Encephalitis lethargica occurred suddenly in 1916–17 as an epidemic that was simultaneously reported in Vienna and in France, and continued until 1926, when it gradually disappeared. Reports of the disease spread throughout the world. A previous similar episode occurred in northern Italy in 1889–90 called la nona. Subsequent to the Italian outbreak, frequent reports of encephalitis in Europe were recorded (Hall AJ, Lumilean Lectures, Lancet 1923;i:731, quoted in [1].) During the course of the 1917–1926 epidemic, approximately half a million people world-wide were affected by EL. About one-third of the patients died acutely, one-third developed PEP and the remainder recovered almost completely. Sporadic cases diagnosed as EL have been reported up to the present time, but it has not been clear whether they share the same etiology as classical EL [2–4].

Dale, Church et al. [3], reporting on a series of modern EL cases, suggested that EL is an autoimmune disorder. They based this on clinical presentation, intrathecal oligoclonal bands (OCB), anti-streptolysin-O titers and autoantibodies reactive against human basal ganglia antigens in EL patients. Western blotting showed that 95% of EL patients had autoantibodies reactive against human basal ganglia antigens. While Dale et al. offer indirect evidence (see Discussion), our study presents direct evidence for a specific Enterovirus etiology for both classical and modern EL.

EL was thought to be infectious at the time of the epidemic. However, a controversy arose as to whether there was a direct etiological association with the contemporaneous “Spanish influenza” pandemic. Recently, McCall, Henry et al. [5] found no evidence for influenza RNA in brain from EL or PEP. Also, McCall, Vilensky et al. [6] found little historical data supporting influenza as the etiology of EL.

Anderson, Vilensky et al. [7] in a literature review reported on the neuropathology of 112 cases of acute encephalitis lethargica from the period of 1915 to 1940. They found that cortical damage was present in 75% of the cases, damage to the meninges and brainstem in half the cases, and the substantia nigra was damaged in 13% of the cases.

In the acute stage, EL presented as encephalitis with either somnolent or manic behavior, mask-like facies, muscular rigidity, involuntary eye movements, and a tremor which was distinct from the pill-rolling tremor of Parkinson’s disease. Importantly, some cases of PEP developed without a history of acute EL. Other PEP patients were somnolent for many years but recovered temporarily with L-dopa treatment introduced during the 1960’s. Sacks in “Awakenings” [8] documented the course of the patients’ therapy. The cause of EL has remained a mystery [1].

In routine autopsy procedure the whole brain is immersed in formalin, resulting in slow penetration and suboptimal preservation of ultrastructure. In the epidemic period, bodies were not refrigerated post-mortem. This was the practice even in the city hospital for the American capital as reported in the Washington Post, Morgue Room at Gallingher Held Disgrace, August 18, 1929. “The morgue at Gallingher Municipal Hospital is a disgrace to civilization…without any refrigeration system or other means of refrigerating bodies…the Gallingher morgue is worse than the most unmodern slaughter house in the country.” Hence brain tissue from the EL epidemic underwent autolysis to a greater degree than tissue obtained by modern autopsies. This had the effect of almost completely destroying the structure of neural ribosomes, whereas the VLP in the cases of EL were virtually intact.

In contrast to the fragile architecture of the normal neuron, Booss and Esiri [9] described virus particles in brain that were relatively well preserved in otherwise poorly preserved tissue.

This study seeks to define the cause of classical and modern EL. With the application of TEM, immunohistochemistry and molecular biology, evidence was obtained supporting the hypothesis that EL was caused by a member of the genus Enterovirus.

The hypothesis that EL is caused by an Enterovirus was explored using TEM, immunohistochemistry and RTPCR. Four cases from the 1917–1926 classical EL epidemic were studied. Preliminary TEM data indicated that small virus particles may be present. Possible small DNA viruses were the parvoviridae family [11, 12] and viruses of the annellovirus genus, including Torque Teno virus (TTV) and MTTV [13, 14]. Parvoviruses require actively dividing cells as fetal hematopoietic tissue as their source of DNA polymerase. Brain is therefore an unlikely site for the replication of parvoviruses [11]. Immunohistochemical findings utilizing anti-parvovirus antibodies did not support a parvovirus etiology. No clinical disease has been identified with the annellovirus genus, making it an unlikely candidate for EL [13, 14]. These observations suggested that a small RNA virus such as Enterovirus is the most likely candidate.

Taxonomy of the Enteroviruses is as follows: Family – Picornaviridae. Genus – Enterovirus. Species – e.g. poliovirus or coxsackievirus. Enterovirus is a plus single-stranded RNA virus. The virion diameter has been reported to vary from 25 nm to 30 nm, depending on the virus species and preparation methods for TEM. Purified virus particles possess an icosahedral capsid coat consisting of 60 structural subunits, also referred to as capsomeres, with one copy of four capsid proteins VP1-VP4 in each subunit.

In our study of coxsackievirus B and poliovirus infected cell cultures, apoptosis of supernatant cells was complete at 4 h. This is consonant with the findings by other authors of apoptosis early in the viral replication of poliovirus [15] and coxsackievirus B infected cells [16]. We observed that virus replication in infected cell cultures took place in both the nuclei and cytoplasm. Different stages of virus replication were found without endoplasmic reticulum association detectable by TEM. Large clusters of electron dense, small (22 nm) virus particles were observed both in the nuclei and the cytoplasm of infected cells. Scattered throughout the virus factories were slightly larger virus particles measuring approximately 28 nm. In addition, large (50 nm) intranuclear particles were observed associated with virus factories. TEM of the control cell cultures showed no virus particles. The finding of intranuclear clusters of 22 nm particles surrounded by particles measuring approximately 28 nm suggests that incomplete 22 nm virus particles are first assembled in clusters; they subsequently develop into 28 nm complete virus particles that are then secreted into the cytoplasm and consequently to the extracellular space.

The finding of small (27 nm) and large (50 nm) particles in the same cell in the in vitro cultures that were similar in size to the two types of VLP described in cases of EL could be accounted for by a single infecting virus. The assembly of 50 nm particles may be the result of an alternative pathway for Enterovirus replication. Otherwise, there may be two distinct virus populations, one acting as a helper virus in association with Enterovirus for its expression. This would also apply to the finding of two types of particle in the modern EL cases and the case of PEP. This poses a subject for a future study of Enterovirus infected cell cultures. The presence of a helper virus would go toward explaining the sporadic epidemiology of EL. The possibility that the 50 nm particles in the in vitro virus infected cell cultures were an external contaminant is unlikely due a) to the TEM finding that the control cell cultures were free of virus particles; b) the seed virus was isolated by end point dilution; c) the 50 nm particles were also found in the modern EL brain tissues.

An experimental model for helper virus in Enterovirus infection was described by Cords and Holland [17, 18]. In this study, synergy in virus replication was detected as the result of simultaneous infection in medium lacking in guanidine of cell cultures by two variants of polioviruses, one that was guanidine dependent, the other variant was guanidine independent. Replication of the guanidine dependent type was recovered by the concomitant replication of the independent type.

Being neurotropic, poliovirus and coxsackievirus B were the working hypotheses for the etiology of EL and therefore they were tested immunohistochemically [19]. The anti-poliovirus antibody staining was significantly positive to a high dilution with four classical EL specimens, and anti-coxsackievirus antibody staining was positive for three classical EL specimens, (Tables 1, 2). In case #91558, which was rated as being negative for the anti-coxsackie antibody, the small sample of tissue available did not include neurons; the neuropil and microglia stained moderately. In positively stained EL specimens, the cytoplasm and nuclei stained equally. In specimens that included cerebellar tissue, Purkinje cells stained strongly. This was supported by the finding of a loss of Purkinje cells in the two modern EL cases reported by neuropathology examination, and also in the classic EL cases reviewed by Anderson, Vilensky et al. [7]. The involvement of the cerebellum in EL is consistent with the clinical tremor in EL patients.

The anti-poliovirus and anti-coxsackievirus polyclonal antibodies had been titrated against their respective viruses at their institutes of origin. High titers for these antibodies were found but cross-reactivity was reported (See Methods). Our findings suggest that the anti-viral specificity of the antibodies is related to the disease condition of EL. Two monoclonal antibodies against conserved Enterovirus structures were tested. The results were non-significant because of excessive cross-reactivity with control specimens. The manufacturers test their antibodies but do not guarantee absence of reactivity against human tissues. Because of antigen degradation in fixed autopsy material, the antibody titer for the immunohistochemical test could differ from the reported neutralizing dose.

VLP were not observed in the neuropil of brain tissue in areas affected by EL, which nevertheless stained positively with the polyclonal antibodies and which showed marked disruption by TEM. Therefore the immunogold technique using those antibodies would not discriminate more clearly between the VLP and other tissue components than did conventional immunohistochemistry and therefore was not performed.

In this report, a viral hypothesis is entertained. An alternative hypothesis of autoimmunity for the cause of EL, based on the work of Dale, Church et al. [3] is described in Background. An editorial by Vincent [20] is paraphrased, which analyzed the autoimmunity hypothesis: - Dale identified autoantibodies by western blotting of soluble extracts of basal ganglia homogenate of human brain. The antibodies bound to several different polypeptide bands. Western blotting efficiently detects antibodies to non-conformational epitopes; it is likely to miss those potentially pathogenic antibodies that bind to conformational determinants. It would have been more informative to use a whole tissue or membrane preparation rather than a soluble extract. There are questions regarding the regional specificity since an analysis of different parts of the brain was not done. The term ‘antibasal ganglia antibody’ may be misleading. The presence of antibodies to different protein bands varying between the individual patients suggests they may be due to an immune mediated condition rather than causation. It is not yet clear whether the antibodies to the basal ganglia antigens” are crossreactive with streptococcus A antigens.

We deduce from our own data and that of Dale, Church et al. [3] that the autoimmune phenomena that they describe is secondary to an Enterovirus infection.

The electron microscopic, immunohistochemical and preliminary molecular findings make a prima-facie case for an Enterovirus etiology. Further study of the virology of Enteroviruses in cell culture is planned. TEM and immunohistochemistry of Enterovirus infected and control cell cultures will be carried out, physical purification of the two morphological forms of virus particles, and molecular biology of purified fractions of virus are projected.

The VLP found in EL tissue mirror the polio viral lifecycle in vitro[21, 22]. Salonen, Ahola et al. [21] reported that on entry the viral genome migrates to specific perinuclear sites where the complementary minus strand RNA and plus strand RNA are synthesized. The authors present evidence that “poliovirus replication complexes consist of clusters of vesicles of 70–400 nm in diameter, which after isolation are large rosette-like structures of numerous vesicles interconnected with tubular extensions…The rosettes can dissociate reversibly into tubular vesicles, which carry poliovirus nonstructural proteins on their surface. In poliovirus infected cells a continuous proliferation and loss of ER membranes takes place”. Dales et al. [23] also found that polio replication is associated with ER membranes. In Dales’ TEM observations of poliovirus infected cells, single membrane bound bodies similar to the tubular vesicles that were reported by Salonen were described, commencing 3 h after infection and by 7 h overwhelming the remaining cell structure. In the late stages, complete virus particles, 26–28 nm in diameter, were observed within the tubules with an amorphous matrix. The complete particles were finally assembled in crystalline arrays in the cytoplasm. Dales et al. also observed ‘viroplasm’ in the cytoplasm of cells 3 h after infection, and aggregates of ‘granules’ 17–25 nm in diameter in the cytoplasm, unassociated with membranes, which they assumed to be nascent virus particles. Their micrographs also illustrate large numbers of individual nascent cytoplasmic virus particles outside the aggregates. Dales did not report the intranuclear replication of poliovirus.

However, the importance of the cell nucleus in poliovirus replication is demonstrable. Bienz et al. [24] described the migration of poliovirus proteins into the host cell nucleus. Kawanishi [25] and Anzai and Ozaki [26] infected FL cells (derived from human amnion) with poliovirus at 28°C for 20 h and found intranuclear crystals of assembled poliovirus capsids without cores. Follett, Pringle et al. [27] showed that a number of RNA viruses, including poliovirus, yielded much less virus when enucleated cells were infected.

Lipofuscin bodies are often found in large numbers in neurons of EL brain. They stain acid fast with carbolfuscin and positively with PAS, indicating their origin or association with lysosomes. They are frequently found normally in brain from the age of 9 to patients late in life in increasing frequency and numbers [28, 29]. They are not specific indicators of viral infection. However, the large number of lipofuscin bodies found in the neurons of EL cases was abnormal for patients in the young age group that was typical for classical EL. They were also observed by light microscopy in the histopathological analysis of EL case #91558. Presumably they develop as a cellular reaction to virus pathogenesis.

During the EL epidemic, in vivo experiments were carried out supporting the hypothesis that EL was caused by a virus related to poliovirus. In 1921, Neustaedter, Larkin et al. [30] injected the brains of five macaque monkeys with a suspension of brain from a monkey killed by poliomyelitis. The polio infected brain suspension had been incubated with either convalescent serum from surviving EL patients, or with normal human serum or saline as controls. Monkeys injected with polio infected brain that had been incubated with EL serum were protected, whereas controls either died or were paralyzed. This result suggests that antibodies from EL patients neutralized poliovirus due to antigens common to poliovirus and the EL virus. Neustaedter’s experiment was similar to experiments a decade earlier demonstrating the neutralization of poliovirus by convalescent sera from polio patients. There is no reason to suppose that contemporary technical limitations invalidate this classical experiment.

More recently, persistent coxsackievirus B encephalitis was reported in both immunodeficient [31] and immunocompetent patients [32]. The observations that coxsackievirus B infection may persist in the central nervous system are significant for post-encephalitic parkinsonism, a chronic, presumably infectious syndrome. Evidence for chronic Enterovirus infection has been reported in the post-viral fatigue syndrome [33, 34]. Their observations are consonant with our finding VLP in brain in a case of PEP (see Figures 13, 14).

The rhinoviruses are members of the genus Enterovirus. They are the most frequent cause of the common cold. Infection in children may result in bronchiolitis and asthma. However, there are no reports of central nervous system involvement by rhinoviruses. It is unlikely, therefore, that they may be implicated in the etiology of EL. The rhinovirus capsid is icosahedral in structure, with a diameter of approximately 30 nm [35].

As part of the control study, TEM of brain from a patient with fatal H3N2 influenza was carried out. Intracellular influenza virus particles were found, presumably following an agonal surge of infection and breakdown of the blood brain barrier. No particles resembling VLP were present. This observation demonstrates that influenza virus particles, were they present, may be readily recognized by TEM in autopsy tissue of human brain.

This study is concordant with the conclusions of McCall et al. [5, 6] contesting the influenza theory of EL. Evidence is presented favoring an Enterovirus etiology of classical EL. It is supported by the TEM finding of small VLP in the cytoplasm and nucleus of neurons from classical and modern cases of EL and one case of PEP. The VLP were differentiated from ribosomes by the following findings. 1) The developing VLP were either in putative virus factories or nascent VLP particles attached to ER membranes. 2) There was nearly complete loss of ribosomes in normal control brain from the EL epidemic period in contrast to the predominant preservation of VLP in the cases of classical EL. 3) The diameter of the complete VLP observed by TEM was significantly greater (26.6 nm) and more variable than ribosomes in neurons from normal control brain (20.4 nm).

In two cases of modern EL and in one case of PEP examined, 50 nm intranuclear particles were found in addition to the small VLP. Similarly, in poliovirus and coxsackievirus B infected cell cultures, intranuclear aggregates were observed of 22 nm, 28 nm and 50 nm virus particles. The Enterovirus hypothesis is further supported by positive immunohistochemistry results employing polyclonal anti-poliovirus and anti-coxsackievirus antibodies on EL brain tissue.

Together with the TEM and immunohistochemical findings, preliminary RTPCR leading to sequence data of a 97-bp RNA fragment isolated from one case of epidemic EL strongly point to the genus Enterovirus as the cause of EL. Further information on the authors is available in Additional file 2.

The Sophie Cameron Trust, 107 St Thomas Road, Trowbridge, Wilts., BA 14 7LT. Email: Mail@sophiecamerontrust.org.uk The trust funded essential consumable supplies for Dr Dourmashkin’s research and contributed to the cost of publication.