Acta Neuropathologica

, Volume 109, Issue 1, pp 32–48

Pathogenesis of prion diseases

Authors

  • Ursula Unterberger
    • Institute of NeurologyMedical University of Vienna, AKH
    • Institute of NeurologyMedical University of Vienna, AKH
  • Herbert Budka
    • Institute of NeurologyMedical University of Vienna, AKH
Review

DOI: 10.1007/s00401-004-0953-9

Cite this article as:
Unterberger, U., Voigtländer, T. & Budka, H. Acta Neuropathol (2005) 109: 32. doi:10.1007/s00401-004-0953-9

Abstract

Prion diseases are rare neurological disorders that may be of genetic or infectious origin, but most frequently occur sporadically in humans. Their outcome is invariably fatal. As the responsible pathogen, prions have been implicated. Prions are considered to be infectious particles that represent mainly, if not solely, an abnormal, protease-resistant isoform of a cellular protein, the prion protein or PrPC. As in other neurodegenerative diseases, aggregates of misfolded protein conformers are deposited in the CNS of affected individuals. Pathogenesis of prion diseases comprises mainly two equally important, albeit essentially distinct, topics: first, the mode, spread, and amplification of infectivity in acquired disease, designated as peripheral pathogenesis. In this field, significant advances have implicated an essential role of lymphoid tissues for peripheral prion replication, before a likely neural spread to the CNS. The second is the central pathogenesis, dealing, in addition to spread and replication of prions within the CNS, with the mechanisms of nerve cell damage and death. Although important roles for microglial neurotoxicity, oxidative stress, and complement activation have been identified, we are far from complete understanding, and therapeutic applications in prion diseases still need to be developed.

Keywords

Prion pathogenesisOxidative stressMicrogliaImmune systemPeripheral nervous system

Introduction

Prion diseases or transmissible spongiform encephalopathies (TSEs) may present as sporadic, genetic, or acquired infectious disorders. They occur in humans as sporadic (sCJD), familial (fCJD), or variant (vCJD) Creutzfeldt-Jakob disease, Gerstmann-Sträussler-Scheinker disease (GSS), fatal familial insomnia (FFI), and kuru, as well as in animals, like scrapie in sheep and goats, and bovine spongiform encephalopathy (BSE) in cattle. According to present knowledge, their outcome is invariably fatal. Unique to these diseases is the infectious pathogen, termed prion. Prions have been defined as transmissible particles lacking nucleic acid, presumably composed exclusively of a modified normal cellular protein, the prion protein or PrPC [98, 121]. The pathological prion protein (PrPSc) is a misfolded conformer that is characterized by a high β-sheet content and protease resistance. Thus, it can not be degraded by cellular enzymes and is deposited extracellularly in brain tissue during the course of the disease [121].

Human prion diseases are rare. Less than two per million individuals are affected per year. Only about 10 years ago, the public became interested in these peculiar disorders because of the “outbreak” of a new variant of CJD in Great Britain, which was due to the oral uptake of infectivity, presumably from BSE cattle. No one knows how many people could possibly be endangered, and great efforts have been made to elucidate modes and spread of prion infection. Significant advances have been achieved, especially in the field of peripheral pathogenesis. This part of prion research deals with the routes prions may use to reach the CNS from the body periphery, and the cell types, particularly immune cells, that are involved in peripheral prion propagation.

In contrast, how prions harm the CNS still remains enigmatic. The clinical picture in prion diseases is dominated by central neurological symptoms, which are presently believed to be due to an early synaptic dysfunction [45] and, at later stages, neuronal loss. What we do not know is how neurons are actually affected by the disease? Why are some neurons more readily destroyed than others? What is the role of PrPC in the whole process? And how does the pathological form, PrPSc, contribute? To the neuroscientist and neuropathologist, these questions are certainly most fascinating. They are discussed in detail in the next sections; an overview of peripheral prion pathogenesis is presented in the second half of this review.

CNS pathogenesis of prion diseases

Historically, there were two major approaches to central prion pathogenesis: one was the so-called “gain of function hypothesis”, which put forward a possible neurotoxic effect of an abnormally folded, not further degradable, protein that is deposited in considerable amounts in the brains of affected individuals. On the other hand, one could argue that the continuous conversion of PrPC to PrPSc might lead to decreased availability and/or functional impairment of the former, so that its assumed neuroprotective effects are lost. This was central to the “loss of function hypothesis”. Both theories had their advocates. Although none of them has definitely proven right or wrong, the situation is certainly much too complex to be satisfyingly explained by one simple model.

What seems now clear is the incapacity of PrPSc accumulation alone for causing symptomatic disease [11, 103]. PrP-deficient mice are generally resistant to scrapie [31, 105]. When PrPSc deposition is induced in such animals by grafting neural tissue overexpressing PrPC into their brains and intracerebrally inoculating them with scrapie prions, the grafts accumulate PrPSc, which also spills over to the host brain. However, while the grafts develop severe, scrapie-like neurodegeneration, the brain tissue devoid of PrPC shows no damage at all [11]. Furthermore, if neuronal PrPC is depleted in mice with ongoing neuroinvasive prion infection, non-neural replication and accumulation of prion infectivity continues, but early cerebral histopathological changes are reversed, and neuronal loss and progression to clinical disease are prevented [103]. Thus, expression of the normal prion protein must play a crucial role in the development of neurodegeneration after prion infection. This knowledge implies another question: must PrPC necessarily be present on all types of brain cells (i.e., neurons, astrocytes, and oligodendrocytes) to confer (1) susceptibility to clinical prion disease, (2) formation of PrPSc, and (3) transmission of infectivity and disease?

To address this issue, several transgenic mouse models have been generated, expressing PrPC selectively in neurons [124], astrocytes [126], and oligodendrocytes [120], respectively. Subsequent inoculation and transmission experiments revealed that mere neuron-specific expression of hamster PrPC suffices to support prion infection and disease development [124], while restriction of murine PrPC to oligodendrocytes does not [120]. The role and impact of astrocyte-specific PrPC expression is discussed controversially. Raeber and colleagues [126] reported that transgenic mice expressing hamster PrPC selectively in astrocytes are susceptible to prion infection. On the ultrastructural level, brains of these mice show neuronal lesions typical of TSE, despite lack of neuronal PrPC [80], suggesting that deposition of PrPSc in intimate proximity to neurons and their processes is sufficient to induce TSE pathology. However, the studies mentioned above, using a different approach (neuroectodermal grafting [11] or postnatal neuron-specific down-regulation of PrPC expression [103]), argue against this hypothesis, since close proximity of PrPSc to the neuronal cell surface in these models did not induce obvious morphological alterations, neuronal loss, or clinical disease [11, 103]. Whether these differences reflect distinct pathogenic mechanisms determined by neuronal and astrocytic PrPC expression [80] or certain prion strain properties (hamster 263K versus mouse RML), or if they rely on varying expression levels of astrocytic prion protein in the mouse models used (or on other still unknown factors), is presently not clear.

Finally, it is important to note that mice with selective genetic elimination of the prion protein gene (Prnp) open reading frame (ORF) within the borders of exon 3 (leaving the splice acceptor site of exon 3 intact) display per se only relatively subtle behavioral and biochemical changes [25, 30, 41, 102, 104, 139], so that the severe damage of neuronal tissue in scrapie mice can hardly be related just to loss of PrPC function. In contrast, mice with a deletion of the Prnp ORF including at least the splice acceptor site of exon 3 develop loss of cerebellar Purkinje cells and late-onset ataxia [128, 129]. This phenotype has meanwhile been linked to the aberrant expression of a PrPC-like protein, called Doppel, which is encoded by a Prnp-like gene, named Prnd, located immediately downstream of the Prnp locus [111]. Under physiological conditions, neuronal expression of Doppel is silenced post-developmentally. In PrP knockout mice with deletion of the splice acceptor site of exon 3, its expression is re-activated under the control of the Prnp promoter due to an intergenic splicing between Prnp and Prnd [111], leading to Doppel-mediated neuronal cell death.

In sum, cell culture and in vivo results, taken together, suggest a combination of various mechanisms leading to neuronal death in prion diseases. A few of these mechanisms are discussed in the following sections.

Morphological aspects: local distribution of PrPSc accumulation and neural tissue damage

If one considers the morphological features of the prion-affected brain (Figs. 1A–F, 2A, B) with regard to pathogenic mechanisms, primarily two questions arise: (1) what determines patterns and distribution of PrPSc deposition and neural tissue damage, and (2) is there a direct relation between PrPSc deposition and histopathological signs, like astrogliosis (which is a rather unspecific reaction), spongiform change (which is highly specific for prion diseases), and, finally, neuronal loss?
Fig. 1 A–F

Histological (hematoxylin-eosin staining, A–C) and immunohistochemical (PrPSc immunolabeling, D–F) findings in a case with sporadic CJD. Cerebral cortex with typical spongiform change accompanied by slight neuronal loss and gliosis (A) and a characteristic diffuse synaptic type staining pattern for PrPSc (D). Focal area with large confluent vacuoles in the cortex of the same patient (B) with a typical patchy-perivascular deposition pattern of PrPSc (E). Cerebellar cortex with spongiform change in the molecular layer (C), synaptic PrPSc deposition pattern in the molecular and a coarser, granule-like deposition pattern in the granule layer (F) (PrP prion protein, PrPSc pathological PrP, CJD Creutzfeldt-Jakob disease). Original magnification A, D, F ×100; B, C, E ×200 (photomicrographs kindly provided by Dr. Ellen Gelpi, Institute of Neurology, Medical University of Vienna)

Fig. 2

A, B Bihemispheric sections immunostained for PrPSc. Cerebral (A) and cerebellar (B) cortex of a patient suffering from sporadic CJD with prominent diffuse cortical PrP deposition, but without significant brain atrophy and hippocampal involvement. C, D Sporadic CJD with pre-existing cerebellar infarctions. Adjacent sections immunostained for PrPSc (C) and the neuronal protein synaptophysin (D). In the center of the picture, few cerebellar gyri show preserved histoarchitecture with prominent synaptophysin expression, while the surrounding gyri are scarred, lacking any detectable neuronal staining (D). PrPSc deposition strictly co-localizes with the expression of the neuronal marker protein without significant diffusion into the surrounding tissue (C). Original magnification C, D ×20

Presence and patterns of histopathological changes vary greatly between individual cases and disease subtypes [28, 29]. In sCJD, the morphological and clinical phenotype was shown to depend on the physicochemical properties of PrPSc and the genetic background of the patient [75, 114]. In FFI, there may be little or no spongiform change at all [2], while this disease is specifically characterized by prominent thalamic atrophy with profound astrogliosis [28, 29]. In any case, local PrP deposition seems to require the presence of intact neuronal elements; PrPSc does not accumulate in pre-existing brain lesions, like scarred infarctions with prominent gliosis (Fig. 2C, D) [28, 29].

One factor rendering neurons more sensitive to prion-mediated toxicity might be the level of PrPC expression. PrP0/+ mice, which express about half the normal level of PrPC, display a delayed onset of clinical disease [32, 105]. On the other hand, even cell types strongly expressing PrP, like cerebellar Purkinje cells, are relatively resistant to cell death after prion infection [57, 95]. Selective vulnerability of parvalbumin-expressing GABAergic neurons was found both in human [9, 61] and experimental prion diseases [62]. This vulnerability was detectable already early in the incubation period, and thus represents one of the earliest changes ever described after experimental inoculation [62]. Interestingly, FFI differs in this phenomenon from all other human TSEs [63]. Other vulnerabilities include that of the granular layer of the cerebellum that is frequently depleted in sCJD [29], and the variable involvement of the basal nucleus of Meynert [3, 34]. However, it was shown that in transmitted prion diseases, the respective strain of agent plays a central role with regard to incubation time and neuropathology [27]. Varying strain-dependent lesion profiles in syngenic animals are suggestive of a strain-specific targeting of different neuronal populations [27, 57].

In conclusion, PrPSc deposition and tissue damage is in a way influenced by host factors, like genotype and individual and selective vulnerability of neuronal subsets, as well as by PrPSc properties or, in transmitted prion diseases, the strain of agent. So far, the molecular mechanisms underlying these effects remain obscure.

Concerning the second question, originally, it was proposed that disease-associated histopathological changes in the brain correlate well with PrPSc deposition [81]. Reactive astrogliosis was found to follow PrPSc accumulation by 1–2 weeks in hamster scrapie. Accordingly, it was suggested that etiology and pathogenesis of prion diseases are directly related to PrPSc [81]. Later investigations showed somewhat different results. Not in all cases do amount and distribution of PrPSc actually correspond with type and severity of local tissue damage [2, 90]. In some TSEs, like FFI [46] and GSS [70], lesions may develop without PrPSc accumulation. In a time course study in mice with experimental CJD, spongiform change preceded PrP deposition in various brain regions [90]. A study using brains of the same series found severe loss of a subpopulation of GABAergic neurons in the cerebral cortex 5–9 weeks after the infection, i.e., clearly before the appearance of cortical PrPSc [62] (see above). Furthermore, experiments revealing, under certain conditions, loads of PrPSc in brain tissue but no accompanying tissue damage [11] must be taken into consideration in this context. A consistent relationship between PrPSc deposition and brain tissue damage has never been proven and, with a growing amount of data arguing against it, is becoming less and less likely.

Prion research in vitro: the fragment PrP106–126 and its neurotoxic effects

Until the early nineties of the last century, central pathogenesis of prion diseases was discussed primarily on the basis of morphological observations analyzing the extent and spatial pattern of spongiform change, neuronal loss, astrogliosis, and deposition of PrPSc. In 1993, Forloni et al. [51] described a peptide corresponding to amino acid residues 106–126 of the human PrP, which has a high intrinsic tendency to aggregate into fibrils, thereby mimicking one key feature of PrPSc. This peptide was the only one amongst a variety of peptides tested that was capable of inducing cell death in primary neuronal cultures after chronic exposure. Although the discovery of the “neurotoxic” property of PrP106–126 has provided prion researchers with a highly valuable tool to unravel pathogenic mechanisms, it must be emphasized that this fragment has in fact never been detected in any form of TSE (natural, acquired or experimental). Thus, it should be regarded as an experimental system for prion research, but not as an actual part of the disease.

During the following years, neurotoxic mechanisms were studied extensively using either PrP106–126 or preparations of purified PrPSc. As mentioned in the introduction, the first crucial factor for neuronal death is the expression of PrPC. Neurons that do not express PrPC are resistant to the in vitro toxicity of PrP106–126 [18], as well as the in vivo neurodegeneration mediated by PrPSc [31, 105]. On the other hand, cerebellar cell cultures of mice overexpressing PrPC are more sensitive to the toxicity of PrP106–126 [13]. Another necessity for a toxic effect of the peptide is the co-existence of neurons and microglia. Neurons are resistant to PrP106–126-mediated cell death after depletion of contaminating microglia [19].

The contribution of astrocytes to the neurotoxicity of PrP106–126 is discussed controversially. Based on results of a co-culture model of PrPC-deficient neurons and PrPC-expressing astrocytes, Brown [14] proposed that PrPC-ablated neurons develop an increased sensitivity to glutamate toxicity in the presence of PrPC-positive astrocytes, thus becoming even more dependent on a sufficient astrocyte-mediated protection against glutamate. Incubation of these co-cultures with PrP106–126 inhibited this protective property of astrocytes, resulting in an increased glutamate-mediated damage to the sensitized PrPC-depleted neurons, a pathway, by which astrocytes could indirectly contribute to the in vitro toxicity of PrP106–126 and the neurodegeneration seen in prion diseases. At first glance, this suggestion might be in line with inoculation experiments of a transgenic mouse model expressing PrPC selectively in astrocytes [80, 126]. However, its relevance in vivo has recently been questioned by a conditional knockout model with specific postnatal depletion of neuronal PrPC [103]. In this model, inoculation with RML prions resulted in a progressive accumulation of PrPSc, which was first converted from neuronal PrPC, but later, after the neuronal Prnp gene had been eliminated, was solely derived from astrocytes. Neither obvious neuronal damage nor any signs of clinical disease could be observed in the infected animals. Thus, PrP-negative neurons survived despite an intense formation of PrPSc and the postulated inhibition of glutamate detoxification by astrocytes, indicating that, in vivo, either neurons are not primed for increased glutamate sensitivity in a similar way to that observed in vitro, or they can escape glutamate toxicity by at least one different clearance pathway. Therefore, the role of astrocytes in central pathogenesis in vivo needs to be further determined.

Oxidative stress and antioxidant stress defense

An indirect strategy of elucidating CNS pathogenesis of TSEs tried to explore the physiological roles of PrPC. In this context, much attention was devoted to the unstructured N-terminal region of PrPC containing a unique octapeptide-repeat sequence. This octarepeat region belongs to the best-conserved parts of the mammalian prion protein, although a recent report describes a higher degree of variability than previously anticipated [142]. First insights into the function of this peculiar repeat sequence came from studies in cell-free systems. Using peptides corresponding to three to four octarepeats of mammalian PrP, Hornshaw et al. [76] demonstrated specific and preferential binding of divalent copper ions as compared to other metals. Like the discovery of the neurotoxic effect of PrP106–126, this observation opened new perspectives in prion research and initiated a series of experiments to unravel this new functional property of PrPC.

Copper binding has been reported to influence structural and biochemical properties of PrPC or parts of it. Incubation of aged recombinant PrP or PrPC in the microsomal fraction of brain extracts with divalent copper ions initiated a conformational shift from α-helical to β-sheet structure, accompanied by the formation of aggregates, detergent insolubility and resistance against proteinase K digestion, all biochemical characteristics of the pathological prion protein, PrPSc [122]. However, further studies revealed that the protease-resistant PrP generated by copper treatment exhibits a different structure compared to PrPSc [123], arguing against any gain of direct infectivity of copper-converted PrP. Nevertheless, PrP conversion by a physiological or even excessive load of copper ions might play a supportive role in the development of prion diseases.

Another significant finding was the rapid, reversible stimulation of endocytosis of PrPC from the cell surface by copper ions [115]. This may have important physiological implications, as PrPC could serve as a recycling receptor for copper uptake from the extracellular milieu [115], thus protecting the cell against the toxicity of free copper cations.

In 1997, first studies were published that indicated that the role of PrPC in the cellular defense against oxidative stress might go further than the mere binding of copper. Early results showed that elevated levels of PrPC correlate with enhanced cellular resistance to oxidative stress [21], while lack of PrPC results in higher sensitivity [22]. Primarily, the improved cellular resistance to oxidative stress was ascribed to an auxiliary function of PrPC supporting known cellular antioxidant defense mechanisms. Increased activities of copper/zinc superoxide dismutase (Cu,Zn SOD) and glutathione peroxidase (GPx) were found in neuronal cells expressing higher levels of PrPC [21, 125]. PrP-deficient cultured neurons, on the other hand, have significantly reduced glutathione reductase (GR) activity, combined with enhanced sensitivity to the toxic effects of hydrogen peroxide [146]. The exact mechanism of a PrPC-dependent elevation of cellular antioxidant activity in general and Cu,Zn SOD activity in particular is still discussed controversially. Brown and colleagues described a close correlation between the expression level of PrPC, brain copper content [20], cellular copper uptake, and copper incorporation into Cu,Zn SOD [16]. Accordingly, they suggested that the activity of Cu,Zn SOD is regulated post-translationally by its copper supply [16]. Enhanced cellular copper binding in response to elevated PrPC expression levels, as well as increased antioxidant enzyme activity have also been observed by other researchers, whereas a change of copper delivery into the cell could not be demonstrated [125]. Therefore, it was proposed that PrPC activates a so-far-unknown signal transduction pathway to stimulate cellular oxidative stress defense [125]. In 1999, Brown and colleagues [23] widened the spectrum of potential antioxidant activities of PrP by demonstrating an intrinsic SOD-like activity of native and recombinant mouse and chicken PrPC folded in the presence of copper ions. In an attempt to further elucidate the physiological function of PrPC in response to oxidative stress, our group analyzed the neuronal PrPC expression profile in human neurodegenerative disorders, in which damage by free radicals is thought to play a pivotal pathogenic role. We hypothesized that the proposed protective properties of PrP against oxidative stress should be reflected in its cellular up-regulation. In TSEs and Alzheimer’s [144], Parkinson’s, and diffuse Lewy body disease, as well as in progressive supranuclear palsy and multiple system atrophy [92], we could indeed demonstrate a significant increase of intraneuronal PrP immunoreactivity. In contrast, in motor neuron disease, PrP expression was lost in anterior horn neurons of the spinal chord, i.e., in those neurons, which selectively degenerate during the course of the disease [92]. This indirect approach provided additional evidence for a relevant function of PrPC in the cellular defense against oxidative stress.

The potential link between oxidative stress and prion diseases is marked by two categories of pathophysiological events: (1) enhanced production of reactive oxygen species (ROS) and reactive nitrogen species (RNS), followed by direct oxidative cell damage, and (2) impairment of the cellular oxidative stress defense.

Increased generation of ROS has primarily been reported in in vitro experiments using PrP106–126. Analysis of primary neuronal cultures containing residual amounts of microglia demonstrated that neurotoxicity of PrP106–126 depended on a combination of neuronal PrPC expression and enhanced production of oxygen radicals by activated microglia [19]. Recently, it was shown that the mere co-incubation of PrP106–126 with copper ions in a cell-free system results in the generation of hydrogen peroxide, which is further converted to highly reactive hydroxyl radicals following the Fenton principle [140].

Direct oxidative damage to lipids, proteins, and DNA has been reported in several models of prion disease. In mice with end-stage scrapie, our group demonstrated immunohistochemically a widespread neuronal labeling for nitrotyrosine (NT), a common marker for protein oxidation. In addition, a marked increase in heme oxygenase-1 (HO-1) immunoreactivity was observed [64]. HO-1 expression is generally used as a sensitive marker for oxidative stress. In brains of patients with sCJD and fCJD, we observed oxidative damage to nucleic acids. Immunohistochemical analysis revealed nuclear and cytoplasmic staining for 8-hydroxyguanosine/8-hydroxydeoxyguanosine (8-OHG/8-OHdG) in neurons, indicating an oxidative damage to DNA as well as RNA. Interestingly, staining intensity was correlated with disease duration, but not with the deposition pattern of PrPSc [65].

Abnormal production of ROS/NOS and subsequent oxidative damage in prion diseases seems to be supported by a parallel, direct impairment of the cellular oxidative stress defense. Consistently, a significant reduction of mitochondrial manganese (Mn) SOD activity was described in experimental models of prion disease [35, 97]. An experimental study revealed that co-incubation of PrPC with the neurotoxic fragment PrP106–126 inhibited the SOD-like activity of PrP [24], suggesting that conversion of PrPC to PrPSc might compromise antioxidant protection, thus leading to oxidative damage and neurodegeneration [17]. In this context, one further observation is noteworthy: by analyzing PrPC expression in human prion diseases, we detected increased intraneuronal immunoreactivity for PrP but not PrPSc [91, 144]. If indeed cellular defense against oxidative stress includes up-regulation of PrPC, antioxidant defense in prion diseases might result in a vicious circle, where accumulation of PrPSc promotes generation of ROS, which in turn induce PrPC expression; newly synthesized PrPC again expedites the formation of PrPSc, which increases the oxidative burden, and so on [144].

Little is known about the molecular mechanisms underlying the impaired cellular response to oxidative stress in TSEs. Since ROS and RNS can both attack a broad variety of lipids, proteins and carbohydrates, not to mention DNA and RNA molecules, it is tempting to assume that members of the antioxidant stress defense system are directly damaged by free radicals, either specifically or just by chance. This was described for instance for Cu,Zn SOD [113] and GPx [4]. A similar mechanism for PrPC was reported recently [108]. Brown proposed an additional possibility for an impairment of antioxidant defense specific for prion diseases [15]. Analyzing possible direct protein-protein interactions, he described a site-specific binding of both the neurotoxic fragment PrP106–126 and PrPSc to amino acid residues 112–119 of PrPC. In culture systems, interaction between PrP106–126 and PrPC inhibited copper binding of the prion protein, thus rendering the cells indirectly more susceptible to copper toxicity. Furthermore, cellular copper uptake and SOD-like enzymatic activity of PrPC were inhibited, two events which might directly and indirectly compromise antioxidant defense in prion disease [15].

Aberrant metal binding by PrP has been suggested as a central molecular event related to the significant decline in antioxidant protection by PrPC. Independent analysis of brains of scrapie-infected mice by two groups revealed significant perturbations of divalent metal ions with a strong decrease in copper content [138, 151] and a major increase in manganese content already in early stages of the disease [138]. These changes were paralleled by a significant decrease in copper binding by PrPC and a proportional decline in its SOD-like activity [138, 151]. Similar results were obtained in brains of patients with sCJD [152]. However, it is still too early to judge whether and to what extent imbalances in metal ions contribute causally to the pathogenesis of prion diseases, or whether they reflect only secondary changes in metal occupancy of PrP subsequent to alterations in PrPC conformation, the latter initiated by other disease-related molecular mechanisms.

Oxidative stress, defined as an imbalance between burden of ROS/NOS and cellular antioxidant defense, is often regarded as an “either-or” mechanism, where either the disproportional generation of oxygen species predominates over the decline in antioxidant protection or vice versa. In prion diseases, the situation is different. Both mechanisms are linked together by the pathological isoform of PrP, PrPSc. Thus, they act cooperatively instead of alternatively. Formation of PrPSc enhances the production of ROS and, subsequently, the rate of oxidative damage to brain tissue. In parallel, the disease-immanent interaction between PrPSc and PrPC inhibits specific functions of normal PrP, including binding and detoxification of copper, cellular uptake and possibly cellular delivery of copper, as well as intrinsic SOD-like activity, altogether a loss of function compromising the antioxidant defense system directly and indirectly. In sum, oxidative stress in prion diseases is a combination of two equally important processes, the increase in oxidative damage and the decrease in protection.

In general, oxidative stress is a mechanism of cell damage, although not a direct mode of cell death. Nevertheless, it may induce neuronal death by initiating one of two pathways: apoptosis or necrosis. Both pathways have been described in the nervous system in response to oxidative damage [131, 134], depending on the individual disease entity. The pathway in prion diseases is discussed in the following section.

The mechanism of neuronal death in TSEs

Two distinct modes of cell death have been described: (1) necrosis, in most instances initiated by a sudden cell or tissue injury and followed by a marked inflammatory response, and (2) apoptosis or programmed cell death, initially regarded as a physiological type of cell death occurring during tissue development and lacking any overt inflammatory reaction [83]. Meanwhile, various conditions have been discovered including loss of trophic support, disturbance of calcium and potassium homeostasis, and cytolethal toxic damage, which all can induce post-developmental apoptosis [131]. However, although primarily separated, both modes of cell death show some molecular and cell biological overlap [127]. The ladder type of DNA fragmentation, for instance, in principle one biochemical hallmark of apoptosis, is a specific sign for internucleosomal DNA strand breaks, but it is neither an obligatory requirement for, nor a feature solely restricted to, programmed cell death. Accordingly, DNA fragmentation was absent in some models otherwise typical for apoptotic cell death [38, 49, 153], but present in certain forms of necrosis [42, 52]. Similarly, the terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) technique (also called in situ end-labeling, ISEL) [54], originally introduced to visualize DNA fragmentation on a histological level, marks DNA strand breaks of every origin and is therefore unable to discriminate unequivocally between apoptosis and necrosis [50, 131]. Furthermore, mitochondrial damage, initially regarded as a mere feature of necrotic cell death, has nowadays “lost” its discriminatory specificity, and is thought to be additionally involved in some apoptotic pathways. In sum, single observations like DNA fragmentation or mitochondrial impairment lack any cogency to classify a certain mode of cell death. Only the combination of several distinct cell death features to an integrative picture offers the possibility to define a pathway with sufficient reliability. This fact has to be kept in mind when assessing the modes of neuronal death in prion diseases.

In 1993, by analyzing the neurotoxic properties of the PrP fragment 106–126 in rat hippocampal neurons, Forloni et al. [51] detected morphological changes and a ladder type of DNA fragmentation in degenerating cells, two features primarily suggestive of apoptotic cell death. In line with these results, DNA fragmentation and DNA strand breaks have been demonstrated with biochemical and/or histological methods (TUNEL labeling) in vitro in neuronal culture systems incubated with either PrP106–126 or PrPSc [33, 132], in murine scrapie [55, 56, 149], and in several forms of human prion disease including FFI [46] and sCJD, fCJD, iatrogenic CJD, and vCJD [50, 60]. In addition to DNA breaks, upon evaluation by light and/or electron microscopy, some of the cell culture and mouse studies showed morphological changes reminiscent of apoptotic cell death [33, 55, 132, 149]. In contrast, a study using human biopsy and autopsy tissue failed to detect any morphological hallmarks of apoptosis in TUNEL-labeled neurons despite a high degree of TUNEL positivity in certain brain areas of single cases [50]. Although these results might be explained by additional DNA damage due to the time interval between agony, death, autopsy and tissue fixation, they have questioned the use of in situ end-labeling as an appropriate screening method for cell death in autopsy tissue in general [50]. To evaluate the position of apoptosis in prion pathogenesis, time course experiments have been performed to determine the temporal and spatial correlation between DNA fragmentation and neuronal loss. In two studies, prominent TUNEL positivity of retinal [55] and hippocampal CA1 [149] neurons preceded a massive cell loss in precisely congruent tissue areas in terminally ill animals. A few more studies analyzed the spatial correlation between the degree of neuronal loss and apoptosis (as revealed by TUNEL staining) in autopsy tissue. The results, however, were incongruent, since two studies from the same group reported a close correlation between the two parameters [46, 60], while a third study described only an inconsistent correlation [50]. As outlined above, these differences might again reflect the limited suitability of DNA fragmentation as a marker for apoptosis in human autopsy tissue.

A more indirect approach to detect apoptosis-related DNA strand breaks was used by Bürkle and colleagues [33]. In primary mouse neuronal cultures, the authors observed a strong nuclear labeling for poly(ADP ribose) 30–48 h after the addition of PrP106–126 in the same subset of cells that showed a morphological phenotype typical of apoptosis. The delay in nuclear staining indicated that this activation of poly(ADP-ribose) polymerase (PARP) occurred primarily in response to DNA fragmentation and not as a result of early toxic effects, like the formation of ROS, induced by the PrP fragment [33].

Unlike these earlier experiments, several more recent studies on apoptosis in prion diseases focused on cell biological events upstream of the final DNA breakage. They describe a variety of pathways possibly involved in the initiation and perpetuation of prion-related apoptotic cascades. However, the current picture is complex and far from being complete. In vitro, analyses were performed in different neuronal culture systems treated with either PrP106–126 or PrPSc. The central and common finding in all these experiments was the activation of members of the caspase protease family, especially the activation of the executioner caspase-3 [43, 73, 112, 147]. Nevertheless, the pathways triggering this caspase activation differed considerably. O’Donovan et al. [112] proposed depolarization of the mitochondrial membrane as the initiating event in PrP106–126-induced apoptosis in SH-SY5Y cells. The depolarization was followed by a release of cytochrome c from the outer mitochondrial membrane with subsequent activation of caspases and a release of mitochondrial calcium with activation of calpains, a second protease family involved in programmed cell death. Other researchers, using the same experimental setup, described a p38 MAP kinase-dependent activation of caspase-3 as an underlying mechanism in PrP106–126-induced apoptosis [43]. Hetz et al. [73] suggested increased endoplasmic reticulum (ER) stress and calcium release from the ER as pathways triggering caspase-12 dependent caspase-3 activation in scrapie-infected N2a cells. To support their hypothesis, the authors also investigated brain tissue from a murine scrapie model and patients with sCJD and vCJD. In line with their in vitro results, they demonstrated a significant up-regulation in the expression of selected ER-stress-associated chaperones, as well as the generation of active fragments of caspase-12 [73].

Considering all results outlined above, the currently available information indicates a crucial role of apoptosis in prion-related neuronal cell death, and suggests that it indeed represents the primary cell death pathway in TSEs. Recently, it was proposed that another form of programmed cell death, neuronal autophagy, could be of importance in prion diseases [99, 133]. This proposal was based on ultrastructural studies of brain biopsy specimen of various CJD patients and one FFI patient, where autophagic vacuoles were found in many synapses, but morphological features of apoptosis were absent [133]. The authors speculated that synapses already damaged by ongoing neurodegeneration are removed by macroautophagy [133]. Such mechanisms may provide an explanation for loss of synapses in TSEs.

Neuroinflammation in prion disease

The microglial response

The presence of an inflammatory response to prion infection in the brain was doubted for a long time. We know now that there are in fact some inflammatory features, but that they differ from those seen in other infective disorders involving the brain. A classical, conspicuous perivascular leukocyte accumulation is normally absent in TSEs. In contrast, microglia, which represent the mononuclear phagocyte system in the CNS, were shown to be activated in human [130] and murine [56] prion disease. The effects microglia elicit in vitro indicate a potent role of these cells in prion pathogenesis. The neurotoxicity of the human PrP fragment PrP106–126 was shown to depend on the presence of microglia which respond to infection by enhanced secretion of oxygen radicals, resulting in oxidative damage to co-cultured neurons [19]. This phenomenon is believed by several researchers to be one of the keys (beside diminished anti-oxidative resistance) to central pathogenesis in TSEs, and was discussed in detail earlier in this review. In vivo findings are also consistent with the proposal that microglia may have a direct impact on neural tissue damage. In the brains of scrapie-infected mice, the microglial response is largely confined to regions with vacuolation and PrP deposition [148]. As time course studies have elucidated, microglia activation precedes neuronal death, and is thus unlikely to be just a sequel of cell destruction in prion disease [56, 149]. On the other hand, Perry and colleagues [116] have argued that the immunological phenotype of microglia in vivo closely resembles that of macrophages having ingested apoptotic inflammatory cells, i.e., a profile including the absence of pro-inflammatory cytokines and concomitant domination of anti-inflammatory pathways, which may be in accordance with the widespread neuronal apoptosis occurring in TSEs. However, the reports on the cytokine expression pattern in prion disease are incongruent [44, 150], maybe due to different animal models and/or detection methods. In vitro infection experiments brought even more contradictory results, revealing an up-regulation of pro-inflammatory cytokines like IL-6 [7, 117], and IL-1β [117]. This might be explained by the “artificial” setup, where an acute response to the acute addition of a neurotoxic peptide is monitored, while the natural process in prion disease is characterized by the slow accumulation of PrPSc with consecutive neurodegeneration [116]. A recent in vivo study has demonstrated a relatively late onset of generally low levels of cerebral cytokine gene expression in the ME7/CV murine scrapie model [12]. Together with the fact that tumor necrosis factor α (TNF-α)-, as well as IL-6-deficient mice are fully susceptible to prion disease when challenged intracerebrally with the ME7 strain [100], this renders a crucial role of pro-inflammatory cytokines in central prion pathogenesis unlikely. Interestingly, it has been proposed that, although in prion disease microglia do primarily not express significant levels of typical pro-inflammatory cytokines, they could nevertheless be in a “primed” state, and, if further stimulated by peripheral infections, secrete inflammatory mediators which promote the neurodegenerative process [116].

Taken together, virtually all results speak in favor of an important role of microglia in the disease process in TSEs, but we do not yet know with certainty, which of the involved pathways can finally be disastrous for the neuronal cell. Even more vague are assumptions about the direct causes of microglial activation. As microglia are closely related to peripheral macrophages, it seems logical that they are responsible for the phagocytosis of cells and/or extracellular material like deposited PrP. In vitro, microglia were shown to internalize to some extent fibrillar PrP106–126, which in turn interferes in a way with the phagocytic process [36]. If this cell type primarily responds to PrPSc with activation and cytokine secretion in vivo is still not clear.

Complement activation

Activation of the complement system has been described in various neurodegenerative disorders including Alzheimer’s and Huntington’s disease [53]. For CJD and GSS, it was demonstrated that factors of the early complement cascades co-localize with amyloid plaques [78]. In a recent study that was done in our laboratory, active complement compounds, like C1q and C3b, were detected in extracellular PrP deposits in CJD, and the most important effector of the complement system, the membrane attack complex (MAC), was found on neurons in human TSEs [93]. Localization of early and late complement components seems to correlate with the severity of disease-specific pathology; in the respective study, areas showing neuronal TUNEL reactivity overlapped with those exhibiting MAC deposits in most of the cases that were examined. Brains without correlation exhibited advanced pathology, with only a few remaining neurons which displayed MAC immunoreactivity. C3b, a member of the early complement cascade, was seen also in better preserved regions. However, the mere presence of active complement components in the brain does certainly not prove their unequivocal role in central pathogenesis of TSEs. It is not surprising that some complement activation should happen under the given circumstances, as the complement system can be activated via oxidative stress [40, 143], which is supposed to be a central event in prion pathogenesis. In turn, complement components stimulate the cellular production of reactive oxygen species [1, 47], so that the complement system may contribute to a vicious cycle of events threatening neuronal cells. On the other hand, mice which are temporarily depleted of C3, as well as mice deficient of C1q, develop full blown scrapie if infected via the intracerebral (i.c.) route [101], and mice lacking early complement components display, despite delayed onset of disease, a cerebral histopathological picture equal to that of wild-type mice after intraperitoneal (i.p.) inoculation [88]. Thus, complement-mediated cell toxicity and cell lysis may be part of the pathomechanisms causing cell death in prion disease, but according to the results obtained from knockout mice, a pivotal role of the complement system seems rather unlikely.

A possible role for the transmembrane form of PrP

PrP exists in multiple topological variants, including a secretory form that is completely translocated through lipid bilayers [68], and two transmembrane forms [67]. A possible relation between these variants and neurodegenerative disease was first investigated by Hegde et al. [70]: certain mutations within the PrP coding region dramatically alter the ratio of the topological forms in favor of one of the transmembrane forms, termed CtmPrP. In mice carrying such mutations, CtmPrP was found to be associated with severe neurodegeneration with some neuropathological features typical of prion disease, while at the same time PrPSc was virtually absent in analyzed brain tissue, suggesting elevated CtmPrP as the primary cause of pathology. Accordingly, in human GSS, the A117V mutation was proposed to lead to a relative preference for the synthesis of the transmembrane forms of PrP. The analyzed GSS brains indeed contained increased levels of CtmPrP, whereas no protease-resistant PrPSc was detectable [70]. Subsequently, a pathogenic mechanism involving CtmPrP was also suggested for other inherited prion diseases, which, in contrast to GSS, feature accumulation of PrPSc, and for infectiously acquired prion diseases [71]. In an elegant inoculation study, double-transgenic mice, carrying both a murine MoPrP and a hamster SHaPrP transgene, were inoculated with mouse RML prions. In this model, accumulation of PrPSc during the incubation period triggered in parallel the de novo formation of hamster CtmPrP as a possible cause for disease development and neurodegeneration [71]. On the other hand, only mutations within or near the transmembrane domain of the PrP sequence were found to enhance the formation of CtmPrP [70, 135], while pathogenic mutations in other regions showed no effect with respect to the formation of transmembrane forms of PrP [135]. Furthermore, cell culture and in vivo studies, where the amount of CtmPrP is not altered after prion infection, speak against a general, obligatory role of this molecule in TSEs [136]. Thus, the pathogenic relevance of CtmPrP for most prion diseases is still unclear. Similarly, the mechanism by which transmembrane prion protein might evoke cell death is presently not known. While some authors denied any significant accumulation of CtmPrP in the ER [70, 71], others reported it to be retained in the ER and suggested that it could stimulate the activation of pro-apoptotic, ER stress-response pathways [137]. An involvement of ER stress and caspase-12 activation in the brain was recently demonstrated for sCJD and vCJD [73].

Given all the inconsistencies mentioned above, further research is necessary to delineate whether accumulation of PrPSc in prion diseases indeed induces the enhanced formation of CtmPrP and, if yes, whether CtmPrP really represents a potent neurotoxic agent in TSEs.

Peripheral pathogenesis of prion diseases

In the previous sections, we discussed the current knowledge about central pathogenesis in prion diseases. These mechanisms describe pathogenic events occurring in sporadic or genetic TSEs, and perhaps some types of iatrogenic CJD. However, other forms of TSEs also exist in which prion infectivity is not generated “endogenously”, but is acquired from outside the host organism. This group includes vCJD and Kuru in humans and BSE, scrapie, and chronic wasting disease (CWD) in animals. In these instances, a process designated as peripheral pathogenesis precedes the disease process in the CNS.

Many experimental modes of peripheral infection (like skin scarification, conjunctival instillation, and intravenous, perivenous, or intraneural injection) have been established to study peripheral pathogenesis in detail. Of these routes, oral transmission of prions and the experimentally related, but not completely identical i.p. route are of major importance for the analysis of naturally occurring peripherally acquired TSEs. Here we focus mainly on these two routes, and briefly summarize the current concept of peripheral pathogenesis, including (1) possible entry sites for infectivity, (2) cellular transport mechanisms, (3) peripheral target organs, (4) cellular accumulation and replication of prion infectivity, and (5) neuroinvasion by prions.

Entry of infectivity

To enter an organism from the inside of the gastroenteric tract, prions have to cross an epithelial barrier consisting of a monolayer of enterocytes. Cells which may support this process are the so-called membranous epithelial cells (M cells), a population of cells involved in transepithelial transfer of enteric antigens and pathogens to the mucosa-associated lymphoid tissue system (MALT). To investigate the properties of M cells in vitro, a few years ago, a group of researchers developed an elegant co-culture model of human colon carcinoma (Caco-2) cells and human lymphoblastoid Raji cells [82]. Initially, the cells grow independently in two compartments separated by a filter membrane containing multiple micropores, thus forming two independent cell layers. During the following culture process, single lymphoblastoid cells migrate through the micropores to the basolateral side of the tightly closed Caco-2 cell layer and induce differentiation of adjacent Caco-2 cells into M cells. At this stage, the co-culture represents an ideal model to study the capability of M cells for transporting macromolecules and pathogens. Transferring this approach to TSE research, Heppner and colleagues [72] demonstrated an efficient transepithelial transport of prion infectivity through differentiated M cells when the cultures were incubated with a high dose of RML scrapie prions, suggesting M cells as a potential candidate for the enteric entry site of orally consumed prion infectivity. Recently, an alternative scenario was proposed, putting forward the involvement of migratory bone marrow-derived dendritic cells (DCs) in enteric prion invasion [77]. In general, DCs are capable of migrating from the blood vessel compartment to the inner surface of the intestinal wall and trapping antigens in the enteric tube; these antigens are subsequently transported to the local lymphoreticular system including mesenteric lymph nodes and probably more distant sites. Combining in vitro and in vivo inoculation experiments, Huang et al. [77] showed that DCs can acquire PrPSc and transport infectivity from the gut lumen directly to lymphoreticular tissues. Whether or not these pathways contribute to the intestinal entry of prions in humans remains to be established. In this context, it is important to note that orally administered prion infectivity can immediately get into direct contact with the lymphoreticular system at the site of the palatine tonsils, thereby perhaps bypassing the intestinal pathway. Certainly, the contact time between prions and the tonsil surface is very short; nevertheless, prions have been shown to stick to several surfaces very rapidly, rendering the tonsils an additional plausible entry port of orally acquired infectivity. The fact that PrPSc has been consistently detected in tonsil biopsies of patients with vCJD might further point towards this direction [74].

Cellular transport mechanisms

Apart from migratory DCs [77], lymphocytes and macrophages have been implicated as vectors for the initial transport of infectivity to lymphoid organs. However, the available information is incomplete, and research on this topic faces a specific difficulty: prion transport does not necessarily coincide with prion replication; there is not necessarily an accumulation of infectivity in the transport system itself. On the contrary, primary spread of prions and secondary propagation of infectivity are most likely separated processes. Thus, initial vectors presumably contain only traces of infectivity, and more sensitive detection methods are needed to determine which cell type contributes to this initial spread of prions and which does not.

Peripheral target organs for prion infectivity

After oral challenge, prion infectivity usually colonizes the lymphoreticular tissue early in the incubation period. Accordingly, PrPSc accumulation has been demonstrated for example in the distal ileum, the spleen, and lymph nodes of scrapie-infected sheep [141], as well as in the ileum of cattle with experimental BSE [145]. In the ileum, Peyer’s patches (PPs) constitute the main representative of the MALT system, and thus the primary site of intestinal PrPSc deposition. The importance of PPs for oral transmission in prion diseases has recently been emphasized by Prinz and colleagues [119], who analyzed knockout mice with either destroyed architecture but normal number of PPs, or with a significant reduction of PPs, and found that only the number of PPs, but not their cellular composition, especially with regard to the presence or absence of PP-associated B lymphocytes, is crucial for the development of prion disease.

Cellular accumulation and replication of prion infectivity

Decades ago, the lymphoreticular tissue, including spleen and lymph nodes, was recognized as the major site of prion accumulation in the body periphery. Nevertheless, the cellular background responsible for the amplification of infectivity has remained obscure for many years. As a consequence, speculations about several cell types, ranging from migratory cells like splenocytes, lymphocytes or macrophages to resident cells of the lymphoid tissue stroma, were put forward for quite a long time — despite the fact that early fractionation experiments with spleen tissue from scrapie-infected mice revealed a prominent role for the stromal compartment, but not the pulp fraction, as a reservoir for prion infectivity [37]. The credit for unraveling this issue goes primarily to the group of Kitamoto [85]. At a time when sophisticated genetically modified mouse models like transgenic or (conditional) knockout mice were not yet readily available, the simple inoculation of two mouse strains with impaired immune systems via two different routes, combined with the development of immunohistochemical staining techniques capable of differentiating between PrPC immunoreactivity and PrPSc accumulation in frozen and paraffin-embedded tissue sufficed to imply follicular DCs (FDCs) as the central cell population in prion propagation in peripheral lymphoid organs. In a first set of experiments with spleens of CJD-infected wild-type mice, Kitamoto and colleagues [85] reported that PrPSc-specific fluorescence labeling was selectively restricted to the germinal centers of lymphoid follicles, to cells morphologically reminiscent of FDCs. These results were further supported by co-localization studies with antisera against PrPSc and TP-3, a marker recognizing DCs in lymphoid tissue. Subsequently, Kitamoto’s group performed inoculation studies comparing the transmission efficiency of i.c. and i.p. administered CJD prions in wild-type mice, nude mice, and mice with severe combined immunodeficiency syndrome (SCID). T cell-deficient nude mice showed incubation times similar to wild-type controls on either route (i.c. and i.p.) [109]. SCID mice, harboring a combined deficiency for B and T cells, could easily be infected by i.c. administration of prions, but did not develop any signs of clinical disease after i.p. challenge, even after very long observation periods. In line with these surprising clinical findings, immunohistochemical and immunoblot analysis revealed no accumulation of PrPSc in brain and spleen of i.p. inoculated, and spleen of i.c. infected SCID mice [85]. The number of FDCs in the spleen of SCID mice did not differ significantly from that in wild-type mice. Thus, the discrepancy in CJD susceptibility could not be explained by a mere decrease in FDCs. On the other hand, previous studies had indicated that the development of mature FDCs is B cell dependent. Kitamoto et al. [85] concluded that the depletion of functional B cells in SCID mice with consecutive impairment of FDC maturation was most likely responsible for the inability of these animals to replicate infectivity, as well as for their consecutive resistance to the development of prion disease after peripheral exposure. Meanwhile, this concept has been validated and extended by many researchers. Accordingly, in a huge study using SCID mice and several genetically engineered knockout mice, selective or combined B cell deficiency, but not selective T cell deficiency, prevented the development of clinical scrapie after i.p. inoculation [86]. A year later, the same group showed by means of hematopoietic reconstitution experiments that repopulation of B cell-deficient mice with fetal liver cells containing B cell progenitors restored susceptibility to scrapie after i.p. challenge irrespective of the fact whether these B cells expressed PrPC or not [87]. In an approach based on chimeric mice with a mismatch in PrPC expression between FDCs and other immune cells, Brown et al. [26] demonstrated that replication of mouse-adapted ME7 prions strictly depends on PrP expression of FDCs, but not of lymphocytes. In line with these results, a recent study using immunoelectron microscopy revealed an extensive accumulation of presumably disease-specific PrP in the extracellular space around highly reactive or hyperplastic FDC processes and a moderate PrP labeling of the FDC plasmalemma in spleens of terminally diseased mice inoculated i.c. with ME7 prions [79]. These observations at the final stage of the disease are in contrast with sparse labeling of mainly simply structured FDC dendrites in the spleen of mice in the pre-clinical phase, suggesting that pathological PrP may be continuously released from the cell surface of FDCs during the whole course of the disease and accumulates extracellularly to a high titer plateau [79]. During the last years, the role of B cells with respect to the maturation of FDCs and organization of splenic germinal centers has been further clarified. According to the current scientific view, the development of lymphatic follicles with mature FDC networks depends on the expression of lymphotoxin (LT)-α/β heterotrimers and tumor necrosis factor (TNF) by differentiated B cells, and of LT-β receptors (LT-βR) by FDCs [48, 59, 106]. This concept initiated further experiments in TSE research. In one mouse model, mature FDCs were successfully depleted by repeated administration of soluble LT-βR. This treatment significantly delayed disease development and prevented detectable splenic prion replication in mice challenged with scrapie prions via the i.p. route, further indicating a central role for FDCs in splenic prion amplification [110]. Complete prevention of prion disease development and accumulation of prion infectivity in spleens and lymph nodes after i.p. inoculation was finally shown in mice lacking different members of the LT signaling cascade (LT-α, LT-β, LT-βR) [118]. Despite their central role in peripheral prion pathogenesis, FDCs are not necessarily the only site of prion replication and accumulation in the periphery. A recent study demonstrated that other immune cells can replace FDCs under certain experimental conditions. Mice with impaired TNF signaling, lacking either TNF-α or TNF receptor 1 (TNFR1), do not develop mature FDC networks and germinal centers in lymphoid follicles of spleen and lymph nodes. Nevertheless, the majority of these mice (i.e., all TNFR1- and some TNF-α deficient mice) were fully susceptible to i.p. prion infection with incubation times comparable to wild-type controls. In addition, all mice accumulated high titers of prion infectivity in lymph nodes, but not in the spleen [118]. In a further experiment, transferring fetal liver cells to PrP knockout mice, the same researchers detected co-localization of PrP with a subset of macrophages in lymph nodes, suggesting that these macrophages might participate in lymphatic prion replication and neuroinvasion [118]. Whether these results are relevant only in the artificial absence of mature FDCs, or whether they document an alternative pathway bypassing FDCs remains to be established.

Neuroinvasion by prions

Some years ago, Blättler and colleagues [10] reported that transplantation of PrP-positive hematopoietic cells into PrP knockout mice sufficed to support prion propagation in the spleen after i.p. inoculation, but failed to induce spongiform encephalopathy in PrP-producing neurografts implanted into the brains of these mice, indicating that a further PrP-expressing tissue compartment, such as the peripheral nervous system (PNS), is necessary for neuroinvasion of prions. The exact nature of prion transfer to the PNS is currently poorly understood. So far, two types of mobile cells have been identified that are capable of acquiring and transporting prion infectivity to lymphoid tissues, namely macrophages [118] and DCs [77]. Therefore, it is tempting to speculate that at least one of these cell types might also be responsible for the delivery of prion infectivity to peripheral nerve terminals. In support of this concept, it was recently demonstrated that living DCs, isolated from spleens of scrapie-infected mice and injected into B cell-deficient mice, successfully transferred prion disease to the previously prion-resistant host without detectable splenic prion amplification [5]. In this context, it is noteworthy that mobile DCs might also contribute to spread of prion infectivity from the CNS centripetally to the periphery, as was suggested by a recent study from our laboratory [89].

Apart from cellular components of the immune system, the innate immune system, and here especially the complement system, was reported to play a supportive role in the early phase of peripheral prion pathogenesis. Two independent studies revealed that splenic prion propagation and disease onset after i.p. exposure were significantly delayed in mice with temporary or permanent ablation of early components of the complement system or complement receptors [88, 101]. These results suggested that opsonization of the infective agent with C3d/C4b facilitated transport to and uptake of infectivity by the lymphoreticular system.

The PNS represents the second system (apart from the immune system) necessary for peripheral prion pathogenesis [10]. However, the role of the PNS seems to be restricted to neuroinvasion, to bridging the gap between the periphery and the CNS, and perhaps to some degree to replication and accumulation of infectivity [58]. In general, several pathways for peripheral neuronal prion transport have been described, including the sympathetic nervous system, the vagus nerve, and the sensory nervous system. The reported results show some discrepancies, since some researchers observed an exclusive participation of one of these pathways in prion spread, while others found a parallel or sequential involvement of two or three PNS compartments. The incongruent results may be related to the different routes of infection used in individual experiments, and perhaps by the main focus the researchers put on the different peripheral neural pathways. When prions were administered orally, initial neural spread was reported to occur along the vagus nerve to parasympathetic nuclei of the medulla oblongata, followed by neuroinvasion along the sympathetic nervous system [8]. In the same animal model, in an advanced stage, prominent PrPSc deposition was observed in sympathetic and sensory ganglia, while vagal nerves showed only sparse labeling for pathological PrP [107]. After i.p. exposure to prions, the sympathetic nervous system was shown to transport prions exclusively [39, 58] or coincidentally [6] to the CNS. Finally, intraneural injection of prion infectivity proved that prions are efficiently, albeit “slowly” transported via ascending sensory nerve tracts [84, 96]. However, the exact mode of neuroinvasion along nerve fibers is currently enigmatic. Whether the occasional detection of PrPSc in peripheral nerve tissue in fCJD [66] and vCJD [69] represents centripetal or centrifugal spread of prions remains to be established. For the CNS, we could recently document granules of PrPSc within axons, so that intra-axonal transport of prions might be important for their spread not only in the CNS, but also in the PNS [94].

Conclusions

Looking back at two decades of intensive research activities devoted to the understanding of prion diseases, the general scientific approach to prion pathogenesis has changed significantly. In former times, investigators tried to explain peripheral prion pathogenesis as a simple process, in which a chain of mobile and resident cells transports prions along one single route from the periphery to the CNS. Similarly, the mechanism of neurodegeneration was reduced to mutually exclusive alternatives: either “loss of function” of PrPC or “gain of function” (i.e., neurotoxicity) of PrPSc. Today, these one-dimensional concepts have been replaced by a more holistic point of view. We now have sufficient evidence that many, if not all aspects of prion pathogenesis share parallel pathways. In the body periphery, at least two cell types (M cells or DCs) have been identified as potential entry portal for prion infectivity. Subsequently, PrPSc may either be locally amplified prior to further spread, or directly transported to various lymphoreticular tissues, and perhaps also to nerve terminals. Accordingly, neuroinvasion of prions was shown simultaneously along different fiber tracts of the PNS. In the CNS, several potential mechanisms of nerve cell damage have been identified, including the coincidence of reduced antioxidant protection and increased oxidative stress, induced by the formation of PrPSc, as well as by PrPSc-dependent activation of microglia. However, there is currently no well-established model to explain nerve cell death in TSE completely. At present, research results in prion pathogenesis appear to be somewhat more advanced with regard to the periphery as compared to the CNS. Nevertheless, further efforts are needed in both areas to elucidate the concerted action of all pathogenic mechanisms finally contributing to the development of prion disease. Only this will provide a rational basis for the development of effective therapeutic strategies.

Copyright information

© Springer-Verlag 2005