key: cord-0970877-l80yeyj1 authors: Dal Canto, Mauro C.; Rabinowitz, Stanley G. title: Experimental models of virus‐induced demyelination of the central nervous system date: 2004-10-07 journal: Ann Neurol DOI: 10.1002/ana.410110202 sha: b605e8cdfef41ab98ac7ffa2397db268dad830df doc_id: 970877 cord_uid: l80yeyj1 One of the arguments in favor of a viral pathogenesis for multiple sclerosis is the existence of several experimental and natural animal models of virus‐induced primary demyelination. This review deals comprehensively with such models. Well‐known examples of demyelinating viral infections in their natural host are JHM, Theiler, visna, and canine distemper encephalomyelitides. Recent reports of experimental murine infections with pathogens such as vesicular stomatitis, Chandipura, herpes simplex, Venezuelan equine encephalomyelitis, and Semliki Forest viruses are also discussed. The thrust of the review is to include viral models suspected of producing primary demyelination on an immunopathological basis. Since Pasteur first developed an antirabies vaccine [72] it has been recognized that viruses, either in- alomyelitis, which starts one to two weeks after infection, as well as its pathological similarity to EAE, have long suggested that demyelination in this disease is most probably caused by the host immune response [4 11 . A more common disease of humans, multiple sclerosis (MS), is also characterized by inflammatory demyelination [ 59, 781. The course, however, generally is not acute but rather is chronic, with exacerbations and remissions over a period of many years [59] . Epidemiological El], serological [68, 691, and virological [9, 131 data are compatible with the suggestion that a viral infection plays a role in the pathogenesis of MS [41] . In addition, because of the pathological similarities between MS and EAE as well as immunological data from human studies, several investigators [17, 11 . Nucleocapsids contain a single protein that may be cleaved by proteolytic enzymes used for cell dispersion. The virion envelope contains one large and one small glycoprotein [3] . Entry of virus into host cells is accomplished through virus attachment and fusion of the viral envelope with the cell membrane. Virus replication is mainly cytoplasmic, and maturation is by budding of assembled nucleocapsides through a virally modified cell membrane. One of the most characteristic effects of infection by CDV and related viruses is the induction of syncytium formation by cell fusion. This phenomenon has been observed both in vitro and in vivo [34, 43, 83, 721. CDV produces a natural disease in the dog characterized by an incubation period of three to six days, fever, nasal discharge, and both intestinal and respiratory involvement. Commonly, central nervous system (CNS) symptoms ensue either during the acute phase of the disease or late, after three or four weeks of illness. In some cases CNS infection may appear without previous signs of disease or after a period of recovery from systemic manifestations [6] . Initial studies at the ultrastructural level were performed in animals with clinical symptoms extending from ten days to twelve weeks prior to death. In all dogs, both demyelinating and destructive lesions were observed [78] . Demyelination occurred in areas with variable degrees of inflammation, was often observed around blood vessels (Fig l) , and was generally associated with cells. Processes of mononuclear cells were seen to encompass myelinated fibers and proceed either to phagocytose myelin from the outer toward the inner lamellae or to penetrate the myelin sheaths and strip myelin lamellae from their respective axons. Quiescent lesions were characterized by groups of denuded axons surrounded by activated glial processes in the absence of residual inflamma- alterations, the association of demyelination with inflammatory cells also suggested the possibility of an immune-mediated process of myelin injury. In an attempt to resolve this question, studies were recently done to determine the sequence of events leading to demyelination in the early phase of the acute form of natural distemper [83] . Abundant evidence of early myelin breakdown was found in the absence of inflammatory cells except for occasional macrophages. In well-established lesions, on the other hand, inflammatory cells were numerous. Important cellular components of the acute lesions were multinucleated giant cells which, under electron microscopy, appeared to have been derived from fusion of astrocytes. Viral inclusions were numerous in both giant cells and astrocytes as well as in cells considered to be oligodendrocytes (Fig 2) . A major conclusion from these studies was that demyelination in CDV infection was not necessarily related to the presence of an inflammatory response and that inflammatory 13-28, 1976.) cells probably represented a secondary late response to viral-induced tissue injury [83] . Because of the difficulty in studying the precise temporal development of natural distemper in the laboratory, investigators have tried to develop experimental models of the disease. To this point, however, the dog remains the only animal to develop demyelination after experimental infection with CDV. Initially, even in the natural host, CDV produced demyelination in only a small percentage of dogs. The isolation of R252 CDV by cocultivation with Vero cells and its subsequent inoculation into dogs expanded the incidence of demyelination to almost 50% of infected animals [62, 631. These data highlight the fact that even in the natural host, ordinary nonadapted strains of CDV do not appear capable of producing consistent demyelination. When R252 CDV was inoculated into 4or 8- week-old dogs, three clinicopathological syndromes were observed [62] . In some dogs, pathological evidence of acute encephalitis developed two to five days after the first neurological signs appeared. In these animals, multiple areas of demyelination were observed in the medulla and cerebellum. Inflammatory cells were absent in such demyelinating lesions. Reactive astrocytes were numerous, and many astrocytic nuclei contained eosinophilic viral inclusions, as did some oligodendroglial nuclei. Immunologically, dogs with the acute syndrome were affected by severe, persistent lymphopenia. Subacute encephalitis presenting with intermittent neurological dysfunction occurred in other animals. Affected animals survived for at least twelve weeks after infection. Pathologically, there were multifocal demyelinating lesions in association with perivascular cuffs of mononuclear cells. Inflammatory cells were also found in nondemyelinated areas of white matter, however, and some demyelinated areas were free of inflammatory cells. Intracytoplasmic and intranuclear accumulations of CDV nucleocapsids were present in several types of cells, especially astrocytes. Immunologically, these animals developed transient lymphopenia. The last group of animals displayed asymptomatic CDV infection with no pathological lesions but were found to have transient lymphopenia. The distribution of the acute versus subacute demyelinating disease appeared to be influenced by the animal's age at the time of inoculation. Thus, newborn dogs most often displayed the acute encepha-litic form of CDV infection, while the subacute form was more common when dogs were infected as weanlings [45] . Such data imply a delicate balance between CDV infection and mechanisms of host immunity. Altering the balance profoundly affects the outcome of the disease. Thus, when viral factors predominate, acute fatal encephalitis ensues; if host defenses predominate, persistent infection develops [45] . Support for this hypothesis can be found in experimental studies in which anti-CDV antibody passively transferred to newborn gnotobioric (germ-free) dogs resulted in prolongation of the disease course and, more importantly, in modification of the pathological lesions in the CNS [47] . Extensive involvement of gray matter, multiple viral inclusions in large neurons, and sporadic demyelination in dogs receiving anti-CDV antibody contrasted sharply with only sporadic foci of neuronal degeneration in untreated animals. Although the mechanisms of antibody-induced modification of CDV encephalitis are not well understood, these studies suggested that both clinical expression of the disease and severity of pathological lesions are at least partly dependent upon immunological factors [471. In this context, the possibility exists that demyelinating lesions themselves may be modulated by the host immune response rather than being produced by direct viral cytolytic activity. From the cited studies of both natural and experimental CDV encephalitis, several conclusions emerge. Two lines of evidence favor the hypothesis that demyelination is immune mediated: (1) mononuclear inflammatory infiltrates can be found in lesions of both the subacute and chronic forms of CDV infection, i.e., "old dog encephalitis" [62, 981; and ( 2 ) antimyelin antibodies have been shown to develop prior to the onset of encephalitis [44] . On the other hand, data against immune-mediated injury include the following: (1) the acute form of demyelination is not associated with inflammatory cells [83] ; ( 2 ) in the subacute form, many of the demyelinating lesions can be anatomically separated from mononuclear infiltrates, and vice versa [62, 83] [83] . In this regard, fusion of glial cells by virus, similar to the situation in subacute sclerosing panencephalitis, has been invoked as one possible mechanism leading to myelin disruption [83] . [60] . This makes further investigation of the model cumbersome and expensive. Furthermore, CDV is highly contagious, so studies must be conducted under strict conditions of isolation. Finally, immunological manipulations in dogs are not standardized and cannot be conducted under controlled genetic conditions such as exist for inbred strains of mice. JHM virus .(JHMV) is a neurotropic strain of mouse hepatitis virus, a member of the coronavirus family [84] . Coronaviruses owe their name to a crown, or corona, of widely spaced, bulbous peplomers or spikes, 12 to 24 nm in length, arranged around a spherical particle measuring 80 to 160 nm in diameter. They are enveloped, RNA-containing viruses that bud solely from the membranes of the endoplasmic reticulum, where nucleocapsids assume a horseshoe or flasklike configuration. The virion genome is a large, single piece of single-stranded infectious R N A of messenger polarity with a molecular size of about 5.4 million daltons [84] . The molecular events in coronavirus absorption, penetration, and uncoating, in biosynrhesis of viral macromolecules, and in viral assembly occurring in infected cells both in vitro and in vivo are not completely understood. During maturation, nucleocapsids are assembled in the cytoplasm in association with the endoplasmic reticulum and Golgi apparatus. At the stage of budding from the membranes, peplomers are added to the virion. Release of the virus occurs by lysis of the cell membranes rather than by budding from the cell surface [84] . JHMV is endogenous to mice and was isolated by Cheever et a1 in 1949 [15] when its capacity to produce demyelination was first recognized. Recent studies in mice, rats, and in vitro have produced useful data aiding in the interpretation of this model. The outcome of the infection in mice is known to depend on viral dose and age of the host. The importance of the dose of virus was clearly demonstrated by infecting 4-week-old Swiss mice "971. For example, demyelinating lesions could be obtained with inrracerebral inoculation of 10 to 100 SMLD50 (suckling mouse lethal dose) of virus, while inocula of 1,000 to 10,000 SMLDSO mainly resulted in gray and white matter necrosis. With increasing viral dose, higher mortality was observed [97] . As in CDV infection, demyelination was best observed in animals of a certain age. Consistent demyelinating lesions could in fact be obtained in 4- week-old mice inoculated with 10 to 100 SMLD,, of virus, while mice inoculated at 2 weeks of age always developed severe encephalomyelitis and died even after very small viral doses [97] . From studies in SJL mice, it appears that resistance to fatal JHMV infection of the CNS may depend in part upon age-related maturation of a specialized non-T, non-B, Ia-negative adherent cell population [89] . What specific antiviral mechanism these cells are responsible for is not yet clear [89] . Demyelination in 4-week-old mice develops in the first three to five days after infection and is not temporally associated with the appearance of mononuclear cell infiltrates, which may be observed shortly thereafter. Ultrastructurally, the virus is known to be pantropic since every cell type of the CNS becomes infected. Oligodendrocytes, however, appear to be the principal target for this virus [30, 521. JHMV replication in oligodendroglia, as in other types of cells, is characterized by virus budding into cytoplasmic vacuoles and cisterns of the endoplasmic reticulum [30, 521. Infected oligodendrocytes quickly undergo pathological alterations characterized by hyperproduction of haphazardly oriented membranes, abnormal multiple connections with myelin sheaths (Fig 3) [77], and formation of syncytia by fusion with other oligodendroglial cells [30] . The presence of morphological alterations in oligodendrocytes, the lack of a temporal and anatomical relationship between demyelination and inflammation, as well as the failure of immunosuppression to prevent demyelination [97] strongly suggest that myelin degeneration in this model is due to primary viral injury of the myelin-supporting cell rather than to immune-mediated mechanisms. An important feature of this model is the presence of active demyelination during the chronic phase of infection [38] . Strain and species differences influence this effect. Only 3 out of 110 weanling rats showed evidence of demyelination during chronic disease [64] . Among the strains of mice that have been studied, 15% of Balblc mice and 46% of C57BL/6 mice infected at 12 weeks of age showed evidence of chronic demyelination [91] . In these mice, virus could be isolated from the CNS only during the acute phase of the disease. N o detectable virus could be found later than fourteen days after inoculation, coincident with the increase in detectable antibody. Viral antigen, however, could be seen sporadically on immunofluorescence study in the CNS of infected animals for months after infection [91] , suggesting that a persistent and perhaps defective infection develops in this model. The mechanisms responsible for persistence of virus in JHMV infection have not been fully clarified. In vitro studies demonstrated that persistent infection of neuroblastoma cells is not associated with alterations in the properties of the virus, such as selection of temperature-sensitive or plaque mutants or production of defective interfering particles. Such studies have also failed to demonstrate the presence of interferon secreted by infected cells [90] . O n the other hand, the introduction of an antiviral factor such as antibody can be responsible for modulating productive into persistent infection [90] . Addition of low-concentration antiviral antibody, in fact, produced a JHMV carrier culture characterized by the presence of cytoplasmic viral antigen but without surface viral antigen, production of infectious virus, or cytopathic effects [90] . In analogy with the in vitro situation, in vivo host immune responses may be a critical element for the establishment and maintenance of persistent JHMV infection [90] . JHMV infection is an important model for the following reasons: (1) it is a persistent infection; (2) it is characterized by both acute and chronic demyelination; and ( 3 ) it produces murine infection, which allows easy immunological, genetic, and virological manipulations in the laboratory. Furthermore, coronaviruses have been isolated from the brains of two MS patients [9] . These observations await further confirmation but, if valid, would lend credence to the importance of coronaviruses in human disease. Visna virus belongs to the family of Retroviridae [4] . Virions are 60 to 90 nm in diameter and consist of a lipoprotein envelope enclosing an icosahedral shell that contains a filamentous nucleocapsid. The virion genome consists of a single-stranded RNA molecule composed of subunits and arranged in a coiled configuration. The genome is transcribed by a virion-associated reverse transcriptase into a D N A provirus that is incorporated into the cellular D N A [36] . This mechanism allows the virus to persist in cells without being recognized by the host immune system. Progeny RNA is transcribed from provirus D N A and acts as a messenger RNA. Provirus D N A is infectious. Viral maturation takes place by budding through the plasma membrane, which carries viruscoded glycoproteins. An important aspect of visna virus-cell interaction is the induction of s yncytia by cell fusion [4] . Visna is a natural slow virus disease of sheep involving both the respiratory system and the CNS 170, 881. Experimentally, intracerebral inoculation results in neurological disease after an incubation period of two months to eight years [76] . Symptoms also progress over periods of months to years, sometimes in a remitting-relapsing pattern eventuating in paraplegia or quadriplegia and ultimately death. Virus may be isolated from many tissues, including the CNS, for several years after infection 1751. Low viral titers, however, have made virus localization impossible by either immunofluorescence or electron microscopy. Pathologically, two phases of the disease may be recognized 133, 75, 881 . The first is a subacute encephalitis characterized by inflammatory infiltrates and relatively little necrosis of tissue. Apparently the virus infects only a few cells, with moderate cytolysis but with enough production of antigen to stimulate an antibody response. The second phase is more severe and does not follow a stereotyped pattern, since different animals become ill at intervals ranging from two months to more than six years after infection. Lesions are more focal than in the subacute encephalitic phase and may show varying degrees of severity in different foci. Two types of alterations are seen [34, 751. The first is characterized by inflammation and tissue necrosis, more severe around the ventricles and involving both gray and white matter in brain and spinal cord. The second type of lesion is characterized by primary demyelination that appears to be directly related to mononuclear inflammatory infiltrates. Acute and chronic lesions may be found at the same time, indicating that separate waves of demyelination may occur [ 3 2 ] . In some of the older lesions, remyelination may be observed (Fig 4) . Studies of the immune response to intracerebral [61, 67, 751 . Complement-fixing antibodies are detected in serum two to three weeks after infection, while virus neutralizing antibodies appear by six weeks to three months. Antibody titers then persist at a fairly constant level for years. High titers of neutralizing antibodies are also present in the cerebrospinal fluid (CSF). The lack of correlation between serum and CSF antibody titers suggests that viral antibodies are synthesized locally in the CNS. In support of this conclusion, typical plasma cells are generally present in both CSF and CNS tissues, and oligoclonal immunoglobulin bands have been observed in some CSF specimens. There is evidence that the immune response is important in limiting the spread of virus, thus contributing to the slowness of the infection [61] . During study of cell-mediated immune responses, a transient virus-specific lymphocyte blast transformation was observed in cells from both blood and CSF of infected sheep. The response, however, lasted only four or five weeks and was observed in animals without clinical disease [61] . Recently, aspects of the disease have been uncovered which may help to explain the persistence of infection. During the course of infection, neutralizing antibodies to variants of visna virus antigenically distinct from the original inoculum were observed [75] . In addition, isolates recovered years after infection were antigenically different from the original inoculum [65] . This shifting of antigenic characteristics, or "antigenic drift," may be a mechanism by which virus escapes host defenses, thus being able to persist in the host for prolonged periods. The molecular mechanisms responsible for antigenic drift are not understood. An important aspect of visna virus infection is the immunopathological nature of the early lesions, as demonstrated by immunosuppression studies with antithymocyte serum [66] . Whether myelin degeneration in the chronic phase is also immune mediated has yet to be determined. Two of the important characteristics of visna are persistence of the infection and relapsing course. The mechanisms underlying restriction of virus replication and the role of antigenic drift in the expression of clinical disease are potentially of great relevance to MS. As a model of demyelination, visna has some limitations. The use of sheep as a laboratory animal is cumbersome and expensive. Second, the lack of available inbred strains of sheep limits immunological manipulations. Finally, primary demyelination is part of a more complex pathological process. Little is known about the molecular biology of Theiler murine encephalomyelitis virus (TMEV), an R N A virus that belongs to the family of Picornaviridae. Mature virions possess three major structural polypeptides in the range of 25,000 to 35,000 daltons and a smaller fourth major polypeptide of 6,000 daltons [ 5 8 ] . Virions measure 2 5 to 29 nm and have icosahedral symmetry. Picornaviruses are freed from infected cells by cell lysis [ 5 4 ] . Very little is known about the mechanisms leading to viral persistence in some infected cells. In 1934 Max Theiler recovered several isolates from the CNS of young mice that had developed spontaneous flaccid paralysis. Inoculation of other mice with these brain-derived isolates produced a similar disease with pathological lesions closely resembling human poliomyelitis [94] . Theiler estab- lished the viral nature of these isolates and further demonstrated that surviving paralyzed mice had chronic infection of the C N S [94] . It was later recognized that Theiler viruses may be harbored in the intestinal contents of asymptomatic mice and that virus only rarely spread to the CNS to produce paralytic disease. Isolates from the CNS of paralyzed mice and from stools of asymptomatic mice were able to reproduce experimentally the same CNS paralytic disease as Theiler's original strains, and they are referred to as TO strains of TMEV. They contrast with two other strains of TMEV, GDVII and FA, which are less prevalent in nature and produce an acute encephalitis rather than chronic infection [ 571. In 1952 Daniels et a1 [29] first described myelin destruction in the spinal cord of mice that were inflammation and a chronic phase of white matter degeneration [ 5 5 ] . Only a small percentage of animals die of the acute gray matter disease, while the majority recover from this phase and later develop spastic paralysis indicating white matter destruction. Most mice that enter the phase of chronic white matter degeneration survive for months after infection [ 5 5 ] . Growth of TMEV is essentially limited to the C N S after a very brief initial period of viremia followed by limited growth in extraneural organs. During the first three weeks there is logarithmic growth of virus in the CNS; at this time viral antigen is mainly localized in neurons and their processes, though other types of cells may already be infected [ 5 5 ] . From six weeks to more than a year after inoculation, low levels of infectious virus may still be recovered from the CNS although viral titers steadily decline with time [ 5 5 ] . During the chronic phase of the infection, viral antigen may be observed by ultrastructural immunohistochemical study to be localized in several cells of the spinal cord, i.e., neuronal terminals, astrocytes, inflammatory cells, and especially macrophages in and around white matter lesions [23] . Pathologically, the first phase of spinal cord gray matter involvement is characterized by areas of focal necrosis, neuronophagia, and mononuclear cell infiltration. These changes are readily observed from the first to the fourth week after infection but decline progressively during the second month. Gray matter changes are rarely seen after six to seven weeks of infection [55] . The first phase of the infection is therefore an acute poliomyelitic disease that appears to be self limiting. The arrest in progression of gray matter lesions correlates temporally with the maximum rise in neutralizing antibody [ These changes are practically identical to those of EAE, the prototype for autoimmune diseases of the CNS [5 11 . In contrast to JHMV infection, oligodendroglial cells are often observed to be intact in areas rich in demyelinated axons, indicating that oligodendroglial infection is not the primary cause of myelin degeneration (Fig 6) [2 11. Ultrastructural immu-nohistochemica1 studies have also indicated that oligodendrocytes in and around demyelinating lesions appear to be free of viral antigen [23] . Although a low-level infection of these cells cannot be excluded, it appears that myelin degeneration in this disease is not primarily cytolytic. Immunosuppression studies strongly support this contention. When infected mice are treated with cyclophosphamide, white matter disease is greatly decreased or completely prevented, while gray matter disease is enhanced [56] . This The mechanisms underlying host-mediated virusinduced demyelination in this model are still under investigation. The immune response to TMEV infection is rather peculiar. Although serum neutralizing antibody is present by one week, there is an unusually slow rise in antibody titer, which reaches its peak at around two months after infection and remains at that level for the entire period of the disease [57] . Similarly, the cellular immune response, as measured by in vitro incorporation of ["]-labeled thymidine into DNA of spleen cells in the presence of ultraviolet-inactivated TMEV antigen, is not detectable until two months after infection. Once established, however, it gradually rises and is present for the entire duration of the disease [81] . The delay in immune responses may be due in part to limited exposure of the immune system to the virus, which replicates only in the CNS. Under these conditions, stimulation of the immune system may be totally dependent on chronic leakage of virus across the blood-brain barrier. Macrophages, which by ultrastructural immunohistochemistry have been observed to contain large amounts of viral antigen in diseased white matter for months after infection, may be instrumental in carrying the virus across the barrier [23] . Recent studies have shown that by changing the strain of the host or by attenuating the virus through cell adaptation, a remitting-relapsing course may be obtained in this infection [22] . Relapses are characterized by new waves of inflammatory demyelination in previously normal white matter or by renewed demyelination of areas that had previously been remyelinated, mostly by invading Schwann cells (Figs 7, 8 ) [22] . Remyelination of CNS axons by Schwann cells is very prominent in this model. Normal function is probably restored in axons remyelinated by Schwann cells since mice appear completely normal during the remyelinating phase. MS patients have also been shown to have Schwann cell-remyelinated axons in chronic lesions [35] . It is intriguing to speculate that stimulation of Schwann cell activity in demyelinated CNS lesions might result in amelioration of neurological function in MS patients as well. Ultrastructural immunohistochemical studies have shown that Schwann cells in areas undergoing a second wave of demyelination involving Schwann cellremyelinated axons are free of viral antigen, while other cells, including macrophages, contain abundant immune reaction product [20] . These results add support to the contention that demyelination in this model is not due to a primary viral attack on the myelinating cells, but rather is dependent on the host immune response. Theiler virus encephalomyelitis is an important model of human disease for the following reasons: (1) it is a persistent infection that develops in the majority of inoculated animals; ( 2 ) it utiIizes the natural host of TMEV, the mouse, which is the best laboratory animal for immunological, virological, and genetic manipulations; ( 3 ) acute lesions are characterized by inflammatory demyelination, as is the case in MS; and ( 4 ) the host immune response appears to play a major role in the pathogenesis of myelin injury. Although the infections described in the preceding sections have been considered the best available experimental viral models of primary demyelination, recent reports have detailed heretofore unexpected evidence indicating that many conventional viruses have the capacity to induce primary demyelination. Among these models are: temperature-sensitive (ts) mutants of vesicular stomatitis virus (VSV); a temperature-sensitive mutant of a human rhabdovirus, Chandipura virus (CV); Venezuelan equine encephalomyelitis virus (VEEV); Semliki Forest virus (SFV); and herpes simplex virus (HSV). While these viral models represent examples of artifically induced diseases, they nevertheless have the potential for advancing our knowledge into mechanisms underlying the production of demyelination. Furthermore, and most important, each viral model appears to represent primary demyeIination mediated b y immunopathological mechanisms. Vesicular stomatitis virus is a member of the rhabdovirus family. It is a negative-strand R N A virus with a genome coding for five structural proteins. The virus grows in many tissue culture cells, and substantial research has been done on its mechanisms of replication, maturation, assembly, and genetics [3 7, 391. In terms of clinical disease, VSV is a natural pathogen for cattle, producing a benign, self-limited, vesicular eruption [39] . Because the early appearance of the eruption is similar in some respects to that in cattle infected with foot-and-mouth disease virus, VSV infection can be mistaken for this more formidable disease and cattle quarantined until evidence is forthcoming to prove that the infection is benign. Temperature-sensitive mutants of VSV have been readily produced by mutagenizing standard parental virus and have been extensively studied to unravel the biochemical and genetic events underlying host-virus interaction [79] . Of particular interest in our present discussion is the fact that at least two ts mutants of VSV, ts G41 and ts G32, as part of a complex pathological process, produce primary demy elination. Inoculation of 3to 4-week-old Balb/c mice with ts G41 VSV produced in more than 90% a subacute neurological disease initially characterized by the development of lethargy, hunched posture, and ruffled fur within five to seven days after infection [SO] . In 60% of infected mice the initial neurological signs proceeded to hind limb paralysis. These signs appeared by seven to ten days after infection and lasted 21 to 28 days. Only 10 to 15% of the mice died as a result of the infection. By four weeks most of the mice had recovered, although up to 10% remained paralyzed. Virus was isolated from brains of infected mice for three weeks after infection, and the recovered virus remained temperature sensitive, ruling out the possibility of revertant virus [80] . Pathologically, ts G41 VSV infection of mice produced a spectrum of abnormalities [26] . By six to seven days after infection, meningitis and diffuse microglial infiltration of the anterior horns of the spinal cord were apparent. Between one and three weeks after infection, numerous foci of primary demyelination developed in the white matter of the spinal cord. These areas of demyelination were associated with inflammatory infiltrates of mononuclear cells [26] . In contrast to the clinical illness caused by ts G41 VSV, ts G32 VSV produced disease in only 50% of infected Balb/c mice. The animals became lethargic about five days after infection, most dying by two weeks. In some less severely affected mice, only hind limb paralysis developed [2 51. Pathologically, ts G32 VSV infection also produced both gray and white matter alterations. Infected mice displayed striking inflammation of the gray and white matter and prominent foci of primary demyelination. Again, as in ts G4 1 VSV infection, areas of demyelination were characterized by strict association with mononuclear inflammatory infiltrates and macrophages [2 51. Because demyelination appeared late in the course of both infections, was intimately associated with inflammatory cells, and did not appear to involve oligodendrocyte destruction, speculation arose that the mechanisms for demyelination might be immune mediated. T o determine if immune-mediated injury did contribute to demyelination in this model, nude and heterozygous littermates were infected with ts G32 VSV. Heterozygous mice demonstrated pathological alterations indistinguishable from those in Balb/c infected mice. Prominent inflammation and demyelination occurred. Homozygous nude mice, on the other hand, showed no evidence of myelin degeneration and little inflammation [25] . Gray matter disease, however, was enhanced in nude mice, and extensive necrosis occurred. Immunofluorescence studies performed in both heterozygous and nude mice demonstrated viral antigen in white and gray matter of both groups. In fact, in nude mice the amount of viral antigen detected as well as virus titers were greater than in heterozygous littermates, whereas demyelination was found only in heterozygous mice [25] . The studies in nude mice support the hypothesis that immune-mediated in jury contributes in a major way to the pathogenesis of primary demyelination in ts G32 VSV infection. Chandipura virus also is a member of the rhabdovirus family [7] . As such, it is a negative-strand RNA virus that produces infection in a host of tissue culture cell lines and matures by budding from the surface of infected cells. In contrast to VSV, CV is a human virus. Isolations of CV have been made in India and Africa from the blood of infected individuals, in whom the virus appears to produce a benign, self-limited, febrile illness [7] . To our knowledge, no reports have appeared of pathological studies conducted on tissues of infected persons. As with VSV, CV has been mutagenized in vitro, and clones of ts mutants of this virus have been produced by Gadkari and Pringle [31] . Whereas mice infected with wild-type CV succumbed to infection in three to four days with minimal neurological signs, infection by the ts mutant 472 CV was characterized by a slower disease process, beginning five to seven days after infection and lasting up to three weeks. Most mice demonstrated hind limb paralysis, and all affected mice eventually died. Pathologically, the most intriguing and important findings were the development of extensive inflam- matory infiltrates and plaques of demyelination in spinal cord white matter (Fig 9) [27] . Both the time of appearance and the close anatomical relationship between inflammation and myelin injury suggested that ts 472 CV infection was similar to ts G32 VSV and to Theiler virus myelitis in terms of the pathogenesis of primary demyelination [27] . T o evaluate this possibility, nude and heterozygous control mice were infected. Heterozygous nude mice showed perivascular mononuclear cell infiltrates in spinal cord white matter with surrounding primary demyelination. Homozygous nude mice, on the other hand, did not show white matter inflammation or demyelination, although, as with ts G32 VSV infection, extensive gray matter degeneration ensued [27] . These findings support the contention that whereas gray matter injury results from direct viral cytolytic activity, a host inflammatory response is required for production of primary demyelination. Venezuelan equine encephalomyelitis virus belongs to the family Togaviridae, genus Alphavirus (arbovirus group A) [5] . Togaviruses contain a nucleocapsid with cubic symmetry, 20 to 40 nm in diameter, enclosed in a lipid-containing envelope. The virus genome is a continuous molecule of single-stranded R N A with a molecular weight of 4 x lo6 daltons. Maturation of the virus occurs in the host cell cytoplasm. Nucleocapsids are formed in association with cell membranes and eventually bud from modified plasma membranes. Nucleocapsids possess antigenic determinants common to viruses of the same group, while the viral envelope contains both virus-specific and cross-reactive antigens [51. VEEV may cause encephalitis in both animals and humans [53] , but it had not been known to produce white matter alterations. Recently, however, extensive primary demyelination was observed in the spinal cords of infected mice starting nine days after infection [24] . Inflammatory cells were intimately associated with degenerating myelin, again suggesting an immunopathological mechanism of myelin injury (Fig 10) [24] . T o probe a possible immunopathogenesis of white matter alterations, congenitally athymic nude mice and heterozygous controls were similarly infected. Whereas heterozygous controls demonstrated both necrosis of gray matter and areas of inflammatory demyelination in white matter, nude mice showed only moderate gray matter alterations with no evidence of inflammation o r demyelination in white matter [24] . When immunofluorescence studies were performed, nude mice demonstrated a heavier antigenic burden in both gray and white matter than heterozygous controls; absence of primary demyelination could not be explained by lack of viral replication [24] . These studies support the contention that white matter changes in VEEV infection depend upon the host immune response rather than being produced by primary viral cytopathic effect. Semliki Forest virus is an arbovirus of the family Togaviridae and belongs to the same Alphavirus genus as VEEV [5] . SFV and VEEV therefore have similar morphological and physicochemical characteristics. In contrast to VEEV, SFV does not appear to be associated with any known illness of humans [5] . The virulent wild-type virus causes a fatal encephalitis in young mammals [8] . Infection of adult mice with the A774 avirulent strain of SFV, on the other hand, produces a focal, self-limiting encephalitis that is usually subclinical [ 161. Demyelination has recently been shown to be a feature of infection by this avirulent strain. Lesions were focal and were randomly distributed in the brain by ten days after inoculation [161. The pathogenesis of demyelination in SFV infection is controversial. A direct viral cytopathic effect on oligodendrocytes has been suggested by some investigators [86] , while others have implicated the host immune response in the pathogenesis of the CNS lesions, including demyelination [93] . In a recent study, demyelinating lesions were usually seen to be associated with mononuclear infiltrates rich in lymphocytes. A close anatomical association between lymphocytic processes and astrocytes or macrophages was interpreted as supporting evidence for immune-mediated myelin injury [42] . Experiments in nude mice also support this contention, as SFVinduced demyelination was reported to be less severe in these animals than in normal mice [40] . Relevance to MS These recent models are important primarily because of the following characteristics: (1) they are examples of possible immune-mediated primary demyelination produced by conventional viruses; (2) they employ viruses that differ widely in their biochemistry, genetics, and replicative cycle; and (3) they involve murine hosts in which extensive data on host immune processes and genetics are available. Among the limitations of these models are that: (1) infection occurs in an unnatural host; (2) none of these infections are chronic or characterized by a relapsing and remitting course; and (3) in many of these infections demyelination is part of a complex pathological picture. Up to this point only RNA viruses capable of producing primary demyelination experimentally have been discussed. A recent model employing the D N A virus herpes simplex will now be reviewed. HSV belongs to the family Herpetoviridae [2] . Two types, H S V l and HSV2, have been recognized, the first usually being responsible for herpes labialis and the second for genital infections. HSV has an icosahedral capsid, 85 to 110 nm in diameter, that comprises 162 capsomers. The capsid contains a core that appears as a ring of D N A surrounding a protein plug. The capsid is surrounded by a tegument that is wrapped in an envelope. Spikes project from the envelope to the outside. Virus development begins in the nucleus and continues in the cytoplasm until maturation by budding occurs. Cell-to-cell transfer of viral particles is common. The viral genome is made of double-stranded, linear D N A molecules [2] . Inoculation of the Rodanus strain of type 1 HSV into the cornea of mice and rabbits resulted in an ascending infection of the trigeminal nerve that reached the root-CNS transition in three to four days [95] . As soon as the infection reached the CNS portion of the trigeminal root entry zone, severe myelin destruction ensued accompanied by mononuclear inflammatory cell infiltration. In contrast, no demyelination was observed in the peripheral portion of the root entry zone or in the trigem-inal nerve [95] . By immunofluorescence, viral antigen was rarely seen in trigeminal ganglion neurons, while it was always present in the CNS portion of the root entry zone as early as four to six days after inoculation. Astrocytes appeared to be the most heavily infected cells, but oligodendrocytes were also affected [95] . Immunosuppression of mice with cyclophosphamide prior to infection caused a marked reduction of mononuclear infiltrates in the CNS accompanied by a decrease in myelin destruction when compared with infected nonimmunosuppressed controls [96] . Both groups of animals showed comparable viral titers and serum neutralizing antibodies. From these studies the authors concluded that the cellular immune response plays a definitive role in the process of myelin destruction in this animal model of HSV infection. In contrast to other models of virus-induced demyelination, the HSV model is characterized by a single lesion in a well-defined, predictable site. The model lacks the multicentric nature of the demyelination found in both MS and the other experimental diseases. O n the other hand, the fact that HSV is a ubiquitous virus in humans together with its known predilection for establishing latent infection makes potentially important the observations relative to its capacity of producing primary demyelination. From the models presented, it appears that viruses possess the capacity to cause widespread and extensive myelin destruction. At least two different mechanisms may be involved in the production of demyelination in viral infections. First, a direct viral cytopathic effect on oligodendrocytes can occur. JHMV encephalomyelitis of mice is the prototypic example of such a mechanism. A second mechanism, involving host immune reactivity to virus or CNS antigens, appears to play a major role in other models. Theiler virus encephalomyelitis may be the best example of such immune-mediated injury. In most infections, both mechanisms probably operate but their relative role varies. Since participation of the host immune system appears to play an important role in MS, models based on combined virus-immune pathogenesis may be especially relevant to human d' isease. There could be several mechanisms of virusinduced myelin injury by an immune pathogenesis. First, viral antigenic material could be inserted into the plasma membranes of oligodendrocytes, which in turn could be damaged by the ensuing host antiviral immune attack. This is not likely to happen in Theiler virus infection, as picornaviruses do not bud from cells. Second, virus and myelin could share some antigenic determinants. Although cross reactivity has been demonstrated between measles virus and myelin basic protein [71] , this is unlikely to be the case with several different and unrelated viruses. Third, virus could cause primary myelin injury that would lead to release of free myelin antigenic material. The resulting series of events would be quite similar to EAE. Although this appears to be an attractive possibility, data exist which are not supportive. For example, experiments measuring cellmediated immune responses to myelin basic protein in Theiler virus infection have given inconclusive results (unpublished data), in contrast to studies of EAE [73, 741. Finally, in addition to the foregoing classic theories, considerable attention has been devoted in the last few years to the possibility that myelin may be damaged as an innocent bystander in the midst of an immune reaction against unrelated antigens. This concept originates from work done in 1968 by Ruddle and Waksman [85] , who demonstrated nonspecific bystander killing of rat embryo fibroblasts in the presence of heterologous antigen and sensitized lymphoid cells. More recently, Wisniewski and Bloom [loo] extended this concept to the CNS in vivo. They demonstrated primary demyelination following local injection of purified protein derivative in the spinal cord of guinea pigs previously sensitized to that antigen. It has been proposed that a similar mechanism may operate in viral infections, in which virus would act as the target antigen and myelin would be nonspecifically damaged by a surrounding antiviral immunological attack [ 2 1, 56, 791. Such a mechanism of nonspecific myelin injury is supported by recent in vitro studies that demonstrated myelin vulnerability in the presence of neutral proteases, including plasminogen activator, secreted by nonspecifically activated macrophages [ 121. The ultrastructural feature of vesicular disruption of myelin, often observed in demyelinating models, might well be the morphological expression of such biochemical events [ 2 8 ] . A mechanism of bystander injury in viral infections is particularly attractive because the various models described by different investigators employ viruses that are largely unrelated to each other. The host immune response therefore could represent the unifying factor behind the capacity of different viruses to cause similar pathological alterations in the CNS white matter. Primary demyelination in response to a range of viruses may not be an unusual or rare event, but may represent a common consequence of the resolution of many viral infections. The clinical expression of diseases like MS may require multiple virological at-tacks engrafted on the appropriate immunological and genetic milieu. 2) . Molecular weight standards as in Figure 2 . Risk of multiple sclerosis related to age at immigration to Israel Viruses of Vertebrates Viruses of Vertebrates Viruses of Vertebrates Viruses of Vertebrates Pathogenesis of canine distemper Chandipura, a new rhabdovirus isolated in India from patients with febrile illness The virulence of original and derived strains of Semliki Forest virus for mice, guinea pigs and rabbits Two coronaviruses isolated from central nervous system tissue of two multiple sclerosis patients Multiple sclerosis, house pets and canine distemper: critical review of recent reports Measles, canine distemper and respiratory syncytial virions and nucleocapsids. A comparative study of their structure, polypeptide and nucleic acid composition Degradation of basic protein in myelin by neutral proteases secreted by stimulated macrophages: a possible mechanism of inflammatory demyelination Decreased percentage of polymorphonuclear neutrophils in mouse peripheral blood after inoculation with material from multiple sclerosis patients Postinfectious perivenous encephalitis and acute hemorrhagic leukoencephalitis Demyelination in mice after two or three infections with avirulent Semliki Forest virus In vztro cellular responsiveness in multiple sclerosis patients to a purified measles virus nuclear core and to other viral antigens A possible association between house pets and multiple sclerosis Neutralizing antibodies to canine distemper and measles virus in multiple sclerosis Uncoupled relationship between demyelination and primary infection of myelinating cells in Theiler's virus encephalomyelitis Lipton HL Primary demyelination in Theiler's virus infection: an ultrastructural study Schwann cell remyelination and recurrent demyelination in the central nervous system of mice infected with attenuated Theiler's virus Ultrastructural immunohistochemical demonstration of Theiler's virus in acute and chronic Theiler's demyelinating encephalomyelitis Central nervous system demyelination in Venezuelan equine encephalomyelitis infection. An experimental model of virus-induced myelin injury Murine central nervous system infection by a viral temperature-sensitive mutant Subacute infection with temperature-sensitive vesicular stomatitis virus mutant G41 in the central nervous system of mice. 11. Immunofluorescent, morphologic and immunologic studies Virus-induced demyelination. Production by a viral temperature-sensitive mutant Vesicular disruption of myelin in autoimmune demyelination Observations on encephalomyelitis of mice (DA strain) Further ultrastructural observations of virus morphogenesis and myelin pathology of JHM virus encephalomyelitis Temperature-sensitive mutants of Chandipura virus. I. Inter-and intragroup complementation Primary demyelination in visna. An ultrastructural study of Icelandic sheep with clinical signs following experimental infection The pathology of visna and maedi in sheep The ultrastructure of early visna lesions Remyelination in multiple sclerosis with peripheral type myelin The detailed structure of visna-maedi virus Experimental measles encephalitis: a genetic analysis Mouse hepatitis virus-induced recurrent demyelination Vesicular stomatitis and related viruses The pathogenesis of avirulent Semliki Forest virus infection in athymic nude mice The role of viral infections in demyelinating diseases An EM study of the focal demyelinating lesions in the mouse CNS during 29 days after an IP injection of avirulent Semliki Forest virus Ultrastructure of canine distemper virus in explant tissue cultures of canine cerebellum Myelin specific autoantibodies associated with central nervous system demyelination in canine distemper infection Age-related susceptibility to infection with canine distemper virus in gnotobiotic dogs Canine distemper virus and multiple sclerosis Canine distemper virus-associated encephalitis: modification by passive antibody administration Multiple sclerosis and canine pets Multiple sclerosis in the Faroe Islands: I. Clinical and epidemiological features Autoimmune and virus-induced demyelinating diseases. A review Electron microscopic studies on the vascular permeability and the mechanisms of demyelination in experimental allergic encephalomyelitis Mechanism of demyelination in JHM virus encephalomyelitis. Electron microscopic studies Sequelae of Venezuelan equine encephalitis in humans: a four-year follow-up Reproduction of picornaviruses. Comp Virol 55. Lipton H L Theiler's virus infection in mice: an unusual Neurological Progress: Dal Canto and Rabinowitz: CNS Virus-induced Demyelination 125 biphasic disease process leading to demyelination Theiler's virus-induced demyelination. Prevention by immunosuppression Chronic Theiler's virus infection in mice Purification of Theiler's murine encephalomyelitis virus and analysis of the structural virion polypeptides: correlation of the polypeptide profile with virulence Multiple sclerosis Induction of chronic neurologic disease in mice with canine distemper virus Animal models of virus-induced demyelination Experimental canine distemper virus-induced demyelination Demyelinating activity of canine distemper virus isolates in gnotobiotic dogs Demyelinating encephalomyelitis induced by a long-term corona virus infection in rats Antigenic drift of visna virus in persistently infected sheep Pathogenesis of visna. 11. Effect of immunosuppression upon early central nervous system lesions Pathogenesis of visna. IV. Spinal fluid studies Comparison of antibodies against different viruses in cerebrospinal fluid and serum samples from patients with multiple sclerosis Immunoglobulin abnormalities in multiple sclerosis. Relation to clinical parameters: exacerbations and remissions Maedi and visna in sheep Antibodies to measles virus react with myelinic basic protein Pasteur L MCthode pour prkvenir la rage apres morsure The demyelinating diseases: clinical and experimental studies in animals and man Neuroimmunologic diseases of animals and humans Pathogenesis of visna I. Sequential virologic, serologic and pathologic studies Immunopathogenesis of visna. A slow virus disease of the central nervous system Oligodendrocytes and their myelin-plasma membrane connecrions in JHM mouse hepatitis virus encephalomyelitis The etiology and pathogenesis of multiple sclerosis Genetic characteristics of conditional lethal mutants of vesicular stomatitis virus induced by 5-fluorouracil, 5-azacytidine and ethyl methane sulfonide Subacute infection with temperature-sensitive vesicular stomatitis virus mutant G41 in the central nervous system of mice. I. Clinical and virological studies Cellular immunity in chronic Theiler's virus central nervous system infection Experimental allergic encephalomyelitis and related conditions On the development of CNS lesions in natural canine distemper encephalomyelitis Comprehensive Virology Cytotoxicity mediated by soluble antigen and lymphocytes in delayed hypersensitivity, Part 1 (characterization of the phenomenon) Demyelination in mice resulting from infection with a mutant of Semliki Forest virus Cellular hypersensitivity to basic myelin (Al) protein and clinical multiple sclerosis Sigurdsson B: Rida, a chronic encephalitis of sheep Resistance to fatal central nervous system disease by mouse hepatitis virus, strain JHM. 11. Adherent cell-mediated protection Stability of neurotropic mouse hepatitis virus UHM strain) during chronic infection of neuroblastoma cells Chronic central nervous system demyelination in mice after JHM virus infection The effects of canine distemper virus on explant tissue cultures of canine cerebellum Virusassociated demyelination. A model using avirulent Semliki Forest virus infection of mice Spontaneous encephalomyelitis of mice, a new virus disease Central nervous system susceptibility to herpes simplex virus infection The expert technical assistance of Jannie Tong, Jayashree Huprikar, and Mary Grover as well as the secretarial assistance of Nancy Starks are gratefully acknowledged.