1976). Interaction of mouse peritoneal macrophages with different arboviruses in vitro. Journal oJ General Virology 34, 353-361. WEBB, rI.E. & FAZAKERLEY,J.K. (1984). Can viral envelope glycolipids produce auto-immunity, with reference to the CNS and multiple sclerosis? Neuropathology and Applied Neurobiology 10, I 10. WEBB, H.E., CHEW-LIM, M., JAGELMAN, S., OATEN, S.W., PATHAK, S., SUCKLING, A.J, & MACKENZIE, A. (1979). Semliki Forest virus infections in mice as a model for studying acute and chronic central nervous system virus infections in man. In Clinical Neuroimmunology, pp. 369-389. Edited by F. Clifford Rose. Oxford: Blackwell Scientific Publications. WEBB, H.E., MEHTA, S., GREGSON, N.A. & LEIBOWITZ, S. (1984). Immunological reaction of the demyelinating Semliki Forest virus with immune serum to glycolipids and its possible importance to central nervous system viral auto-immune disease. Neuropathology and Applied Neurobiology 10, 77 84. WISNIEWSK1, H, M., LASSMANN, H., BROSNAN, C. F., MEHTA, P. D., LIDSKY, A. A. & MADRID, R. E. (1982). Multiple sclerosis: immunological and experimental aspects. Recent Advances in Clinical Neurology 3, 95-124. WOODWARD,C.G. & SMITH, H. (1975). Production of defective interfering virus in the brains of mice by art avirulent, in contrast with a virulent, strain of Semliki Forest virus. British Journal of Experimental Pathology 56, 363 372. WOODWARD,C.G. & SMITH,H. (1979). Temperature-sensitivity of an avirulent strain of Semliki Forest virus. FENS Microbiology Letters 5, 107-110. WOODWARD,U.G., MARSHALL,LD. & SMITH, H. (1978). Investigations of reasons for the avirulence of the A7 strain of Semliki Forest virus in adult mice. British Journal of Experimental Pathology 58, 616-624. ZLOTNIK, I. & HARRIS, W.J. (1970). The changes in cell organelles of neurons in the brains of adult mice and hamsters during Semliki Forest virus and encephalitis. British Journal of Experimenlal Pathology 51, 37 42. ZLOTNIK, I., GRANT, D.P. & BATTER-HATTON,D. (1972). Encephalopathy in mice following inapparent Semliki Forest virus (S.F.V.) infection. British Journal of Experimental Pathology 53, 125-129.