M. Jeffrey, Veterinary Laboratory Agency, Lasswade laboratory, Pentlands Science Park, Bush Loan, Penicuik, Midlothian, Scotland EH26 0PZ, UK. Tel: +44 131 4456169; Fax: +44 131 4456166; E-mail: firstname.lastname@example.org
Scrapie is a prion disease or transmissible spongiform encephalopathy (TSE) of sheep, goats and moufflon. As with its human counterparts, pathology consists of vacuolation, gliosis and accumulations of abnormal forms of a host prion protein (PrPd) in the brain of affected individuals. Immunohistochemical methods can be used to identify both the intracellular truncation sites of PrPd in different cell types (PrPd epitope mapping) and the different morphological patterns of accumulation (PrPd profiling). Differences in the inferred truncation sites of PrPd are found for different strains of sheep TSEs and for different infected cell types within individual strains. Immunochemical methods of characterizing strains broadly correspond to PrPd mapping discriminatory results, but distinct PrPd profiles, which provide strain- and source-specific information on both the cell types which sustain infection (cellular tropisms) and the cellular processing of PrPd, have no immunoblotting counterparts. The cause of neurological dysfunction in human is commonly considered to be neuronal loss secondary to a direct or indirect effect of the accumulation of PrPd. However, in sheep scrapie there is no significant neuronal loss, and relationships between different magnitudes, topographical and cytological forms of PrPd accumulation and clinical signs are not evident. PrPd accumulation also occurs in lymphoid tissues, for which there is indirect evidence of a pathological effect, in the peripheral nervous system and in other tissues. It is generally assumed that neuroinvasion results from infection of the enteric nervous system neurones subsequent to amplification of infectivity in lymphoid tissues and later spread via sympathetic and parasympathetic pathways. The evidence for this is, however, circumstantial. Accumulation of PrPd and presence of infectivity in tissues other than the nervous and lymphoreticular systems gives insights on the ways of transmission of infection and on food safety.
The transmissible spongiform encephalopathies (TSEs) or prion diseases are a group of progressive and invariably fatal neurological diseases of animals and human. Classically, these diseases are associated with an abnormal accumulation of a host-coded cell surface glycoprotein called prion protein (PrP) and a triad of histological lesions comprising vacuolation, neuronal loss and astrocytosis. In contrast to sporadic Creutzfeldt–Jakob disease (CJD) of human but in common with variant CJD (vCJD), sheep scrapie is associated with widespread tissue infectivity.
There are numerous mutations of the PrP gene which have been reported in familial prion disease of human, butonly the polymorphisms at codons 129 and 219 of the PrP protein appears to be of importance in iatrogenic, sporadic and variant forms of CJD and in kuru. In contrast, the PrP gene of sheep is highly polymorphic, and there are several sites which influence disease susceptibility . The codons which have most impact are located at positions 136, 154 and 171, with alleles VRQ and ARQ (where letters stand for single letter codes for amino acids at these codons) being most frequently associated with disease. The ARR allele is associated with resistance to clinical disease, and there are no fully substantiated cases of natural disease in ARR/ARR homozygotes.
The cause of the TSEs is now generally, though not unanimously, considered to be a prion. The prion is an infectious variant of PrP, but the use of different nomenclatures in respect of abnormal prion protein can be confusing. The commonly used term PrPscrapie (PrPsc) and associated homologues in different species (PrPCJD, PrPCWD, PrPBSE, etc.) are often used as a substitute for infectivity. Equally frequently, PrPsc is used to mean any combination of the biophysical properties of partial proteinase-resistance, aggregation, insolubility and an increased proportion of β-pleated sheet which define abnormal PrP extracted from infected tissues. Proponents of the prion hypothesis now assert that these biophysical properties of abnormal PrP should only be considered as surrogate markers of infectivity because one or more of these properties may not always be present . Some now suggest that most infectivity is present in protease-sensitive forms of abnormal PrP denoted as sPrPsc[2–4]. For simplicity, we will use the operational definition PrPres to define abnormal protease-resistant PrP, where identification involves the use of a digestion step or the term disease-associated PrP (PrPd) to denote the presence of abnormal accumulation or isoforms of PrP by any method without prejudice as to its biophysical properties, its infectivity, source or host range. We use PrPc to denote PrP found in normal cells and uninfected animals.
For many years, a major challenge to the prion hypothesis has been to explain how an abnormal protein could code for the existence of numerous, stable, well-characterized rodent scrapie strains. According to the prion hypothesis, strain-specific information is encoded within the conformation of different tertiary or quaternary forms of an infectious variant of PrP [5,6]. Murine scrapie strains may be characterized by their relative incubation periods in different inbred mouse lines, different patterns of pathology and immunochemical forms of PrPres[7,8]. A smaller number of laboratory sheep scrapie strains may be identified by their relative incubation times in different PrP genotypes  and pathological phenotypes . SSBP/1, CH1641 and cattle bovine spongiform encephalopathy (BSE) have been transmitted to sheep many times and are sufficiently well characterized to be considered as strains, but most field isolates of natural sheep scrapie have not been passaged and we regard these as sources. All throughout this review, these experimental strains will be considered together with natural scrapie sources as aetiological agents of classical sheep TSEs.
Currently, prion disorders of sheep may be divided into those classically recognized and the recent descriptions of Nor 98  or atypical scrapie. Classical scrapie is transmissible and contagious, characterized by variable neurological signs, classical spongiform changes in the brain and accumulation of PrPd in many tissues, and PrPres is detected following robust digestion conditions . Nor 98 scrapie or the unfortunately adopted term atypical scrapie has mostly been recognized by screening methods applied to cull animals , and only limited information is therefore available. Neurological signs may occur but are poorly defined . Classical vacuolation is absent, although a different form of microvacuolation is inconsistently present . PrPres can be detected but only under mild denaturing conditions, and PrPd is not detected outside the central nervous system (CNS). Epidemiology shows that atypical scrapie is present throughout Europe at about 1/10 000 of the sheep population, irrespective of the concurrent presence of classical scrapie. Although infectious, this prion disorder does not appear to be contagious and mainly involves sheep PrP genotypes considered resistant to classical scrapie .
In this review, we will describe the pathology of classical sheep TSEs in the CNS, and its relationships with strain diversity and clinical disease. We also describe changes in the peripheral nervous system (PNS), in the lymphoreticular system (LRS) and in other tissues, and their relevance to neuroinvasion, food safety and disease transmission.
Pathology of the CNS
The morphological pathology of TSEs in sheep and other species has been extensively described, and is characterized by vacuolation and astrocytosis, whereas neuronal loss is not usually conspicuous. The proportions of neuropil and neuronal vacuolation and their neuroanatomic distribution may differ according to source or strain , although individual variability is very high . Unlike in mice, where it is possible to use the pattern of vacuolation to characterize strain, that variability precludes the use of vacuolar profiling as a strain-typing tool for sheep scrapie.
The other hallmark of sheep TSEs is the accumulation of PrPd in the brain (Figure 1). In common with previous studies of orally challenged rodents and cattle, pathogenesis studies of naturally occurring and orally challenged sheep show that initial PrPd accumulation in the CNS occurs in the parasympathetic or dorsal nucleus of the vagus nerve and in the sympathetic pathways of the intermediolateral columns of the thoracic spinal cord . As both of these sites have connection with the enteric nervous system (ENS), via the vagus or splanchnic nerves, respectively, and because the ENS is consistently PrPd positive in most infected animals, it is widely assumed that neuroinvasion follows infection of the ENS. The evidence is therefore circumstantial, and our own ongoing studies on preclinically infected sheep indicate that the initial target sites of PrPd accumulation in the brain are the same regardless of the route of inoculation. These findings point towards alternative, perhaps simultaneous or complementary, routes of neuroinvasion. With increasing duration of infection, PrPd accumulation becomes more widespread, and the overall magnitude and distribution are affected by route of inoculation, the PrP genotype of the host and the source or strain of the infection. Thus, for experimental sheep BSE, intracerebral challenge results in overall greater amounts of PrPd accumulation, particularly in fore-brain areas, when compared with oral or intravenous routes . The PrP genotype also affects the magnitude of PrPd accumulation although the precise effect depends also on the incubation period: for example, the total levels of PrPd after intracerebral BSE infection are similar in clinically affected ARQ/ARQ sheep and in VRQ/VRQ sheep, but these last have a much longer incubation period, comparable with that in ARR/ARR sheep, which, however, display much lower magnitudes of PrPd accumulation .
Phenotypes of PrPd accumulation in the brain
The detection of PrPd can be achieved by several methods, including: (i) the presence of characteristic fibrils by electron microscopy in tissues and extracts ; (ii) immunochemistry – Western blotting and different immunoenzymatic assays – in brain homogenates; and (iii) immunohistochemistry (IHC) in tissue sections. While most immunoenzymatic methods – the so-called ‘rapid tests’– are used for the active screening of sheep TSEs, the others – including IHC – are officially considered as confirmatory tests. However, these techniques are increasingly being used to obtain additional information on TSE strain diversity and the role of host and environmental factors on TSE phenotypes. Thus, molecular TSE phenotypes, based on differences in Western blot glycoprofiles, are regarded by some as indicative of the existence of different human TSE strains and are currently used to discriminate between classical scrapie, atypical scrapie and sheep BSE.
At clinical stages of disease, IHC phenotypes in the brain of TSE-affected sheep can be defined on the basis of the immunoreactivity to antibodies recognizing different epitopes of PrPd, and of the recognition and scoring of different morphological, topographical and cell-associated PrPd types. These two approaches are termed, respectively, PrPd epitope mapping and PrPd profiling. While these methods allow the characterization of immunohistochemical phenotypes and the subsequent interpretation of the host and infectious agent factors that account for them, they rely on good-quality tissues and are subjective, which compromises comparisons between different laboratories.
PrPd epitope mapping
Numerous monoclonal and polyclonal antibodies are currently available for the IHC detection of PrPd in tissue sections. These antibodies are elicited against, and react to, different epitopes of the sheep prion protein, which may or may not be available for immuno-recognition, depending on extrinsic (TSE agent or strain) or cellular factors. Thus, the patterns of immunoreactivity with a panel of PrP antibodies recognizing different epitopes can provide the basis to differentiate between sheep TSEs. Some of the most commonly used antibodies in sheep TSEs and their corresponding epitopes are listed in Table 1.
Table 1. Antibodies to PrP (*rabbit polyclonals; **rat monoclonal; all others, mouse monoclonals), indicating the epitope to which they react according to the work by the referenced authors
A standard panel of PrP antibodies include those directed to four different clusters of epitopes according to their distribution within the PrP molecule: (i) downstream (numerically lower amino acid sequence) N-terminal epitopes, such as those recognized by antibody BG4 or FH11); (ii) upstream (numerically higher amino acid sequences) N-terminal epitopes, such as those reacting to antibody SAF 32, P4, 12B2, 521, IdPrP4, 8G8 or 505; (iii) globular domain epitopes, such as those recognized by antibody F89, 6H4, L42, Bar224 or 2G11; and (iv) C-terminal epitopes, such as those reacting to antibody R145. When such a panel of antibodies is used on serial sections of brain affected by different TSE agents or sources (natural and experimental scrapie and experimental BSE), the following patterns of immunolabelling – summarized in Table 2 and illustrated in Figure 2– can be observed:
Table 2. Epitope mapping results for different sheep TSEs
PrPdtype and infection
Summary of reactivities of extracellular and intraneuronal PrPd to antibodies recognizing epitopes in different clusters of the PrP molecule, as indicated in Table 1.
• Extracellular and cell membrane-associated deposits in the grey and white matter are revealed with antibodies to all the four epitope clusters described above in all sheep TSEs examined, with the exception of CH1641 infection [18,19]. In contrast, antibodies to the downstream N-terminus of PrP do not recognize intraneuronal, intra-astrocytic or intramicroglial PrPd, suggesting that cell-membrane and extracellular PrPd are full-length protein, whereas intracellular PrPd is cleaved at the N-terminus. This agrees with the predominant intralysosomal location of PrPd observed by electron microscopy in mouse  and sheep scrapie (Jeffrey et al. 2007 submitted), and also with all documented evidence on ‘in vitro’ digestion of PrPd with proteolytic enzymes. The almost complete absence of detectable extracellular PrPd in CH1641 experimental scrapie (Figure 2) would be due to a peculiar cellular processing of abnormal PrP in this infection, where all PrPd generated would accumulate intracellularly.
• Intraneuronal PrPd accumulation is detected with some antibodies but not with others, depending on the infecting agent or strain. Thus, all upstream N-terminal, globular domain and C-terminal antibodies detect with similar efficiency for intraneuronal (and intraglial) deposits in all natural scrapie cases (Figure 2) examined and in experimental SSBP/1 infection. However, P4 antibody (reacting to PrP amino acid sequences 93–99) does not detect intraneuronal PrPd either in BSE [18,19] or in CH1641 infection, and IDPrP4 monoclonal consistently reacts with intraneuronal PrPd in BSE, but most neurones are negative by IDPrP4 in CH1641 infection (Figure 2). Accumulation of PrPd within the cytoplasm of neurones is prominent in this last experimental scrapie infection, but it is only revealed with antibodies to globular domain and C-terminal epitopes. These findings suggest that the principal intraneuronal cleavage sites of the N-terminus of PrPd occur at different points of the amino acid sequence, depending on the infecting strain; because the host factors – ovine neurones – are the same in all cases, the distinct PrPd truncation patterns are presumably due to differences in the conformation of the PrPd that is generated in those different infections. While immunoblotting methods for PrPres are also able to discriminate BSE and CH1641 from natural scrapie, they do not reliably differentiate between BSE and CH1641 .
• In experimental ovine BSE, but not in natural or SSBP/1 scrapie (and so far undetermined for CH1641 infection), the inferred cleavage site of PrPd is further upstream in astroglia and microglia than in neurones. Thus, in sheep BSE, intraneuronal PrPd is labelled with IDPrP4 whereas intraglial PrPd is not. These different cellular labelling reactions are not apparent when immunoreactions are performed with globular domain or C-terminal antibodies such as 6H4 or R145, respectively. Further differences have been observed, also in experimental ovine BSE, when comparing brain cells and tingible body macrophages (TBMs) in LRS tissues (Figure 3) . This suggests that, at least in some sheep TSEs, intracellular truncation occurs at different segments of the N-terminus of PrP depending on the cell type involved. These differences may be due to variability in the denaturing conditions of lysosomes in different cell types, or to differences in the tertiary or quaternary structure of PrPd associated with diverse cells.
To summarize, the IHC approach of PrPd epitope mapping allows differentiation between some sheep (and also caprine and cervine) TSEs, based on the intracellular cleavage patterns of PrPd, which are different for natural and SSBP/1 scrapie, CH1641 scrapie and sheep BSE. As far as ovine BSE is concerned, experimental evidence shows that these differential immuno-reactivity properties of intraneuronal PrPd are maintained after primary and secondary passage in sheep . These peculiarities of enzymatic cleavage of BSE-derived PrPd have been observed also ‘in vitro’ and extended to differential diagnosis between scrapie and BSE by immunochemical techniques. It is also likely that, as more antibodies to selected epitopes of the ovine PrP molecule become available, the possibilities to distinguish between different sources of natural sheep scrapie will increase.
It has been recognized for some time that accumulation of PrPd in the sheep brain occurs as different morphological, topographical and cell-associated forms or types (Figure 1) [23–25]. Initial studies, however, did not address whether or not those different types occur at random or are arranged to constitute particular IHC phenotypes. If the event that different patterns of morphological forms of PrPd constitute different phenotypes, it would become possible to determine which factors – host-associated or extrinsic factors – are influential. Defining the IHC phenotype of PrPd accumulation – the PrPd profile – involves: (i) the definition of the different types of PrPd that accumulate in the brain of sheep clinically affected with TSEs (Figure 1); (ii) the application of a systematic method for their quantification; and (iii) the determination of the neuroanatomical areas where the protocol is to be applied.
In recent studies [9,16,19,26], several different morphological types of PrPd accumulation have been described in detail and grouped into four different cell-specific, cell-membrane or extracellular PrPd patterns and three intracellular, truncated forms of PrPd. The cell-membrane/extracellular PrPd patterns are either neuropil-associated (linear, fine punctate, coarse particulate, coalescing and perineuronal types), glial cell-associated (stellate, perivacuolar, subpial, subependymal and perivascular types), ependymal cell-associated (supra-ependymal type) or endothelial cell-associated (vascular plaques type) (Figure 1). The intracellular patterns are the intraneuronal, the intra-astrocytic and the intramicroglial types of PrPd accumulation. Ultrastructural studies indicate that the majority of extracellular PrPd deposits are in intimate contact with the cell membrane of affected cells, irrespective of the light microscopic morphology of the accumulation (Jeffrey et al. submitted and Figure 4). The magnitude of accumulation of the different PrPd types is scored as 0 (absent), 1 (mild), 2 (moderate) or 3 (severe) in six different neuroanatomical sites: frontal cerebral cortex, corpus striatum, thalamus/hypothalamus, midbrain, cerebellum at the vermis and medulla oblongata at the obex. Average values of each PrPd type are obtained for the whole brain and are graphically represented to provide the profile (Figure 5). These can be expressed as absolute values, in which case they also indicate overall magnitude of PrPd in each brain, or as percentage values of the total PrPd present in each brain.
Extensive studies have been carried out in several natural and experimental sheep TSEs, in which the PrPd profiling approach has been used to define their IHC phenotype. In summary, these studies show that:
•Experimental ovine BSE gives the same PrPd profile, with abundant intraneuronal, intramicroglial and neuropil-associated types, regardless of factors such as the source (cattle or sheep), type (brain or blood) and dose of inoculum, the route of challenge (oral, intracerebral or intravenous), the breed of sheep and their PrP genotype [16,27]. Therefore, the PRNP genotype-derived phenotypic variability observed in murine BSE  does not apply to sheep. In contrast to prion hypothesis predictions, different sheep PrP haplotypes do not alter the specific PrPd profile of BSE.
• The PrPd profile of experimental CH1641  and SSBP/1 [9,26] infections, with clear predominance of intracellular PrPd types, is also consistent despite host and environmental factors, and clearly different from that of sheep BSE. However, some phenotypic variation is observed for SSBP/1 infection, which is apparently related to the polymorphism at codon 136 of ovine PrP (valine or alanine).
• The PrPd profile of natural scrapie in ARQ/ARQ Suffolk sheep, with clear predominance of cell-membrane/extracellular, glia-associated PrPd, is different from that in VRQ/VRQ Cheviot or Welsh Mountain sheep, where vascular amyloid plaques are conspicuous (Figure 5). These are also different from the profiles of experimental sheep TSEs described above and also shown in Figure 5.
The diversity of PrPd profiles observed in sheep brains results from differences in cell tropism and PrPd processing between infections by different TSE agents or strains . This appears to be the case when comparing stable experimental agents, but it is more problematic to interpret for natural infections, in which numerous factors are not controlled. It is therefore premature to dismiss the influence of the PrP genotype on PrPd profile in natural scrapie.
Thus, detailed IHC studies of sheep brains combining two different approaches, the PrPd epitope mapping and the PrPd profiling, allow analysis of the factors that influence the phenotype of PrPd accumulation and help in the definition of TSE strains in their original ovine host.
Molecular phenotypes are immunochemically defined in proteinase-treated tissue extracts by the relative proportions of three possible glycosylation states and the migration of the aglycosyl fragment of PrPres. Differences in molecular phenotypes are often used to infer strain characteristics in prion diseases of human and animals. The differences described above in the N-terminal truncation of intraneuronal PrPd of different sheep TSE strains broadly correspond with the migration of the aglycosy fragment of PrPres. Thus the migration of BSE and CH1641 PrPres, which are often indistinguishable, corresponds to a molecular weight of approximately 2 kDa less than that of other sheep scrapie sources. Similarly, although IHC profiles show different IHC phenotypes, natural sheep scrapie sources all give a relatively homogeneous Western blot profile. The IHC peptide mapping clearly indicates that PrPd accumulates as a combination of different proportions of heterogeneous populations of PrPd aggregates, only a minor fraction of which is truncated in vivo. The PrPd profile informs on both the proportions of different cell types which are infected and on the proportions of PrPd which are released and endocytosed from the cell membrane. It appears that molecular phenotypes, as currently constructed, do not differentiate between different cell types which release PrPd or between the proportions of released and endocytosed PrP.
The causes of neurological deficits in sheep TSEs
Although standard texts often suggest that clinical disease is distinctive, there is marked variation in the incubation period, and some differences in the nature of clinical presentation and duration of illness, for different sheep and goat scrapie strains and sources of TSEs [30–34]. Pruritus and ataxia are often described as features of UK sheep scrapie; yet these signs are often absent, and some sheep with scrapie lesions characteristic of advanced disease may apparently die without any overt clinical signs . When neurological signs are present, locomotor deficits occur for a relatively short interval late in the incubation period, but other more subtle signs, such as abnormalities of cardiac rhythm and rumination rates [36,37], and weight loss, may occur relatively early. Behavioural changes may allow observant farmers to identify affected sheep months before profound neurological deficits are evident . Some suggest that affected sheep have a preference for concentrate food rather than hay, spend longer time in recumbency and have reduced social interactions with other sheep when compared with controls . However, these changes are subtle, difficult to monitor under field conditions, and have not been formally characterized for other than end-point stages of disease.
Relationship between morphological changes and clinical disease
Because of the imprecise onset and vague nature of many of the early clinical signs, establishing relationships between clinical deficits and pathological changes is difficult in large animals. While good data are available on the onset and progression of human prion disease, there is little opportunity to perform systematic intervention to correlate with morphological changes. Several studies which attempt to correlate behavioural and morphological studies have now been undertaken for rodent scrapie. Following ME7 infection of mice, affective, locomotor behaviours and cognitive functions were altered and the earliest behavioural changes were identified from 55% of incubation period, which coincides with loss of synaptic terminal from the hippocampus .
It is unlikely that any of the classical triad of histological lesions, neuronal vacuolation, astrocytosis and neuronal loss are primary causes of clinical disease. It has long been recognized that vacuolation is not always a prominent feature of the pathology of scrapie-affected sheep and mice . Vacuolation may be absent from some experimental strains of sheep scrapie, such as SSBP/1 [14,41] or CH1641 . Vacuolation is also virtually absent in some other prion diseases, such as transmissible milk encephalopathy (TME)-infected mink of the Chediak–Higashi genotype  and scrapie infection of Beige mice.
Similarly, in large animals, astrogliosis correlates with the magnitude of PrPd accumulation, and is often subtle or absent in many naturally occurring cases of cattle BSE and scrapie. It is often undetectable in the instances cited above where vacuolation is inconspicuous (pers. obs.). In mice, the astrocytic response is restricted to those brain areas showing vacuolation  and appears to precede it by about 2 weeks. Astrocytosis is often a subtle feature of natural cases of cattle BSE (pers. obs.) and scrapie . In sheep, most gliosis is of an astrocytic origin with a minor microglial fraction, which is mostly confined to white matter (Figure 6). In contrast, microglia form the substantial majority, and astrocytes the minority, of glia in rodent scrapie (Figure 6). Some inflammatory gene products are up-regulated early in rodent scrapie, and transgenic or pharmacological manipulation to decrease expression of such chemokines may increase incubation period. In ME7 rodent scrapie, there is a correlation between microglial activation and microgliosis, which precedes neuronal loss, and the onset of early behavioural changes .When MEe7-infected transgenic mice deficient in a monocyte chemo-attractant protein and wild-type mice are compared, neither the onset of initial clinical signs nor the sequence with which clinical signs appear, nor the timing and sequence of the morphological changes are altered . However, despite this loss of neurones and microgliosis, the overt late-stage clinical signs were delayed and incubation periods increased. This suggests that progression to terminal disease is not dependent on microgliosis or neuronal loss.
Neuronal loss and cortical atrophy are often prominent features of sporadic CJD . However, human beings with shorter-duration vCJD do not always show significant cortical atrophy . In other human prion diseases, there is marked end-point variability in neuronal populations, which tends to correlate with the length of clinical course and severity of microgliosis [48,49]. GABAergic neurones show high levels of the normal PrP protein in mice , but selective alterations to this population of neurones have included both increased and decreased GABAergic neurones [51,52]. While selective losses of particular subclasses of GABAergic neurones in different diseases and model systems [53–55] might partly account for differences between studies, it is difficult at present to find indication that selective loss of GABAergic neurones is a central event leading to clinical disease.
Neuronal loss is present in some food animal prion diseases and is conspicuous in some murine scrapie models . However, decreased numbers of neurones are not evident on routine examination of cattle clinically affected with BSE or most sheep scrapie brains, and even when present, there is no clear anatomical link with neurological deficits. For example, BSE-affected cattle show alterations in mentation, presumably related to cerebral or thalamic alterations, but subjectively determined neuronal loss is not detectable at either of these sites even though extensive PrPd accumulation does occur in the latter. While morphometric calculations of neuronal density indicate that there is loss of neurones within some brainstem nuclei of BSE-affected cattle [57,58], no neuronal losses are present in the dorsal motor nucleus of the vagal nerve, a site in which there is both intense PrPd accumulation and the likely neuroanatomic origins of neurological deficits of decreased rumination rates and heart rate variability [37,59]. Furthermore, alterations in heart rate variability may be detected in experimentally challenged cows from at least 29 months post challenge , when evidence of neurological disease is absent and neither vacuolation nor PrPd accumulation is present in the brainstem [37,61]. Markers for apoptosis which might indicate an underlying tendency to neuronal loss also show no significant alterations in BSE-affected cattle . When present, neuronal losses of rodent scrapie occur relatively late in the incubation period, and usually occur substantially after the onset of behavioural and locomotor changes [39,63].
Other lesions of TSE-infected brains include degenerative changes in axon terminals and in the numbers of synapses [39,55,64]. In murine ME7 scrapie, axon terminal degeneration and decreases in synapses were shown to occur prior to neuronal loss , which may correlate with the onset of behavioural change . However, no significant evidence of axon terminal degenerations and synaptic alterations has been seen in sheep and cattle brains [65,66] examined at clinical end point at sites which neuroanatomically correlate to specific clinical deficits. While axon terminal degeneration may therefore occur in murine models of disease and presage the degeneration of neurones, there is no evidence to suggest that such lesions may underpin fundamental changes leading to neurological deficits in TSE-affected ruminants.
In summary, when compared across species and diseases, it is impossible to find consistent correlations between morphological changes and the cause and onset of clinical deficits.
Relationship between PrPd accumulation, morphological change and clinical disease
If we accept that there is currently no good evidence to link morphological change and disease, is it possible that accumulation of PrPd might affect neuronal function without any morphological changes? There are two paradigms concerning the relationship between PrPd and putative cellular dysfunction. It has been suggested that neuronal dysfunction could be caused by the loss of function of PrPc, possibly caused by conversion of the normal isoform into abnormal forms, or by some other interaction between PrPd and normal PrPc function (the loss of function premise). Alternatively, it has been suggested that PrPd may be directly toxic to cells (the gain of function premise).
Evidence to support the ‘loss of function’ hypothesis arose initially from grafting experiments in which neural tissues from wild-type mice were grafted into brains of mice from which the PrP gene had been deleted (PrP–/–mice). When such mice are intracerebrally inoculated with scrapie, they develop typical scrapie lesions within the graft. Despite the presence of PrPd within the adjacent PrP–/– tissues, no scrapie-like changes were present. The absence of pathology in PrP–/– tissue, which apparently contained PrPd, was interpreted to suggest that PrPc was a necessary prerequisite for development of lesions and disease . In contrast, recent in vitro studies have suggested that the neuronal cytoplasmic accumulation of a weakly protease-resistant form of PrPd may be highly cytotoxic in certain neuronal cells, and an important mediator of pathogenesis in TSE diseases [68,69]. There are other studies which support, and yet others which conflict with, each of the above data and the associated neurodegeneration paradigms. For example, mice with PrP–/– neurones developed scrapie lesions in vivo when exposed to astrocyte-derived PrPd, suggesting that PrPd can be directly toxic to PrP–/– cells , and other data show that cytoplasmic expression of mutant or wild-type PrP in human primary neurones in vitro is not cytotoxic . Whether the direct toxicity or loss of function paradigms is sufficient to explain clinical disease remains unproven.
Intracellular and cell-membrane or extracellular forms of PrPd
Animals and human with terminal disease often have a wider distribution of PrPd than histopathological lesions [72,73]. Time-course studies show that PrPd accumulates prior to vacuolation [61,64], suggesting that the level or durations of exposure may be critical to the development of lesions. However, clinical disease is occasionally described in various species without any, or only minimal, PrPd accumulations [74–78]. Heterogeneity of molecular forms that can be labelled as PrPd, and duration of exposure might be important for the development of lesions. Some immunochemical studies  and epitope mapping indicate that PrPd is present in a wide range of molecular forms; hypothetically, only some of these, or others that are not detected by existing methods, might be toxic.
Although extracellular fibrillar forms of PrPd accumulation can be visibly correlated with tissue changes in the electron microscope, neither the fibrillar forms of PrPd accumulation, nor the associated morphological changes, can be found at all brain sites or in all disease situations. In particular, visibly aggregated or fibrillar PrPd accumulations which are associated with morphological changes in mice, are usually absent in sheep [65,66] and cattle (pers. obs.), and cannot in these species be the cause of the majority of neurodegeneration and clinical deficits in scrapie or BSE. The peptide fragments used in most of the in vitro studies which demonstrate neuronal toxicity, are not found in vivo.
Some sheep affected with natural scrapie produce abundant cell-membrane and extracellular PrPd at some specific neuroanatomic sites, while other sites produce predominantly intracellular PrPd accumulation  (Figure 5). Several observations point to a relationship between cell-membrane and extracellular PrPd and vacuolation. First, in the brainstem olivary nuclei of many natural scrapie-infected sheep, vacuolation is sparse or absent. At this site, most PrPd is present as intraneuronal accumulations, irrespective of the infecting source or strain. In contrast, in the dorsal motor nucleus of the vagal nerve, where most PrPd may be cell membrane associated, marked vacuolar changes are found. Second, the experimental sheep scrapie strains SSBP/1 and CH1641 lack significant vacuolation, and almost all the PrPd which accumulates in both of these strains is found within neurones and glial cells . Although these examples suggest a direct relationship between the presence and abundance of extracellular PrPd and vacuolation, the absence of vacuolation in some scrapie strains and sources indicates that vacuolation per se is not sufficient for clinical disease.
While cell-membrane and extracellular PrPd may be associated with some neurodegenerative changes, principally vacuolation and gliosis, there is little morphological evidence that intracellular PrPd is associated with significant changes. Some intracellular PrPd is associated with cell death, although this appears to happen only to a minority of cells in only some strain host combinations, such as in the lenticular nuclei of BSE-affected sheep , and the numbers of lysosomes are increased in the perikarya of otherwise healthy PrPd containing neurones (Jeffrey et al. submitted).
Although there are some differences in clinical presentation, different sheep scrapie strains and sources show some broadly similar clinical signs. Yet, as stated above, some sheep strains produce only intracellular forms of PrPd. Furthermore, intraneuronal PrPd accumulation does not occur in TG3Prnp–/– mice, which do not express PrP or accumulate PrPd in neurones; yet these mice develop typical rodent TSE pathology and clinical disease . Similarly, although the spectrum of clinical TSEs includes behavioural and mental changes which relate to higher brain functions, some clinical cattle BSE cases and naturally occurring cases of scrapie in the UK, particularly those of the VRQ/VRQ genotype, do not show significant PrPd accumulations rostral to the hindbrain and may show none in thalamus and cerebrum. Therefore, it is difficult to make any association between the intracellular localization of PrPd and clinical disease.
When semiquantitative immunohistochemical PrPd profiling methods are used to determine the total amounts of PrPd throughout the whole brain, marked magnitude variations are found for different sources or strains of scrapie , which cannot be explained by differences in the gross neuroanatomic targeting of those agents. Similarly, when incubation periods of sheep of different genotypes, and challenged by different routes, are correlated with the amount of PrPd in the brain, all possible combinations of PrPd abundance and incubation period occur  (Figure 7). As the same cells and neuronal populations are targeted in these different situations, the absence of correlation between amount of PrPd and incubation period argues against a clear relationship between the involvement of PrPd in the onset of clinical disease. Thus, when the amounts of PrPd in the brain are scored by subjective or semiquantitative methods, neither the distribution nor the magnitude of PrPd correlates with the onset or the duration of clinical disease.
The discrepancies between PrPd accumulation and clinical disease in animals may have many different explanations. First, different scrapie strains may result in the accumulation of different forms of PrPd, which may induce neuronal dysfunction by different mechanisms; that is, there may be different toxic molecules or mechanisms with different strains of agent or different host species. If fundamental pathogenetic mechanisms in TSEs are altered by species or strain or genotype, then many of the general associations and comparisons we describe above would break down. Second, the ‘toxic’ PrP molecule may be present in minute amounts, or be a species of PrPd molecule which may be below the level of detection by current immunohistochemical or immunoblotting techniques. Given the broadly similar morphological changes and neurological signs of many TSEs, it is more likely that the basic pathogenic mechanisms involved and the basis of neurological dysfunction are common to all species, strains and genotypes. In our opinion, it is unlikely that molecules which accumulate over prolonged preclinical and clinical phases of disease remain both below the threshold of detection and yet are present continuously and in sufficient amounts to cause toxic change and neurodegeneration and neurological deficits. A third possibility therefore needs to be formally considered: that the cause(s) of neurological deficits are not fundamentally linked to the accumulation of PrPd, and that PrPd is simply a by-product of the infectious process. While PrPd clearly does appear to be temporally and spatially related to some morphological tissue changes (vacuolation, dendritic damage, abnormal endocytotic mechanisms), and may induce degenerative changes (neuronal loss) which exacerbate neurological dysfunction once clinical signs are evident, the primary pathogenetic events leading to CNS degeneration and neurological dysfunction in our view still remain to be established. Greater consideration needs to be given to determining the physiological and/or neuropathological changes that cause clinical disease.
Pathological changes outside the CNS
Peripheral nervous system
Infectivity and/or PrPd accumulations are found in peripheral nerves and associated ganglia, particularly in the ENS, and in patterns or associations suggestive of nerve terminals in a number of other organs, such as the tongue and adrenal gland. PrPd is infrequently demonstrated in peripheral nerve trunks [81,82] in sheep TSEs or indeed in other prion diseases, although transmission studies more often demonstrate the presence of infectivity. Some rodent models have previously shown that infectivity may be transported via nerves to the CNS from inoculation sites  and from the infected CNS to peripheral sites through nerves . Given the evidence for transportation of infectivity via nerves, the low frequency of detection of intra-axonal PrPd could be a reflection of the sensitivity of detection methods, but more probably is connected to the amount of PrPd in nerves and the weak relationship between infectivity and PrPd.
In scrapie-affected sheep, peripheral nerves and ganglia frequently accumulate PrPd (Figure 8) and, where tested, infectivity [15,85]. Accumulations of PrPd occur as intracytoplasmic accumulations within both neurones and satellite cells. In contrast to the CNS, where routine immunohistochemical methods detect little PrPc, ENS neurones of scrapie-naïve sheep and mice are strongly labelled by several anti-PrP antibodies , suggesting that these neurones differ from the bulk of those of the CNS either in the predominantly cytoplasmic distribution of PrPc or in the expression levels of the protein.
In most cases of natural sheep TSE, there is widespread PrPd accumulation in the PNS at the terminal stages of disease, so that cranial nerves, and spinal and autonomic ganglia, are all affected. The ENS may be one of the first PNS sites to accumulate PrPd, at least in highly susceptible sheep exposed to high doses of infectivity . PrPd-positive ENS neurones are often most abundant next to PrPd-positive Peyer's patches. Such data have prompted some to suggest that neuroinvasion may take place following infection of ENS after amplification of infectivity in the Peyer's patches of the gut. However, there are a number of problems with the suggestion that infection of the LRS is required before the ENS becomes infected. Previously, demonstrations of nerves in lymphoid tissues have been limited to descriptions of sympathetic fibres surrounding blood vessels, but we have recently shown that nerve terminals may be detected in the Peyer's patches of scrapie-affected sheep, including some in direct physical contact with follicular dendritic cells (FDCs) and TBMs . Such nerves are not present in all follicles or in all scrapie-infected individuals, but are never detected in secondary follicles of lymph nodes or in control sheep (pers. obs.). Thus the presence of nerves in LRS tissues, including that of the Peyer's patches, would not appear to occur with sufficient frequency to supply infectivity to PNS neurones in every case. Furthermore, PrPd accumulation within follicles does not occur in all cases of natural scrapie, particularly in VRQ/ARR sheep, and therefore, hypothetical ENS infection following amplification of PrPd in FDCs cannot be a universal mechanism of neuroinvasion.
ENS neurones are extensively networked throughout the intestine, and synaptic nerve endings extend well into the villi and can readily be found immediately beneath the basement membrane of enterocytes . In a study of the transportation of scrapie brain homogenate across the alimentary mucosa of sheep, we showed that PrPd passed from the lumen across or between enterocytes to reach the lamina propria . It is thus possible that infectivity may directly contact ENS terminals without prior amplification of infectivity in the LRS. In the same experiment, we also showed that PrPd entered the lymphatic system and, within minutes, could be identified in afferent and efferent lymphatic channels of draining lymph nodes. Thus, the possibility that neuroinvasion may occur following lymphatic and haematogenous dissemination should not be dismissed.
At clinical stages of disease, infectivity and PrPd accumulations are generally found throughout all lymphoid tissues with the exception of the thymus. PrPd accumulations in lymphoid tissues are mainly found in association with mature FDCs of the light zone of secondary follicles. TBMs also label for PrPd, and may be found in both the light and dark zones of follicles and, also, in a density which decreases with increasing distance, in paracortical and epithelial structures adjacent to such follicles. As in mice , PrPd accumulation in the LRS of sheep is associated with the plasma membrane of FDCs and the lysosomes of TBMs (McGovern et al. submitted).
In some studies of naturally or experimentally infected sheep at preclinical stages, the first tissues to accumulate PrPd are those of the Peyer's patches or mesenteric lymph nodes [90,91], but other animals may show initial or simultaneous accumulation in tonsil or retropharyngeal nodes [91–93]. In all serial kill experiments, the time difference between the first LRS tissues becoming positive for PrPd and all LRS tissues becoming positive is short. Similarly, the recto–anal mucosa-associated lymphoid tissues become positive for PrPd in similar proportions of scrapie sheep and, at the same time, as in other LRS tissues of sheep of the same genotype, irrespective of the route of inoculation [94,95]. These data suggest that all LRS tissues are exposed to infectivity at more or less the same time, presumably via blood exposure.
Using IHC, a simple temporal progression of PrPd accumulation can be recognized within follicles: the earliest accumulations begin as single puncta within TBMs of the light zone, followed by a weak curvilinear accumulation associated with FDCs, and subsequently granular PrPd accumulation in TBMs of the dark zone. Later stages of presumptive secondary follicle infection include diffuse involvement of all FDCs process networks in the light zone, and prominent multigranular accumulation in TBMs of light and dark zones and surrounding tissues (Figure 3). In addition to the generalized infection of all LRS tissues, most, or all, secondary follicles of each lymph node and Peyer's patches, as well as acquired lymphoid nodules such as those associated with chronic parasitic or viral infections [93,96], show PrPd accumulation. In one systematic study of the rectal lymphoid tissue, 84% of clinically affected sheep had at least 70% of secondary follicles with PrPd accumulation, while 35% and 39% of preclinically and clinically affected sheep respectively showed 100% involvement of secondary follicles . Lymphoid follicles of the alimentary tract normally undergo an age-related involution, but this effect is negligible in scrapie-affected sheep , suggesting that the physiological involution of alimentary tract lymphoid tissues is inhibited or decreased due to PrPd accumulation in the gut-associated lymphoid tissue.
Although the onset of PrPd accumulation in the LRS occurs significantly before clinical disease, the presence or absence of PrPd in the LRS, and the precise time at which this accumulation starts, varies according to genotype, route of inoculation, dose and strain. As previously described in mice, sheep intracerebrally inoculated show fewer PrPd-positive LRS tissues, mainly those of the head and neck , while those challenged by intravenous routes have heavier involvement of spleen than do oral-challenged animals . A proportion of naturally infected VRQ/ARR sheep [90,91,94,98] and, to a lesser extent, of VRQ/ARQ and ARQ/ARQ sheep [92,94], and most experimentally challenged ARR/ARR sheep, lack evidence of infection or PrPd accumulation in LRS tissues. It is likely that the ARR allele, especially when present in the homozygous state, confers resistance to infection or amplification of infectivity in peripheral tissues which can be partially overcome by large doses of infectivity. Cattle naturally infected with BSE also lack PrPd accumulation in peripheral tissues, but do have limited evidence of replication in Peyer's patches following massive 100 g oral challenge.
Two separate studies show early amplification of infectivity in the LRS at 2–3 months of age in sheep, which subsequently died at around 24 months of age (approximately 8–12% incubation period [90,91]. These studies involved heavily infected flocks where the scrapie infection mainly targeted the VRQ allele. However, in other flocks where infection was not so widespread and also targeted the VRQ allele, PrPd accumulation at such an early stages was not found . In other heavily infected sheep flocks and experimental BSE, where infection targets mainly the ARQ allele, PrPd accumulations in lymphoid tissues are not evident until at least 8–12 months of age (approximately 33–40% incubation period) [92,93,95]. Therefore, the dynamics of PrPd accumulation in the LRS and the CNS differ, depending on the PrP genotype of the host, and also perhaps on the agent strain and the prevalence (or dose) of infection.
The accumulation of PrPd in lymphoid tissues at relatively early stages of infection was initially exploited for diagnostic and antemortem tests using biopsy of the palatine tonsil . More recent approaches have successfully used biopsy of the more accessible lymphoid tissues of the rectum for antemortem diagnosis, which have now been developed as potential rapid screening methods [94,95]. These and other results have provided the basis for screening of surgical tissues for incidence of vCJD in the UK.
In addition to infectivity and PrPd accumulation in the CNS, PNS and LRS, a wider distribution of PrPd and infectivity has more recently been recognized in other sheep tissues (Figure 8). Many of these sites of scrapie agent replication may have relevance for environmental contamination or for human food safety.
Environmental contamination and transmission of infection
Scrapie is both infectious and contagious. For many years, it has been recognized that classical sheep scrapie spreads along family lines in infected sheep flocks . Early studies suggesting that in-utero or trans-ovarian infection occurs in sheep contained flaws of design and implementation, and it is most likely that there is no transmission of infectivity prior to parturition . The long-recognized transmission of scrapie down family lines probably relates to disease susceptibility through Mendelian inheritance of susceptible PrP genes. With the exception of two instances of vaccine contamination with scrapie [101,102], it is widely assumed that horizontal transmission occurs via the oral route. A number of epidemiological studies have shown that farms which do not dispose of placenta by incineration or burial, have a higher risk of scrapie in their flocks . Unlike cattle, which calve throughout the year, commercial lambing is restricted to relatively short periods of the late winter and spring. Mathematical modelling can therefore help identify times of increased exposure to infectivity, and evidence is now available to show that there is an increased risk of scrapie transmission at lambing time  and the timing of disease is most consistent with exposure to placental tissue. The duration of exposure after lambing also affects transmission rates to offspring. Those lambs which are removed at birth from their scrapie-infected dams and flock had a lower incidence (10%) of scrapie in adulthood than those separated later (16%) or those not segregated at all (41%) .
In addition to the epidemiological evidence of risk and exposure, transmission , immunohistochemical  and immunoblotting studies  have confirmed the presence of infectivity and PrPd in placenta of scrapie-infected sheep. By IHC, PrPd appears to accumulate mainly in the foetal trophoblast (Figure 8), although some maternal epithelial cells, which form syncytia with foetal trophoblasts, also accumulate PrPd. Despite this cellular location of PrPd, there remain no confirmed cases of in-utero transmission of infection. Although placental structures differ, there is also still no evidence of in-utero transmission for cattle with BSE or following experimental infection of simians, which, in common with man, possess haemochorial placentation .
Whether PrPd is present in placenta or not, depends on both the infection status of the dam and the genotype of the foetus. Only the placentas of susceptible genotype foetuses of scrapie-infected dams will accumulate PrPd. Where twins of differing susceptibilities are present in different horns of the uterus, only the placenta of the horn containing the susceptible foetus will accumulate PrPd. Although there is an increased probability of lambs of infected ewes developing disease, infectivity of the dam does not guarantee infectivity in susceptible progeny , as the stage of infection, and particularly whether systemic infection is present, will affect outcome.
It is likely that horizontal transmission between sheep, other than during the lambing period, contributes to the spread of disease within and between flocks. A study performed before the effects of PrP gene susceptibility were understood monitored the outcome of lambs exposed to scrapie-infected sheep for different time intervals. The longer the period of contact between lambs and infected animals, the higher was the incidence of disease in the exposed sheep . A similar experiment was performed more recently introducing adult scrapie-naïve sheep of susceptible genotypes to scrapie-affected sheep, and reached a similar conclusion . It is possible that age of exposure also influences susceptibility .
In addition to placenta, there are a number of other tissues which contain infectivity, or in which PrPd has been demonstrated, and which might potentially contribute to contamination of the environment outside the lambing period. PrPd is present in macrophages of the rectal mucosa , potentially causing contamination of faeces (Figure 8). Accumulations of PrPd in the inter-stitium of the renal papillae  may potentially result in infectivity reaching urine (Figure 8). Similarly, PrPd accumulation in the tongue may result in infectivity of saliva , as demonstrated for chronic wasting disease (CWD)-infected deer. PrPd in bronchial-associated lymphoid tissues might result in aerosolized spread of infectivity (Figure 8) and, in periductular lymphoid tissues of the mammary gland, might lead to contamination of milk.
PrPd is also present in the retina, where it mainly accumulates in the plexiform layers [95,115], and in a pattern suggestive of a nerve association in the medulla of the adrenal gland (Figure 8). These last tissues, while not relevant for transmission in the sheep, are relevant to pathogenesis studies.
Health and food safety
Although a number of epidemiological studies have failed to support a link between the human prion diseases and scrapie, there remains a theoretical possibility of BSE being masked within the sheep scrapie cases. As such cases would potentially represent a risk to human, precautions are in place at the abattoir to prevent tissues containing high levels of infectivity, principally those of the CNS, reaching the human food chain.
As described above, recent studies have shown that PrPd may be found in association with sensory nerves of the tongue . Similar observations in experimental rodent models have prompted some to postulate that such a site might be a natural route of neuroinvasion . However, the hypoglossal nucleus or sites connected with taste perception are not among those to first accumulate PrPd in sheep scrapie (or in cattle BSE), and it is much more likely that PrPd at sites such as the tongue is the result of centrifugal spread of infectivity and accumulation of PrPd at terminal stages of disease and, therefore, probably does not pose a risk for human health while adequate surveillance systems are in place.
There is emerging evidence from transfusion studies that infectivity can be present in blood of scrapie and BSE-infected sheep from early stages of infection [27,117]. The demonstration of infectivity in blood has little relevance for inter-sheep transmission, but in the case of vCJD, it has clear implications for human health issues because blood permeates all tissues of the sheep used for food.
Striated muscles have traditionally been viewed as having negligible levels of infectivity, although more recent studies have shown that muscles may contain infectivity or PrPd in TSEs of rodents, man, deer and sheep. Several different laboratory methods have now confirmed detection of PrPd in a wide range of striated muscles of sheep naturally and experimentally infected with scrapie, mostly in the nuclear chain and nuclear bag fibres of muscle spindles (Figure 8) . PrPd-positive muscle spindles were more frequent in experimentally challenged sheep than in naturally exposed flocks , a finding that agrees with our own observations. The onset of muscle PrPd accumulation detection in one naturally exposed flock was achieved at approximately 61% of the mean incubation period. The levels of infectivity in the muscle are estimated at 1/5000 fold less than in the brain . Thus, even if BSE were to be present in sheep, the low titre, late onset of accumulation and infrequent occurrence in natural disease decrease the probably that muscle may constitute a significant risk for human.
We are grateful to Jim Hope for critical comments on the manuscript.