Volume 235, Issue 2 p. 242-252
Invited Review
Free Access

Molecular biology, pathogenesis and pathology of mumps virus

Steven Rubin

Corresponding Author

Steven Rubin

Center for Biologics Evaluation and Research, Food and Drug Administration, Silver Spring, MD, USA

Correspondence to: Steven Rubin, Center for Biologics Evaluation and Research, Food and Drug Administration, 10903 New Hampshire Avenue, Silver Spring, MD 20993, USA. E-mail: [email protected]Search for more papers by this author
Michael Eckhaus

Michael Eckhaus

Division of Veterinary Resources, National Institutes of Health, Bethesda, MD, USA

Search for more papers by this author
Linda J Rennick

Linda J Rennick

Department of Microbiology, Boston University School of Medicine, MA, USA

Search for more papers by this author
Connor GG Bamford

Connor GG Bamford

MRC–University of Glasgow Centre for Virus Research, Glasgow, UK

Search for more papers by this author
W Paul Duprex

W Paul Duprex

Department of Microbiology, Boston University School of Medicine, MA, USA

Search for more papers by this author
First published: 17 September 2014
Citations: 157
No conflicts of interest were declared.
This article has been contributed to by US Government employees and their work is in the public domain in the USA

Abstract

Mumps is caused by the mumps virus (MuV), a member of the Paramyxoviridae family of enveloped, non-segmented, negative-sense RNA viruses. Mumps is characterized by painful inflammatory symptoms, such as parotitis and orchitis. The virus is highly neurotropic, with laboratory evidence of central nervous system (CNS) infection in approximately half of cases. Symptomatic CNS infection occurs less frequently; nonetheless, prior to the introduction of routine vaccination, MuV was a leading cause of aseptic meningitis and viral encephalitis in many developed countries. Despite being one of the oldest recognized diseases, with a worldwide distribution, surprisingly little attention has been given to its study. Cases of aseptic meningitis associated with some vaccine strains and a global resurgence of cases, including in highly vaccinated populations, has renewed interest in the virus, particularly in its pathogenesis and the need for development of clinically relevant models of disease. In this review we summarize the current state of knowledge on the virus, its pathogenesis and its clinical and pathological outcomes. Copyright © 2014 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Introduction

Before routine mumps vaccination programmes were introduced, 95% of adults had serological markers of exposure, with peak acquisition during childhood 1-4. Following the use of mumps vaccine in the USA in the late 1960s, disease incidence declined dramatically, and by the 1980s very few cases were reported. By 2001 the disease was on the verge of elimination, with < 0.1 cases/100,000 population reported 5, representing a 99.9% decrease in disease incidence compared to the prevaccination era. Similar success in mumps control was achieved in other countries through vaccination. However, within a few years of these historic lows, large, sporadic mumps outbreaks began to appear globally, involving a high percentage of persons with a history of vaccination 6-20. The reason for the lower than expected efficacy of mumps vaccines is a subject of much debate, ranging from waning immunity to the emergence of virus strains that might be capable of escaping immunity engendered by the vaccine. In addition to questions over vaccine efficacy, safety concerns have come to light following reports of meningitis linked to some vaccine strains used outside the USA. This has led to withdrawal of some vaccine strains and, in some cases, complete cessation of mumps vaccination. In Japan, for example, mumps vaccination was removed from the national immunization programme. Japan now has one of the highest rates of mumps among developed countries, with over a million cases reported annually 21, 22. Considering the numerous issues surrounding mumps, including ongoing outbreaks, a review of MuV pathogenesis is timely.

Humans are the only natural host of MuV. The disease is characterized by painful swelling of the parotid glands, but can involve numerous other tissues and organs, resulting in a wide array of inflammatory reactions, including encephalitis, meningitis, orchitis, myocarditis, pancreatitis and nephritis 23. Mumps is self-limiting, often with complete recovery within a few weeks of symptom onset; however, long-term sequelae, such as paralysis, seizures, cranial nerve palsies, hydrocephalus and deafness, can occur. The disease is rarely fatal and the lack of autopsy tissue limits opportunities to examine disease pathogenesis and pathology. Our current knowledge of MuV pathogenesis is therefore mostly based on animal studies, often following unnatural routes of infection. Consequently, the pathogenesis of the virus in humans remains in question. This review will summarize the current knowledge of the virus, its inferred pathogenesis, clinical manifestations, and the importance of an understanding of disease pathogenesis as a prerequisite to the development of safer and more efficacious vaccines.

Mumps virus

Mumps was first described by Hippocrates in the fifth century bc, in his first Book of Epidemics, but a viral aetiology was not demonstrated until the 1930s, when Johnson and Goodpasture fulfilled Koch's postulates by transferring the disease from experimentally infected rhesus macaques (Macaca mulatta), to children in his neighbourhood, using a bacteria-free, filter-sterilized preparation of macerated monkey parotid tissue 24, 25.

The virus, a member of the family Paramyxoviridae, is an enveloped particle containing a non-segmented negative strand RNA molecule of 15,384 nucleotides. Other significant paramyxoviruses that infect humans and livestock include measles virus, canine distemper virus, parainfluenza virus, Newcastle disease virus, respiratory syncytial virus and metapneumovirus. The encapsidated genome (Figure 1) contains seven tandemly linked transcription units, in the order: nucleo- (N), V/P/I (V/phospho-/I proteins), matrix (M), fusion (F), small hydrophobic (SH), haemagglutinin-neuraminidase (HN) and large (L) proteins 26, 27. The template for viral replication and transcription is the ribonucleoprotein (RNP) complex, which is composed of the negative-strand viral RNA encapsidated by N protein. The RNA-dependent RNA polymerase, a complex of the L and P proteins, acts as a replicase to copy the negative sense (–) RNA to a positive sense (+) RNA and as a transcriptase to generate mRNAs from the (–) RNP by entering at a single promoter at the 3′ end of the genome.

Details are in the caption following the image
MuV virion structure. (A) Thin sectioned transmission electron micrograph showing a typical MuV particle alongside (B) a schematic of the particle. The enveloped particles are pleomorphic, in the size range 100–600 nm. Within this structure lies the long, coiled electron-dense ribonucleoprotein (RNP), containing the MuV genome. Small spikes can be observed on the surface of the particle, corresponding to the viral HN and F glycoproteins. The same general features of the MuV particle are shown in the schematic (B). The envelope (blue lines) is studded with the HN (purple) and F (blue) glycoproteins and encases the viral RNP, made up of the RNA genome (3′–5′) in complex with N (yellow), P (orange) and L (gold) proteins. The M protein (red) interacts with the envelope, glycoproteins and the RNP. The V, I and SH proteins are expressed in infected cells, but are not thought to be incorporated within the virion. Photomicrograph courtesy of CDC/A Harrison and FA Murphy (http://phil.cdc.gov/phil/details.asp).

In infected cells the HN and F glycoproteins are transported through the endoplasmic reticulum and Golgi complex to the cell surface. The M protein is involved in localizing the viral RNP to regions of the host cell membrane expressing the F and HN glycoproteins, facilitating budding of the infectious virions from the infected cells 28, 29. The HN glycoprotein is responsible for attachment of the newly budded virus to neighbouring cells via its receptor, sialic acid, which is abundantly present on the surface of most animal cells. The HN glycoprotein, in concert with the F glycoprotein, mediates virus-to-cell fusion and cell-to-cell membrane fusion, facilitating virus spread. The SH protein is thought to play a role in evasion of the host antiviral response by blocking the TNFα-mediated apoptosis pathway 30, 31. This protein is not essential for virus replication, as demonstrated in studies with recombinant (r) MuVs engineered to lack the open reading frame encoding this protein 32. The V and I proteins are encoded by the same transcriptional unit that encodes the P protein 27, 33. Like the SH protein, the V protein is also involved in evasion of the host antiviral response, where it inhibits IFN production and signalling 34-36. The role of the I protein is unknown.

Clinical features, pathogenesis and pathology

Given the incidence of mumps in the pre-vaccine era, comparatively little is known about the pathogenesis of the disease. Much is inferred by comparison with related viruses, from experimental infection of laboratory animals, and from the clinical features and pathology of the disease in humans.

Initial infection: targeting of the upper respiratory tract epithelium?

MuV is transmitted via the respiratory route by inhalation or oral contact with infected respiratory droplets or secretions, as suggested by the aforementioned Johnson and Goodpasture study, and in a subsequent study by Henle and colleagues, who transmitted the disease to children by both oral and nasal routes of inoculation 37. MuV has been isolated from children with respiratory disease without parotitis 38 and has been detected by RT–PCR in nasal samples 39. The spread of mumps among persons in close contact also suggests this mode of transmission. Based on studies of other respiratory viruses, it is assumed that, following exposure, MuV infects the upper respiratory tract, but this has not yet been formally demonstrated. For measles virus, a related paramyxovirus, it was also assumed that the virus initially infects the respiratory epithelium, but this assumption proved to be incorrect when macrophages and dendritic cells in the lung tissue were shown to be the primary target cells infected with this virus 40.

Systemic spread: from epitheliotropic to lymphotropic?

Given the array of symptoms, it is clear that MuV is able to disseminate systemically in the body, which has led to the assumption that, following infection of the upper respiratory mucosa, the virus spreads to regional lymph nodes, resulting in viraemia during the early acute phase (Figure 2). However, despite the high frequency of extra-respiratory symptoms, virus has only rarely been detected in blood 41-43, even in experimentally infected animals 44. One possible explanation is the coincident development of MuV-specific humoral antibodies.

Details are in the caption following the image
MuV clinical presentation and pathogenesis. Mumps is a respiratory-spread, acute, inflammatory disease in humans, which causes a range of systemic symptoms. The incubation period is 2–4 weeks. Approximately one-third of infections are asymptomatic. The prodromal phase is characterized by non-specific, often mild symptoms, such as low-grade fever, headache and malaise. An early acute phase follows, likely representing spread of the virus from the respiratory tract and development of systemic symptoms, typically parotitis, which lasts from a few days to 1 week. During the established acute phase, orchitis, meningitis or encephalitis may appear. Symptoms usually resolve within 2 weeks, coincident with the development of a MuV-specific humoral response. Long-term complications and death are rare

Approximately one-third to one-half of MuV infections are asymptomatic or result in only mild respiratory symptoms, sometimes accompanied by fever 45-49. The hallmark of mumps is salivary gland swelling, typically the parotid glands, which forms the basis of a clinical diagnosis. Parotitis is usually bilateral, developing 2–3 weeks after exposure and lasting for 2–3 days, but it may persist for a week or more in some cases 37, 50, 51. Submaxillary, submandibular and sublingual glands can be involved, but rarely as the only manifestation of mumps. Viral replication in the parotid gland results in perivascular and interstitial mononuclear cell infiltration, haemorrhage, oedema and necrosis of acinar and epithelial duct cells 52. Serum and urine amylase levels may be elevated as a result of inflammation and tissue damage in the parotid gland 53. Virus is excreted in the saliva from approximately 1 week before to 1 week after the onset of salivary gland swelling 37, 54, 55. MuV has also been identified in the saliva of asymptomatic persons 37. Coupled with excretion of virus up to 1 week before symptom appearance, this may explain some of the difficulties in controlling mumps outbreaks.

Orchitis, which is typically unilateral, is the most common extra-salivary gland manifestation of mumps. It occurs in approximately 10–20% of infections in postpubertal men 46, 56, 57. MuV has been recovered from semen and the testis, suggesting that epididymo-orchitis is the result of direct infection of testicular cells 58, 59. However, an indirect immune-mediated mechanism has also been postulated 60. Both Leydig and germ cells are involved, associated with reduced levels of testosterone production 61-63. Necrosis of acinar and epithelial duct cells is evident in the germinal epithelium of the seminiferous tubules of the testes. Orchitis is almost always accompanied by epididymitis and fever, all resolving within 1 week. Atrophy of the involved testicle occurs in approximately half of cases and can be associated with oligospermia and hypofertility, but rarely sterility 58, 62, 64, 65. Mastitis and oophoritis (manifesting as pelvic pain) occurs in 5–10% of mumps cases in postpubertal women 12, 46, 66. Oophoritis has been associated with infertility 67 and premature menopause 68, but such cases are extremely rare.

Virus frequently disseminates to the kidneys, as suggested by the frequency of viruria during the established acute phase of the disease (Figure 2) 69, 70. Epithelial cells of the distal tubules, calyces and ureters appear to be primary sites of virus replication 52. Kidney involvement in mumps is almost always benign, but cases of severe interstitial nephritis have been reported. In such cases, renal biopsy or postmortem necropsy show evidence of immune complex deposition, interstitial mononuclear cell infiltration and fibrosis, oedema and focal tubular epithelial cell damage 71-73. Pancreatitis, diagnosed as severe epigastric pain and tenderness, has been reported in approximately 4% of cases 6, 12, 74. There are conflicting reports on the association between mumps pancreatitis and diabetes mellitus 75-79 and it is unclear whether there is a causal link.

CNS involvement: from lymphotropic to neurotropic

MuV is highly neurotropic, with evidence of central nervous system (CNS) involvement in up to half of all cases of infection, based on pleocytosis of the cerebrospinal fluid 48, 80-82. Symptomatic CNS infection is less common, but significant. Meningitis occurs in approximately 5–10% of cases and encephalitis in < 0.5%. Although these are small percentages, MuV was the leading cause of encephalitis in the USA until 1975, when mumps-containing vaccine gained widespread use 83. In unvaccinated populations, mumps continues to account for a high percentage of viral encephalitis cases 46, 84, 85. Little is known of the CNS pathology, since the disease is rarely fatal. Of the few postmortem cases examined, the pathology includes oedema and congestion throughout the brain with haemorrhage, lymphocytic perivascular infiltration, perivascular gliosis and demyelination, with relative sparing of neurons. These latter observations suggest that in some cases of mumps encephalitis the inflammation stems from a para-infectious process. However, virus can be recovered from CSF early in the course of meningitis 86, 87, as well as from brain tissue in some cases of mumps encephalitis.

Experimental infection in rodents suggests the virus enters the CSF through the choroid plexus, or possibly via transiting mononuclear cells during viraemia. Based on animal data, once in the CSF, virus appears to be carried throughout the ventricular system, resulting in virus replication within ependymal cells that line the ventricles (Figure 3A) 44. From these locations, virus can penetrate into the brain parenchyma, often infecting pyramidal cells in the cerebral cortex and hippocampus 88. The infected ependymal epithelia become inflamed, lose their cilia, degenerate and collapse into the CSF (Figure 3B), a postulated cause of the aqueductal stenosis that is believed to be responsible for the occurrence of hydrocephalus, typically of the lateral and third ventricles, a common outcome in intracerebrally injected animals 89-93. Mumps hydrocephalus has been reported in humans, most often presenting as obstruction of the cerebral aqueduct with dilatation of the lateral and third ventricles. However, obstruction of the foramen of Monro between the lateral and third ventricles, or obstruction of the foramina of Magendie and Luschka between the fourth ventricle and the subarachnoid space, have also been reported 94-98. The finding of ependymal cell debris in the CSF of mumps patients 94, 96, 99 suggests that the pathogenesis of hydrocephalus in experimentally infected animals is similar to the mechanism of hydrocephalus in humans. However, hydrocephalus has been observed before, or in the total absence of, canal obstruction 88, 91, 100, 101, indicating that such events could be a secondary consequence of external compression by surrounding oedematous tissue and not causally related to the pathogenesis of hydrocephalus.

Details are in the caption following the image
MuV infection of the rat brain. The most prominent neuropathological outcome following MuV intracranial inoculation in small animal models (hamsters, rats) is enlargement of the lateral and third ventricles, ie hydrocephalus, which has also been reported in cases in humans. The cause of hydrocephalus is postulated to be denuding of virus-infected ependymal cells lining the ventricles. (A) Sagittal section of rat brain tissue immunohistochemically stained for the MuV nucleoprotein, showing extensive infection of the ventricular ependymal cells (green foci). (B) Approximately 3 weeks later, ependymal cell loss is evident in comparison to the well-preserved ependymal cell architecture in rats injected with the Jeryl Lynn vaccine strain (C); haematoxylin and eosin (H&E) stain

Deafness has been reported in approximately 4% of mumps cases and is the most frequent cause of acquired unilateral sensorineural hearing loss in children. Hearing loss is typically unilateral and transient, but can be permanent 102-106. Pathological findings include lesions and degeneration of the stria vascularis, tectorial membrane and organ of Corti 107, 108. MuV infection of the CSF has been implicated in the pathogenesis of deafness in mumps, given the detection of the virus in perilymph, which freely communicates with the CSF 109. This is also supported by animal studies, where instillation of the virus into the CSF has resulted in infection of the cochlea 110. However, deafness does not occur any more frequently in patients with meningitis or encephalitis than it does in patients lacking signs of CNS infection, suggesting that CSF may in fact not be involved in the pathogenesis of deafness. An alternative explanation could be that virus infects the inner ear via a haematogenous route, ie that mumps labyrinthitis occurs as a consequence of viraemia. This is supported by studies in guinea pigs following intravascular inoculation of the virus 111, and clinical findings by Lindsay 107 and Mizushima and Murakami 112 suggesting 'viral endolymphatic labyrinthitis' in the pathogenesis of mumps deafness in humans. Hearing loss caused by indirect effects of virus infection (eg immune-mediated damage) have also been suggested 113. MuV was also identified in the vestibular ganglia in experimentally infected animals 110, which likely also occurs in humans and explains vestibular symptoms, such as vertigo, which often present in cases of mumps deafness 114, 115.

Based on electrocardiographic abnormalities in mumps patients, MuV likely infects cardiac tissue 116. While this is rarely symptomatic, interstitial lymphocytic myocarditis and pericarditis have been reported 117, which can lead to endocardial fibroelastosis 118. MuV has also been identified in cardiac muscle from patients with these disorders. Clinically apparent cardiac complications are rare, but can be serious 116, 119, 120.

Other rare complications include cerebellar ataxia 121, 122, transverse myelitis 123, 124, ascending polyradiculitis 125, a poliomyelitis-like disease 126, 127, arthropathy 128, 129, autoimmune haemolytic anaemia 130, 131, thyroiditis 132, 133, thrombocytopenia 134, 135, hepatitis 136, 137 and retinitis and corneal endotheliitis 138-140.

Transplacental transmission of the virus has been demonstrated in non-human primates 141 and is suggested by the isolation of virus from the human fetus following spontaneous or planned abortion during maternal mumps 142-144. Aborted fetal tissue from such cases has been found to exhibit a proliferative necrotizing villitis with decidual cells containing intracytoplasmic inclusions 145. Virus has also been isolated at birth from infants born to women with mumps 146 and from breast milk 147, but few cases of perinatal mumps have been described 146, 148 and it is not clear whether breast milk was responsible for these cases. Mumps virus does not appear to cause congenital malformations 149. The major morbidity from mumps is from complications of meningitis, encephalitis and orchitis. The case fatality ratio is 1.6–3.8/10,000 150, 151, with most fatalities occurring in persons with encephalitis.

Animal models and molecular determinants of MuV pathogenesis

Historically, the most widely used animal models of mumps have been the hamster and the monkey, and information from these models serves as the basis for much of our current understanding of MuV pathogenesis and disease. However, the relevance of findings in these models for humans is questionable, given the use of unnatural routes of inoculation (eg intracranial, intraperitoneal or intravenous) and the inability of these models to clearly and reliably discriminate strains that are attenuated for humans from wild-type, virulent strains 152-158. In the one study where monkeys (rhesus macaques) were inoculated via natural routes (intranasal and intratracheal), only wild-type virus was tested 159. Nonetheless, this study demonstrated the potential to identify sites of early and late MuV replication, which supports further evaluation of this model. Mice and ferrets were also explored as model systems; however, virus replication in these species is self-limiting, making them poor candidates for pathogenesis studies 159-164.

A key advance in the study of MuV pathogenesis was the development of a rat model predictive of the neurovirulence potential of MuV strains for humans. In this model, 1 month after intracerebral injection of virus into newborn Lewis rats, brains are removed and evaluated for virus-induced hydrocephalus (Figure 4), the severity of which correlates with the neurovirulence potential of the virus for humans 165. With the advent of plasmid-based reverse genetics systems for MuV, it became possible to examine molecular determinants of virulence, and thus gain a better understanding of virus factors that influence pathogenicity. The first such study was published by Lemon et al 166, who generated Jeryl Lynn vaccine strain-based viruses expressing genes derived from the Kilham MuV strain, a hamster brain-adapted laboratory strain. Of the single gene replacements assessed (M, F, SH and HN), only the F gene was found to significantly increase the neurovirulence potential of the highly attenuated Jeryl Lynn strain 166. However, in a subsequent study using a different virulent MuV strain, 88-1961, Sauder et al 167 found the F gene to have no biologically meaningful effect on the neurovirulence potential of Jeryl Lynn. MuV strain-specific molecular determinants of virulence are also highlighted in other studies. For example, Xu et al 168 identified the SH gene as a virulence factor for the wild-type IA MuV strain, whereas Malik et al 32 found no such role for the SH gene in 88-1961 virulence. Of additional interest, in the Sauder et al study, the Jeryl Lynn genes found to neuro-attenuate the 88-1961 strain (eg M, HN and L), when derived from the 88-1961 strain, did not meaningfully increase the neurovirulence of the Jeryl Lynn virus. Thus, not only do genes that influence virulence of one strain often not affect virulence of another strain, but genes involved in neurovirulence are not necessarily involved in neuroattenuation. Taken together, these results raise doubt as to the prospect of identifying broadly applicable genetic determinants of virulence and attenuation. An alternative, or complementary, approach may be to examine differences between virulent and attenuated strains in terms of sites of primary replication and spread, following infection via natural routes in disease-relevant animal models. To this end, new in vitro and in vivo model systems that recapitulate the diverse features of a natural infection are needed.

Details are in the caption following the image
Hydrocephalus severity in MuV-infected rats. (A) H&E-stained sagittal sections of brain from a representative 30 day-old rat injected with a wild-type (WT) MuV isolate as a newborn (top), compared to a rat similarly injected with the highly attenuated Jeryl Lynn (JL) vaccine strain (bottom). (B) T1 weighted gradient-echo image from MRI of the same brains as in (A) (upper left corner and lower right corner), compared to brain from a rat injected with an insufficiently attenuated vaccine strain, Urabe-AM9 (Ur, lower left corner) and an uninfected rat brain (0, upper right corner). Note that the severity of hydrocephalus tracks with the virus strain's neurovirulence potential for humans. (C) Assembled three-dimensional rendering of MRI slices represented in (B), showing ventricular volume (blue, wild-type; pink, Urabe-AM9)

Conclusion

As a re-emerging pathogen, with concerns over vaccine safety and efficacy, elucidation of mechanisms of MuV pathogenesis is of paramount importance, as this information will help direct the development of improved vaccines. The utilization of existing reverse genetic systems alongside the generation of new, clinically relevant systems and the development of robust animal models for other aspects of the disease will allow a more complete understanding of disease. This review has summarized our current understanding of MuV clinical disease, pathology, and how this relates to viral pathogenesis. However, a number of areas are evidently poorly understood and important questions remain (Figure 5). These include the elucidation of the target cell tropism throughout an infection, the mechanisms by which MuV establishes a systemic infection and the basis of neurotropism. Determination of these likely requires the establishment of a primate model of MuV infection, similar to what has been achieved with measles. The measles system, exploiting a fluorescent reporter-expressing wild-type MV in a clinically-relevant macaque (Macaca fascicularis) model has facilitated the elucidation of key features of measles virus pathogenesis relevant to disease, transmission and immunity 169. Applying what has been learned from the measles model to mumps will provide an ideal basis to understanding determinants of MuV pathogenesis.

Details are in the caption following the image
A number of important questions remain unresolved regarding MuV pathogenesis. This is of particular relevance to renewed efforts towards development of a more efficacious MuV vaccine, in light of the resurgence of mumps in vaccinated populations. The classic method of virus attenuation is extensive blind passage in vitro. While this often leads to the desired effect of a loss of virulence and reactogenicity, it can also lead to loss of immunogenicity and efficacy. Clearly, a more rational approach to virus attenuation is needed, and understanding the natural pathogenesis of the infectious agent is a prerequisite to any such endeavour. This figure highlights this issue, showing our current assumptions of pathogenesis (black text) and unresolved questions (red text)

Acknowledgements

This work was supported by the NIH R01 AI105063 to WPD and by the Oak Ridge Institute for Science and Education through an interagency agreement between the US Department of Energy and the US Food and Drug Administration. We wish to thank Christian Sauder and Laurie Ngo (FDA, Silver Spring, MD) for critical reading of the manuscript, and Jan Johannessen (FDA, Silver Spring, MD) for providing the MRI rat brain images. We dedicate this work to the memory of the late Dr Philip Snoy, DVM, Director, CBER Veterinary Services, respected scientist, admired leader and treasured friend, who laid much of the groundwork towards development of improved preclinical MuV neurotoxicity tests.