Blue Moon Neurovirology: The Merits of Studying Rare CNS Diseases of Viral Origin
- First Online:
- Cite this article as:
- O’Donnell, L.A. & Rall, G.F. J Neuroimmune Pharmacol (2010) 5: 443. doi:10.1007/s11481-010-9200-4
- 125 Views
While measles virus (MV) continues to have a significant impact on human health, causing 150,000–200,000 deaths worldwide each year, the number of fatalities that can be attributed to MV-triggered central nervous system (CNS) diseases are on the order of a few hundred individuals annually (World Health Organization 2009). Despite this modest impact, substantial effort has been expended to understand the basis of measles-triggered neuropathogenesis. What can be gained by studying such a rare condition? Simply stated, the wealth of studies in this field have revealed core principles that are relevant to multiple neurotropic pathogens, and that inform the broader field of viral pathogenesis. In recent years, the emergence of powerful in vitro systems, novel animal models, and reverse genetics has enabled insights into the basis of MV persistence, the complexity of MV interactions with neurons and the immune system, and the role of immune and CNS development in virus-triggered disease. In this review, we highlight some key advances, link relevant measles-based studies to the broader disciplines of neurovirology and viral pathogenesis, and propose future areas of study for the field of measles-mediated neurological disease.
Keywordsmeasles virusneuronSSPECNS infection
canine distemper virus
central nervous system
green fluorescent protein
RNA-dependent, RNA polymerase protein
lymphocytic choriomeningitis virus
mouse hepatitis virus
measles inclusion body encephalopathy
signaling lymphocyte activation molecule
subacute sclerosing panencephalitis
signal transducer and activation of transcription
yeast artificial chromosome
A current view of the impact of measles virus on human health
Before the introduction of an effective live-virus vaccine in 1963, measles virus (MV) was a major cause of infant mortality throughout the world. The vaccine has been remarkably successful in preventing acute infections, resulting in an estimated 95% decrease in the incidence of infection in the US within 5 years of its introduction. While the worldwide incidence of infection continues to decline, largely due to improving vaccine coverage, the World Health Organization (2009) estimates that 150,000–200,000 people still die each year of complications from this infection.
As with most human pathogens, healthcare issues and the challenges to vaccination and pathogen eradication differ depending on what part of the world is being considered. In resource-poor regions, MV remains the leading cause of vaccine-preventable deaths in children (Centers for Disease Control and Prevention (CDC) 2006; Moss and Griffin 2006). The substantial impact of MV in developing countries can be attributed to multiple factors, including: lack of vaccination, inoculation with an unintentionally inactivated vaccine (due to lapses in the “cold chain” necessary to keep the attenuated virus efficacious), or vaccination prior to the waning of maternal antibodies. These problems, relevant as much to the public health infrastructure of the impacted countries as they are to viral biology, are extraordinarily challenging to address and, thus, regrettably, the morbidity and mortality toll of MV is unlikely to substantively improve until concomitant changes in public health policy occur.
In contrast, resource-rich countries such as the United States typically have high vaccination coverage (>90%) and even those few individuals who are not vaccinated benefit from herd immunity that limits access of the virus into the community. However, countries with historically high vaccination rates are facing unique problems as well. While many perceive measles to be a disease of the past, the recent and alarming wave of measles infections in the US and Europe have resurrected valid fears about MV susceptibility in typically well-protected communities (Editorial Team 2008). In 2008, the CDC reported 131 cases of MV in the US, the highest since 1996 (Centers for Disease Control and Prevention (CDC) 2008). Virtually all of the recent acute MV cases have occurred in children who were not vaccinated; of these, many were not vaccinated due to parental concerns about the perceived association of vaccines (measles in particular) with childhood autism. This controversy was initially fueled by a study published in Lancet in 1998 that reported an association of autism diagnosis with the presence of MV RNA in the gut (Wakefield et al. 1998). Despite its eventual retraction in 2010 (by the journal) and abundant studies that refuted any association between vaccination and autism, many families still harbor concerns. Though parents must make the choices they deem most appropriate for their children, the decision to delay or decline vaccination has imperiled the critical threshold needed for effective herd immunity, and—predictably—outbreaks of MV have occurred in many communities. At the very least, this has been a telling example that complete eradication of a pathogen requires both efficacious interventions (such as vaccines) as well as a community that is willing to use them. In that respect, pathogen eradication is as much a matter of marketing, public discourse, and politics, as it is development of a protective vaccine.
Measles virus biology and cellular receptors
A number of comprehensive reviews (Moss and Griffin 2006; Rima and Duprex 2009) have recently been published concerning the replication and pathogenesis of measles; thus, the basics of measles biology and disease will be reviewed only briefly here. MV is a prototypic paramyxovirus, consisting of an enveloped, negative-strand, ∼16-kb RNA genome encoding eight proteins. These proteins include replication factors [RNA dependent, RNA polymerase (L) and phosphoprotein (P)], structural proteins [nucleoprotein (N), matrix (M), hemagglutinin (H), and fusion (F)], and two accessory proteins that play a role in pathogenesis and replication (C and V; Poole et al. 2002; Rodriguez et al. 2003a). The viral ribonucleoprotein (RNP), formed within the cytoplasm of infected cells, consists of the viral genome complexed with N, P, and L. The classical view of MV spread is that, as the virus buds from the cell, it acquires its envelope, including H and F from the plasma membrane. The M protein is required to correctly transport H and F to the plasma membrane (Naim et al. 2000) and may also play a key role in bringing the RNP into physical approximation with the envelope proteins (Manie et al. 2000; Vincent et al. 1999). Infection of new cells is initiated when MV-H on the virion (or the infected cell) attaches to a cellular receptor on a susceptible cell. This, then, triggers exposure of a fusogenic domain on MV-F, resulting in fusion between the virus and host cell membranes or between an infected cell and an adjacent uninfected cell.
Mice do not possess functional MV receptor homologues, and thus are not susceptible to infection by wild-type or vaccine strains. With the identification of two human MV receptors: CD46 in 1993 (Dorig et al. 1993; Naniche et al. 1993) and SLAM/CD150 in 2000 (Tatsuo and Yanagi 2000), it became possible to develop transgenic mice expressing these receptors, with the aim of establishing permissive mouse models (reviewed in Manchester and Rall 2001). As described in greater detail below, such models have been invaluable in defining key elements of replication and pathogenesis for this otherwise human-restricted pathogen.
Following exposure of a naïve individual to measles by an aerosol route (e.g., droplets from a sneeze), there is a fairly long incubation period (5–10 days) prior to the appearance of any symptoms, during which time the individual is able to spread the infection to others. This period culminates in signs of sickness, including fever, cough, and congestion, that are similar to other respiratory infections. However, unlike most respiratory-restricted infections, other symptoms emerge during MV infection, including a characteristic maculopapular rash and the early appearance Koplik's spots in the oral mucosa. Resolution for most infected individuals occurs without consequence, usually within 10 days after the first signs of sickness. Immunity to MV is thought to be life-long.
Two other sequelae associated with MV exposure are immunosuppression and central nervous system (CNS) disease. Of these two, the vast majority of MV-related deaths are attributable to transient immunosuppression, which allows for rapid colonization and unrestricted growth of opportunistic infections. In regions of the world where sanitation is absent or inadequate, unchecked growth of bacterial or parasitic pathogens can precipitate illnesses such as diarrhea that, without medical intervention, can be fatal. Additionally, pneumonia is a common and equally serious consequence of infection. The observation that MV infection impinges on immunity to other pathogens has been appreciated since the beginning of the twentieth century, but only in the past two decades have we begun to discern the mechanism by which this occurs (comprehensively reviewed recently by Schneider-Schaulies and Schneider-Schaulies 2009). Multiple, nonmutually exclusive hypotheses exist to explain how this virus (which maximally infects 2% of T lymphocytes) can result in such widespread immunosuppression. These well-supported theories include: MV blunting of interleukin 12 production by professional antigen-presenting cells, skewing the subsequent T cell response away from the desired T helper 1 (Th1) type toward the less useful Th2 type (Hahm et al. 2007; Karp et al. 1996; Yu et al. 2008); sequestration of key interferon signaling molecules in the cytoplasm, preventing them from entering the nucleus and binding their promoter targets (Palosaari et al. 2003; Ramachandran et al. 2008); viral protein-mediated inhibition of naïve lymphocyte proliferation (Schlender et al. 1996; Schnorr et al. 1997); and infection of bone marrow stromal cells that disrupts maturation of “prelymphocytes” (Manchester et al. 2002). In humans, MV infection is associated with a marked drop in CD4+ and CD8+ T cells, potentially due to the inhibition of T cell proliferation and cell cycle progression (Naniche et al. 1999; Niewiesk et al. 1999; Schnorr et al. 1997), that is ultimately recovered after the primary infection (Ryon et al. 2002) and a dampened response of peripheral blood lymphocytes to antigenic stimulation ex vivo (Borrow and Oldstone 1995; Hirsch et al. 1984). Finally, a possible role for regulatory T cells has been proposed based upon a mouse model of MV infection, which demonstrated an increased frequency of these cells in the periphery and the CNS, accompanied by a decrease in the hypersensitivity response (Sellin et al. 2009). It is likely that many, if not all, of these strategies are operative, underscoring the myriad ways by which this deceptively simple virus can frustrate the mammalian immune response to gain a replicative advantage.
Among the survivors of measles infection, a small percentage of individuals develop neurological sequelae that can lead to mortality weeks to months to years after the initial exposure. Typically, there are three diseases that are attributable to MV infection in the CNS that differ in their timing, frequency, and background of the infected host. Post-infectious encephalomyelitis (PIE) or acute disseminated encephalomyelitis occurs in approximately one of 1,000 measles cases and typically affects children and adolescents (Miller 1964). Symptoms of PIE include seizures, motor and sensory defects, and ataxia, typically begin 5–14 days after the primary MV infection has resolved, though they can be delayed weeks to months after the primary infection. PIE is characterized by demyelination, with or without hemorrhaging and perivascular macrophage infiltration. There is little evidence for MV infection in the brain during PIE, as MV antigen and RNA are undetectable by immunohistochemistry or in situ hybridization, respectively (Gendelman et al. 1984; Moench et al. 1988; Norrby and Kristensson 1997). PIE may be an autoimmune disease, as patients have elevated immune responses to myelin basic protein (Gendelman et al. 1984; Johnson et al. 1984). PIE has a mortality rate of approximately 25%, with survivors demonstrating persistent neurological deficits (Chardos et al. 2003; Schwarz et al. 2001). Currently, there is no effective treatment.
Measles inclusion body encephalitis (MIBE) is a rare CNS complication following MV infection in immunocompromised patients including those receiving immunosuppressive drugs (Colamaria et al. 1989; Johnson 1998; Poon et al. 1998). Neurological symptoms, including seizures, motor deficits, and cognitive changes, appear 3–6 months after primary MV infection (Johnson 1998; Perry and Halsey 2004). MIBE is characterized by the presence of inclusion bodies within neurons, astrocytes, and oligodendrocytes, with accompanying neuronal loss. Both MV antigen and RNA are detectable within brain samples, and virus can be directly isolated from the brains of MIBE patients (Baczko et al. 1988; Johnson 1998). However, no apparent inflammation is present and there is an absence of neutralizing antibodies in the serum and cerebrospinal fluid (CSF; Rima & Duprex 2005; Weissbrich and terMeulen 2003). Death occurs in approximately 80% of MIBE cases within days to weeks of the initial neurological symptoms.
Subacute sclerosing panencephalitis (SSPE) is a slow, progressive neurological disease that appears a mean time of 6–10 years after primary MV infection (Dubois-Dalcq et al. 1976; Modlin et al. 1977). SSPE typically develops in patients under 20 years of age, though adult onset is possible in rare cases (Singer et al. 1997). Symptoms of SSPE commonly begin with personality changes and subtle cognitive losses before progressing to myoclonus, seizures, dementia, coma, and death over a period of weeks to years after the initial onset of neurological symptoms. Unlike PIE and MIBE, SSPE is associated with extremely elevated levels of measles-specific antibodies in the CSF and serum (Ebinger and Matthyssens 1971). Neurons and oligodendrocytes are the main target cell for MV in SSPE brains, though MV infection of astrocytes, endothelial cells, and infiltrating lymphocytes has been noted (Allen et al. 1996; Kirk et al. 1991). Importantly, infectious MV cannot be recovered from SSPE tissues, implicating cellular or viral changes that affect the viral life cycle. While MV RNA is present in SSPE brains, the mechanism for SSPE pathogenesis is unknown, though SSPE is marked by neurodegeneration, astrogliosis, and markers for oxidative stress in the brain (Aydin et al. 2006). Treatments such as ribavirin and interferon have produced conflicting results in SSPE patients, and there is currently no standard treatment protocol for SSPE other than supportive care.
Viruses in the brain: what we know from clinical studies
How common are neuropathological sequelae for other viruses that can target the CNS? While death due to virus-induced neuropathology is uncommon, many human viruses besides MV have the potential to infect CNS parenchymal cells, including polio, influenza, mumps, rabies, West Nile, and some herpes viruses. As with measles, in those rare cases when acute, virally mediated CNS disease occurs, the prognoses are almost always poor. The severity of these diseases is often due to inflammation, resulting in encephalitis and/or meningitis (reviewed in Rall 1998). Perhaps to moderate these potentially deleterious effects of robust immunity, the CNS has safeguards that collectively restrict both the access and function of lymphocytes within the parenchyma (once called “immune privilege,” but now recognized as not simply “less” immunity, but rather as a distinct way to regulate immune responses in this critical tissue). These anatomical and biochemical properties include the blood–brain barrier, the absence of lymphatic drainage, the paucity of major histocompatibility complex class I expression on most neurons, and elevated levels of immunosuppressive molecules such as gangliosides (reviewed in Rall and Oldstone 1997). Together, these CNS features restrict immune cell access into the CNS under normal conditions and may favor noncytolytic clearance strategies during pathogenic challenge (Binder and Griffin 2001; Parra et al. 1999; Patterson et al. 2002).
Fortunately, for most individuals infected with potentially neurotropic viruses, the infection is resolved before neuropathogenesis can occur. This may be attributable to resolution of the infection before viral access to the CNS can be achieved. However, some evidence suggests that this simple explanation may not uniformly be the case. A study of tissue specimens from aged individuals who died of nonviral (and non-CNS) causes revealed that a large proportion of brain biopsies contained MV RNA (Katayama et al. 1995), implying that —at least for MV— asymptomatic, quiescent infections can be established in the CNS. While these studies require validation with more sensitive methods that are now available, they imply that viral neuroinvasion and neuropathogenesis are not inexorably linked, and the goal of the host immune response may not be to clear the infection altogether, but rather to keep the persisting virus from reactivating or spreading.
In addition to acute disease, neurotropic viruses can also contribute to chronic CNS diseases, similar to the long-lasting MV CNS diseases described above. Examples include: spongiform encephalopathies caused by some lentiviruses, chronic neurodegeneration following Borna disease virus infection, post-influenza encephalitis, and mumps meningoencephalitis (Johnson 1998). In addition, some have speculated that chronic CNS diseases of unknown etiology, including schizophrenia, amyotrophic lateral sclerosis, and multiple sclerosis may have a viral component or trigger (Berger et al. 2000; de la Torre et al. 1996; Johnson 1998). While a formal association between these illnesses and a viral agent requires further study, it is certain that we do not yet fully appreciate the degree to which the CNS is influenced by persistent viral infections.
The issue: does the small number of cases of CNS sequelae following MV infection justify efforts to study these diseases?
While CNS diseases due to MV infection are almost unilaterally fatal, they are also notably rare, likely affecting fewer than one out of 10,000 acutely infected persons. However, these estimations almost surely under-represent the neurological impact of MV, since the latency between primary infection and CNS disease is so long and records are often not sufficiently detailed to connect CNS disease with a prior MV exposure. Moreover, individuals who may have eventually developed CNS sequelae may die earlier of other causes, including malnutrition or other infectious agents (reviewed in Rall 2003). Even so, in weighing the impact of MV on human health, these CNS diseases represent a small minority of the total MV burden. Why, then, is substantial effort invested in studying them and the basics of MV growth in the brain?
The point of view of this review is that the study of rare diseases can provide fundamental insights that may be more broadly applicable to human disease. In particular, the establishment of animal models and cell culture approaches to explore virus–neuron and virus–immune response interactions has revolutionized the study of MV biology within the brain. We will focus on three areas of progress in the remainder of this review. First, the unique viral life cycle adopted by MV in CNS neurons of mice (and potentially, of affected patients) offers a window into how viruses adjust their replication and spread within cells in the brain, perhaps leading to evasion of the immune response. Secondly, it is clear from MIBE patients and animal models that a competent, mature immune response is needed to control MV in the brain. What is surprising is the widespread utilization of noncytopathic strategies of viral control, chiefly through the use of interferon gamma (IFNγ). Finally, MV is overwhelmingly an infection of children; and animal models of MV infection mirror the age-dependent susceptibility. By studying MV in the brain, we are able to ask questions about interactions between the immature immune system and a developing brain and to examine why viral clearance fails predominately in children.
Basic principle 1: the viral life cycle is not monolithic and cell-specific influences on the virus can affect pathogenesis
The rarity of deaths ascribed to MV, and the daunting challenge of isolating brain tissues from these individuals of sufficient quality for subsequent analysis, continue to pose challenges to uncovering the basis of MV-mediated neuropathology. Of the small number of tissues that have been available, a few major points have emerged. First, the brains of SSPE patients show dramatic and extensive pathology, sometimes described in the clinical literature as “decorticated” (Anlar 1997). Blood–brain barrier permeability breaches, glial activation, inflammation, and cellular damage are all noted as pathologic hallmarks, though unraveling which of these neuropathological events are initiating (that is, directly attributable to viral infection or immunopahology) and which are secondary to the primary lesions is not possible to ascertain from these end-stage clinical specimens. A further challenge is deducing the state of the virus in these brains over the long period between initial infection and subsequent death. Is the virus quiescent for years, at which point some reactivation event (similar, perhaps, to reactivation of a latent herpes virus infection) triggers rapid replication and precipitous disease, similar to reactivation of a latent herpesvirus infection? Or, is the virus replicating at a slow but persistent rate, and only once a particular brain region or some threshold number of cells are infected do symptoms appear? While we do not yet know the answer, either scenario (MV latency or glacial replication rate) would suggest a major difference from the canonical view of how this virus replicates.
One intriguing observation obtained from human studies was the frequency of clustered mutations in MV genomes that were sequenced from brain specimens. Extensive point mutations, affecting as much as 2% of the genome (Cattaneo et al. 1988) and occurring in clusters, particularly in the F and M genes, were consistently identified in SSPE and MIBE isolates (Billeter et al. 1994). In one MIBE case, 50% of the uridine residues within the M gene were mutated to cytosines (Cattaneo et al. 1988). These mutations lead to protein truncations, elongations, or nonconservative amino acid substitutions (Schmid et al. 1992). A general consensus at the time these reports were published was that these mutations, in some way, conferred a replicative advantage to the virus, perhaps making this virus more “neuroadapted” or neurovirulent. The advent of reverse genetics, which allowed mutations to be selectively introduced into wild-type strains, was a major step in discerning what role these mutations played in viral growth. Interestingly, when a wild-type M protein was replaced with a mutation-laden SSPE M protein, the virus was still infectious in neurons (Patterson et al. 2001), though the lack of a functional M protein did impair viral growth somewhat. We have already alluded to a report that suggested that MV could apparently persist throughout life without triggering overt disease (Katayama et al. 1995). If this is true, sequencing of these viruses and comparing them to isolates from MIBE or SSPE tissues might offer insights into the functional consequences of biased hypermutations. Interestingly, a recent study using a green fluorescent protein (GFP)-expressing, rodent-adapted MV showed that, despite immunological resolution of MV without acute illness in adult immunocompetent mice, a high proportion of these otherwise healthy mice had detectable MV-positive cells within the CNS weeks after apparent clearance (Schubert et al. 2006), potentially paralleling the human autopsy studies described above.
Because limited mechanistic insights can be gained from human specimens, much of what we have learned about MV infection of the CNS, and resultant neuropathogenesis has come from animal models. As mice are not infectable by MV, mouse models of MV infection depend either upon the use of rodent-adapted MV strains or artificial expression of one of the human MV receptors, CD46 or CD150/SLAM. CD46 is now considered to be the principal receptor for vaccine strains of MV, while CD150/SLAM is the primary receptor for wild-type strains, though examples of overlapping receptor usage for both vaccine and wild-type viruses have been demonstrated.
Neuron-specific enolase (NSE)-CD46 mice were among the first transgenic mice to successfully express a human MV receptor and be susceptible to infection (Rall et al. 1997). In this model, transcriptional control of CD46 is governed by the NSE promoter, restricting expression and infection to CNS neurons (Rall et al. 1997). A limitation of this system is that it does not recapitulate SSPE in humans, since no peripheral infection can occur, and susceptible mice die during the acute phase of infection. However, it has been useful to study MV–neuron interactions and to measure immune cell activation, recruitment, and function within the CNS. Infection of NSE-CD46 mice by both an IC and IN route causes a neuron-restricted infection, though the outcome of infection is predicated on the maturity and effectiveness of the host immune response, as outlined below. In addition to their utility as a model of pathogenesis, the NSE-CD46 mice are a source of primary neurons that retain essential characteristics of this cell type (functional synapses, neuronal markers) and, further, are permissive for MV infection. It is important to note, however, that, while these cells faithfully reproduce cardinal aspects of neurons in vivo, the absence of supporting glial cells and the fact that these neurons are hippocampal in origin may not faithfully model MV infection of other neuronal subtypes in the complex cellular environment of the brain.
CD46+ hippocampal neuron cultures can support MV replication and spread, but the basics of the viral life cycle are fundamentally different from the standard textbook descriptions. While infection of fibroblasts is characterized by extensive release of extracellular virus, massive syncytia formation resulting from cell-to-cell fusion, and concomitant target cell death, infection of neurons shares none of these features (Lawrence et al. 2000). Although the virus can spread within neurons via trans-synaptic transmission (which is also true for polio, rabies, herpes simplex virus-1, pseudorabies virus, and other neurotropic viruses), no infectious, free virus is released into the supernatant. Of note, free virus is not detectable in SSPE specimens either, though whether these observations are due to the same replication restrictions is not known. Historically, the absence of infectious virus and viral budding in human SSPE brain specimens has been ascribed to the accumulation of point mutations in the envelope-associated genes (Dubois-Dalcq et al. 1976; Rentier et al. 1978; Waters and Bussell 1974). While these mutations may impact on viral release, they appear not to be essential for it to occur, since virus sequenced from CD46+ neurons is genetically identical to input virus (Gechman and Rall, unpublished observations). This suggests that the neuronal microenvironment alters the manner by which MV spreads, without necessitating selection of adapted viral variants. Similarly, no neuronal syncytia are observed, perhaps due to an inability of the viral ribonucleoprotein to traffic to the cell surface and acquire its envelope from the plasma membrane. Finally, and perhaps most strikingly, neurons infected with MV do not die of MV cytolysis: no differences in cell death staining were observed between infected and uninfected neurons, again resonating with human studies in which months to years of chronic infection can elapse with virtually no evident pathogenic impact.
These data implied that an alternative mode of viral transmission was operative in neurons and that the CNS disease observed in permissive mice could not be attributed to neuronal loss (Patterson et al. 2002). We have previously suggested (Rall 2003) that these key differences underscore the concept that the viral particle and the infectious particle are distinct entities: the viral particle consists of the viral nucleic acids and the proteins that it encodes, whereas the infectious particle consists of the viral proteins as well as the necessary cellular proteins that the virus needs to complete its life cycle. Given that each cell type offers a unique set of “protein tools,” one can extrapolate that the virus may encounter different sets of cellular proteins upon infection that may alter (or prevent) completion of the life cycle. For instance, Oglesbee and colleagues showed that either transgenic expression of heat shock protein 72 in neurons or transient hyperthermia increased the levels of MV RNA by as much as 2 orders of magnitude and correlated with increased neuropathogenesis (Carsillo et al. 2006).
Studies on the trans-synaptic spread of MV revealed potential roles for alternative MV receptors in neurons. While the CD46 receptor was required for initial viral entry, it was dispensable for neuron-to-neuron transmission (Lawrence et al. 2000), implying that the mechanism of cell-to-cell spread—at least for neurons—was not the same process as when a viral particle enters a non-neuronal target cell. Moreover, a neurotransmitter receptor, neurokinin-1 (NK-1), appears to be operative in trans-synaptic transmission and may support the case for a receptor for MV-F, since the natural ligand of NK-1 receptor, substance P, bears homology to the active site of the fusogenic domain of F (Makhortova et al. 2007). Of note, Cosby and colleagues showed that both wild-type and vaccine strains can utilize alternative uptake mechanisms to gain access to both neurons and oligodendrocytes that do not express either CD46 or SLAM (Abdullah et al. 2009). While infection in these murine cells occurred at a low level, this study nevertheless reinforces the belief among many in the measles community that more receptors await discovery.
What specific neuronal proteins are associated with changes in the MV life cycle? Movement of the MV RNP and envelope proteins to the neuronal synapse could not occur by random cytoplasmic diffusion—there must be active transport from the site of synthesis in the perinuclear space to the nerve terminus. Thus, engagement of cellular motor proteins, including dynein and kinesin that govern retrograde and anterograde transport, respectively, would likely be required. Fairly little is known about how viruses engage these proteins, especially in a cell such as a neuron where the distance that needs to be traveled is so extensive. Recombinant MV-expressing marker proteins (e.g., GFP), coupled with new methods such as slice cultures, will be crucial tools to help dissect the replication and spread of MV in situ (Ludlow et al. 2008). Moreover, the use of other morbilliviruses, such as canine distemper virus (CDV), will be informative: a GFP-expressing CDV was shown to gain access to the CNS by both anterograde transport via the olfactory bulb as well as hematogenous spread through the choroid plexus and capillaries within the brain (Rudd et al. 2006).
What is the status of the virus during the long latency between initial exposure and disease? Autopsy studies would suggest that the presence of MV in the CNS is more common than the frequency of MV-triggered CNS diseases would imply. Assuming this is true, it would argue more in favor of the reactivation model over the slow growth model. But what triggers viral resurgence? And how is the virus maintained in a latent state? The issue of whether this is a ficommon event with a rare outcomefl, or a firare event with a common outcomefl remains to be defined.
Is MV associated with chronic CNS diseases of unknown etiology? Related to this prior point, and as we noted earlier, MV has been associated with a variety of chronic diseases, though formal proof of causation of any disease with a neurotropic pathogen has yet to be established. However, the notion that a virus could trigger a CNS disease that looks quite different from the acute disease is intriguing and could possibly manifest as a result of direct viral damage (to myelin-producing oligodendrocytes, for example) or, indirectly, by induction of an autoreactive or overly aggressive immune response in the CNS. A lucid picture of how immune responses typically function in the brain, which is discussed in the next section, will be of key importance as these studies move forward.
Basic principle 2: neurons play a critical role in induction of host immunity and noncytolytic clearance
When viral infections occur in CNS neurons, the immune response faces a unique challenge. Since neurons are largely a nonrenewable cell population, clearance by cytolytic mechanisms, such as through cytotoxic CD8+ T cell interactions with infected cells, could cause irrevocable neuronal loss. In contrast, without an effective mechanism for viral clearance, persistent infections of the brain could be easily established. Viral infections in the brain typically elicit a potent immune response that can include infiltration of CD4+ and CD8+ T cells into the brain parenchyma, elevated levels of virus-specific antibodies in the CSF, and microglial activation. Yet, in many cases of CNS infection, viral clearance can occur in a noncytolytic manner with minimal damage to host CNS tissue (Binder and Griffin 2001; Finke et al. 1995; Patterson et al. 2002; Rodriguez et al. 2003b). Our laboratory and others have identified a critical role for IFNγ, a pluripotent cytokine released by activated T cells and natural killer cells, in this process (Finke et al. 1995; Patterson et al. 2002).
In the model of MV neuronal infection described above, adult NSE-CD46 mice require IFNγ in order to clear MV from neurons in the brain, as NSE-CD46 mice lacking IFNγ develop neurological disease, have widespread MV replication in the brain, and succumb to infection (Patterson et al. 2002). In recombinase-activating gene 1-deficient mice lacking B and T cells that express human CD46 under the control of its endogenous promoter, adoptive transfer of CD4+ T cells, could protect these immunodeficient mice from MV-mediated neuropathology only if the CD4+ T cells expressed IFNg (Tishon et al. 2006). Moreover, adult Balb/c mice control the spread of a rodent-adapted strain of MV (CAM/RB) in the brain in an IFNγ-dependent manner. When IFNγ was depleted by antibody neutralization, Balb/c mice become susceptible to CAM/RB in the brain, and helper T cells from these mice switched from a Th1 to a Th2 profile (Finke et al. 1995). Together, these studies point to a critical role for IFNγ in both directly controlling MV spread (and preventing subsequent MV-associated neurological disease) as well as in effective induction of the host response.
To understand how IFNγ induces viral clearance in neurons, it is necessary to identify the downstream signaling pathways that are triggered upon IFNγ engagement. These pathways have been extensively characterized in non-neuronal cells (reviewed in Stark et al. 1998) and thus will only be summarized here. Upon IFNγ binding and assembly of the receptor complex (consisting of a heterotetramer of IFNγR1 and IFNγR2 subunits), receptor-associated Janus kinases-1/2 are activated, resulting in the tyrosine phosphorylation of the cytoplasmic tail of the IFNγR1 subunits. Upon docking to the phosphorylated R1 subunit, signal transducer and activator of transcription 1 (STAT1) is phosphorylated on tyrosine 701, resulting in its homodimerization. The phosphorylated STAT1 homodimer translocates to the nucleus and binds to gamma-activated sequence elements within IFNγ-responsive genes (ISGs) to initiate transcription. These gene products establish the “antiviral state” in which the cell inhibits viral spread by upregulation of proteins that block viral gene transcription and protein translation, as well as those that cleave viral RNA. Notably, ISGs are induced in neurons and other CNS cells during West Nile and LCMV infections, but ISG induction (ISG-49, ISG-54, and ISG-56) only partially depends on STAT1 expression, suggesting that other signaling pathways in CNS cells may contribute to ISG expression and the antiviral state (Wacher et al. 2007).
Cell-specific responses to IFNγ have been noted in the regulation of STAT1 and in the activation of alternative, STAT1-independent signaling pathways (reviewed in van Boxel-Dezaire and Stark 2007), suggesting that target cells may play a role in how they respond to IFNγ or to other inflammatory cytokines. For example, many mitotically active cells, such as fibroblasts, respond to IFNγ with rapid activation of STAT1, followed by inactivation of STAT1 by suppressor of cytokine signaling 1. In contrast, primary hippocampal neurons respond to IFNγ with delayed and attenuated STAT1 phosphorylation and expression, though the attenuated activation is sustained for days after the initial IFNγ exposure (Rose et al. 2007). Since IFNγ signaling via STAT1 is often associated with antiproliferative and proapoptotic effects, the attenuation in STAT1 signaling in neurons may confer a survival advantage. In addition, many alternative signaling pathways have been implicated in IFNγ signaling, including activation of other STAT proteins and of other kinases such as protein kinase C family members, PI 3-kinase, and p38-MAP kinase. The role of alternative signaling pathways in IFNγ-mediated noncytolytic clearance is unknown, but in light of attenuated STAT1 activation by IFNγ seen in primary hippocampal neurons, it is possible that neurons utilize other signaling pathways in addition to STAT1 to mediate control of MV in neurons.
Even among neural cell types in the brain, IFNγ differentially mediates viral clearance, depending upon the infected cell type. For example, MHV infects astrocytes, microglia, and oligodendrocytes in the brain (Wang et al. 1992); while perforin is sufficient to mediate viral clearance from astrocytes and microglia, oligodendrocytes depend upon IFNγ in order to clear MHV infection (Bergmann et al. 2006; Parra et al. 1999). In addition, retinal explants from embryonic and neonatal mice strongly activate STAT3, but not STAT1, upon IFNγ exposure (Zhang et al. 2005), confirming that neural cells can utilize alternative signaling pathways in response to IFNγ.
IFNγ levels are rarely elevated in the CSF of adults or children with SSPE, suggesting that traditional mechanisms of viral clearance are either not activated or are suppressed in SSPE (Griffin et al. 1990; Ichiyama et al. 2006; Mistchenko et al. 2005). However, levels of IL-10, a Th2-related cytokine, are elevated in the CSF of SSPE patients, which could reflect an inappropriate skewing from a Th1 response (IFNγ) to a Th2 response. T cells isolated from the peripheral blood of SSPE patients are largely unable to produce IFNγ in response to MV infection in vitro, though IFNγ production in response to other viruses is not inhibited (Hara et al. 2000). Indeed, a subset of SSPE patients from this study possessed T cells that were capable of producing IFNγ in response to MV similarly to non-SSPE controls. Those SSPE patients whose T cells were capable of IFNγ production in response to MV maintained receptive function and progressed more slowly in comparison to SSPE patients with T cells that did not produce IFNγ in response to MV (Hara et al. 2000). These findings suggest that, even in SSPE patients, where viral clearance fails in the brain, IFNγ may be beneficial in controlling MV spread and delaying disease progression.
How does IFNγ mediate MV clearance/control from different cell types in the brain? While it is clear that IFNγ is capable of controlling MV spread in the CNS and specifically in neurons, the signaling mechanisms that ultimately lead to the control of MV are unknown. Understanding the downstream ISGs that are critical for MV control and whether there are cell type-specific differences between mitotically active cells and nondividing cells such as neurons, would offer great insight into potential targets for treating SSPE.
Why does MV control fail during SSPE? SSPE is characterized by widespread MV replication in the brain, even though SSPE patients are not typically immunodeficient. How then does MV overcome traditional host defense mechanisms that are capable of limiting MV spread in other individuals?
Basic principle 3: developmental changes in the brain and immune response affect age-dependent pathogenesis
CNS complications with MV infection are overwhelmingly associated with children, though this may reflect the young age at which most people are typically infected. Given the relative paucity of adults with primary MV infections, it is difficult to discern whether the development of MV-associated neurological complications is age-dependent. However, in animal models of MV infection, young age correlates with both severity of infection in the brain and pathological outcome. For example, adult NSE-CD46 mice become infected by MV in the brain following intracerebral or intranasal inoculation, but are able to clear the infection within weeks without neurological damage, whereas neonatal NSE-CD46 mice demonstrate uncontrolled viral spread, develop seizures, and succumb to infection (Lawrence et al. 1999). Of note, both adult and neonatal NSE-CD46 mice recruit approximately equivalent numbers of CD4+ and CD8+ T cells into the brain, and thus the age dependence does not appear to be due to an inability of neonates to initiate an immune response. The basis of this differential pathogenesis may be due to neuronal vulnerability: neonatal NSE-CD46 mice show widespread apoptosis in the brain with astrogliosis and microgliosis throughout the parenchyma, while adult mice show little evidence of neuronal loss or inflammation (Lawrence et al. 1999; Manchester et al. 1999).
Other transgenic models expressing CD46 or SLAM have shown similar age-dependent susceptibility to MV infection in the brain. In transgenic YAC-CD46 neonatal mice, neurons are the initial target of MV, though low levels of oligodendrocyte and microglia infection are seen later in infection. As in the NSE-CD46 model, MV infection in neonates is ultimately associated with widespread MV replication in neurons, T cell infiltration into the brain, apoptosis, and eventually death (Manchester et al. 1999; Oldstone et al. 1999). In another transgenic model, mice were established that ubiquitously expressed BC-Cyt1 or C-Cyt2 isoforms of CD46, the isoforms most abundant in the human brain. Ubiquitous expression of either isoform resulted in 100% mortality when neonates were infected with MV, again correlating with widespread infection and apoptosis (Evlashev et al. 2000). Interestingly, despite robust CD46 expression in lymphocytes from these mice (more than in human lymphocytes), MV replication was less in the murine lymphocytes than in the human lymphocytes, suggesting that host factors other than receptor expression modulate the viral life cycle (Evlashev et al. 2001). Like CD46-expressing transgenic mice, mice transgenic for SLAM also demonstrate age-dependent susceptibility; suckling mice infected intranasally with wild-type MV develop neurological disease and succumb to infection, while transgenic adults mount an immune response to CNS infection and survive (Sellin et al. 2006).
While transgenic mouse models have been useful for analyzing the immune components and neuropathogenesis of neonatal MV infections, none recapitulate the long lag between acute MV infection and CNS symptoms seen in SSPE. A model of chronic MV infection in the brain was recently established where young mice were infected with a recombinant MV virus expressing enhanced green flourescent protein and the H protein from a rodent-adapted MV strain (MV-GFP-CAMH) (Schubert et al. 2006) Replication of GFP-expressing MV was then visualized in infected neurons. Neonates succumbed to infection, but mice inoculated at 2 weeks of age survived and maintained a persistent infection in the brain out to 50 days post-infection, with viable GFP + neurons in the hippocampus and cortex. Though it is unclear how MV-GFP-CAMH persists in immunocompetent mice when infected at a young age, this model provides an opportunity to dissect immune components that may contribute to a chronic MV infection in neurons.
Age is a major variable in the neuropathogenesis of many CNS infections. For many neurotropic viruses, cellular tropism becomes increasingly restricted as the brain matures. Viruses such as LCMV, cytomegalovirus, and Semliki Forest virus readily infect areas of the brain that are rich in neural precursor cells and immature neurons, but become increasingly restricted to certain areas of the brain or to specific neuronal subtypes in the adult (Allsopp and Fazakerley 2000; Bonthius et al. 2007; van den Pol et al. 2002). In in vitro studies, MV readily infects undifferentiated NT2 cells, which are human teratocarcinoma cells that can be terminally differentiated into neurons with retinoic acid treatment (McQuaid et al. 1998). In contrast, differentiated NT2 neurons are initially refractory to MV infection, but can become infected by cell-to-cell contact with infected neuroepithelial cells. In rodent models of MV infection, MV spreads more aggressively in neonatal neurons than in adults, though whether this is due to greater tropism for immature neurons and/or to limited immune control is unknown. It is also unclear whether MV demonstrates greater tropism for immature neural cells in humans. Interestingly, in SSPE samples, MV antigens can be found in many neural cell types, including neurons, oligodendrocytes, astrocytes, and brain endothelial cells, suggesting that MV is capable of infecting many types of cells in the brain. If MV spreads from a limited presence in endothelial cells during acute infection to a more widely disseminated infection in SSPE through trans-synaptic spread or even if MV entry into the CNS occurs during the acute phase of infection in SSPE cases, remains to be determined.
In addition to changes in neural cells that may affect MV biology in the brain, developmental changes in the immune response may also contribute to age-dependent pathogenesis. The efficacy of the neonatal innate and adaptive immune responses has classically been viewed as immature or defective in comparison to adults when responding to viral infections. This is based upon observations that neonatal immune cells differ quantitatively and qualitatively in comparison to adults. For example, neonatal T cells, which are fewer in number in comparison to adults, were historically viewed as deficient due to their limited capacity to proliferate and produce IL-2 (Adkins et al. 2004). However, this hypothesis has been refined to include that the neonatal immune response may be active, but skewed toward the production of “inappropriate” cytokines during viral infections (Bot et al. 1998; Sarzotti et al. 1996). For example, human neonatal CD4+ T cells skew toward the production of Th2 cytokines in response to environmental antigens, whereas adult CD4+ T cells from non-allergic individuals skew toward Th1 cytokines (Smith et al. 2008). In mice, neonatal CD4+ T cells show a Th1 deficiency, including impaired IFNγ production, even when adoptively transferred into an adult host (Adkins et al. 2002). Collectively, these results suggest that neonatal adaptive immune response may struggle to control viruses whose clearance is dependent upon Th1 cytokines such as IFNγ. Given the critical role that IFNγ plays in clearance from the CNS, as described in the previous section, neonatal defects in IFNγ production could impact on control of spread and clearance of MV.
Why do infiltrating T cells fail to control MV infection in the neonatal CNS? CD4+ and CD8+ T cells enter the brain parenchyma during MV infection in adults and neonates, but neonates cannot control MV spread. Recent work into neonatal T cells suggests that the Th2 response dominates over the Th1 response during viral infections, which could cause a deficiency in IFNγ production and hence an inability to control MV, but how this unfolds in the brain is unknown.
What developmental changes in the brain and the immune response contribute to the onset of SSPE? It is likely that MV enters the CNS during or soon after acute infection, which typically occurs under 2 years of age. Does the immune response interact with the developing CNS in a fundamentally different way than with the mature CNS?
The point-of-view of this commentary is that critical basic observations can be gleaned by studying rare diseases. The three main principles that were addressed here—the unique viral life cycle in neurons, the dissection of the noncytopathic host response, and the potential basis of age-dependent pathogenesis—underscored progress that has been made to date and highlighted key questions for future study.
Our title referred to “blue moon” virology, meant to imply the study of diseases that arise infrequently. Indeed, a blue moon is, simply, an extra full moon that appears during the course of a year: 13 instead of the usual 12. Other than its relative infrequency, this extra full moon is like the others, illuminating in the same way as other full moons. We believe that the study of rare, virus-mediated neurological diseases can afford insights into general biological and pathogenic processes, and in that way, can be equally illuminating.
We thank Kevin O'Regan, Christine Matullo, and Sarah Cavanaugh for advice on this manuscript and acknowledge support from the following sources: G.F.R. was supported by NIH grants RO1-NS40500, RO1-NS060701, P30-CA006927, a Pilot Project grant from Autism Speaks, Pennsylvania Department of Health Tobacco funds, and a gift from the F. M. Kirby Foundation. L.O’D. was supported by an NRSA Postdoctoral fellowship from NINDS.