It is currently believed that primary transmission of classical scrapie to wild-type mice is inefficient and characterized by low attack rates and variable incubation periods and lesion profiles. Consequently, strain characterization of classical scrapie in these mice relies on subpassage. The aim of this study was to perform a retrospective analysis of lesion profiles and immunohistochemistry patterns after transmission of a large number of classical scrapie sources to wild-type mice and to investigate trends that might be used to characterize the agent without subpassaging. Scrapie field cases (n = 31) collected from individual farms between 1996 and 1999 were inoculated into RIII, C57BL, and VM mice and profiled using standard methodology and analyzed by immunohistochemistry. Using cluster analysis to resultant lesion profiles produced groups of similar lesion profiles in RIII and C57BL mice. We observed correlations between lesion profile clusters and the ovine prion protein (PrP) genotype. Immunohistochemistry indicated donor-mediated trends in the PrPSc pattern. These results indicate that ovine PrP genotype is a factor that is linked to both the lesion profile and the pattern of PrPSc deposition on primary transmission of classical scrapie to wild-type mice.
Transmissible spongiform encephalopathies (TSEs) are fatal neurodegenerative diseases that affect animals and humans. Misfolding of normal cellular prion protein (PrPC), a protein of predominantly α-helical structure, to a pathogenic and protease-resistant isoform, PrPSc, a protein of predominantly β-sheet structure, is considered to be a ubiquitous marker of TSE (1). This family of diseases manifests in diseased individuals with central nervous system spongiform change (vacuolar degeneration) deposition of PrPSc in various forms, including plaque structures, and neuronal loss and gliosis.
Classical scrapie is a highly characterized TSE of small ruminants that has been endemic in the United Kingdom for centuries (2). Although scrapie is not transmissible to humans, because sheep succumb to experimental infection with bovine spongiform encephalopathy (BSE) (3, 4), concerted efforts have been made to eradicate it from the national flock. The sheep PrP gene, PRNP, is highly polymorphic within the open reading frame (5). In particular, 3 polymorphisms, at codons 136 (alanine [A] or valine [V]), 154 (arginine [R] or histidine [H]), and 171 (glutamine [Q] or arginine [R]), have been linked to susceptibility to classical scrapie in that the presence of valine at codon 136 in one or both alleles confers the highest susceptibility (6-9).
The mouse bioassay remains the gold standard model for assessing the biologic properties of prions and the discrimination of TSE strains. Prion strains are defined as infectious prion isolates that show distinct phenotypes, such as incubation times and histopathological lesion profiles, which persist on serial transmission (10). Primary transmission of a TSE to wild-type (WT) mice is marked by low attack rates, prolonged and variable incubation periods, and arguably uncharacteristic lesion profiles. For the purpose of strain characterization, therefore, isolates are passaged serially through mice until the incubation period stabilizes before it is fully characterized (11). The RIII and C57BL mouse lines (Prnpa) share the same Prnp amino acid sequence, whereas VM mice (Prnpb) differ at codons 108 and 189 (12). This difference in genotype affects both the incubation period and lesion profile of specific isolates (13). Strains can be distinguished based on incubation period and the pattern of TSE-specific vacuolation after inoculation via a consistent route. Moreover, semiquantification of vacuolation in specific neuroanatomical areas permits the construction of lesion profiles based on average scores for vacuolation in response to a given inoculum in a panel of mice (14). Lesion profiles have historically provided the most informative method of strain characterization through the bioassay. Because different strains may give rise to unique patterns and types of PrPSc deposition in the mouse brain, attention has more recently turned to the use of immunohistochemistry (IHC) for identification and characterization of strains. Early indications suggest that this method could offer greater sensitivity and the capability of identifying agent strains at the level of individual mice compared with lesion profiling, which requires an average profile plotted from at least 5 clinically and pathologically positive mice because of variation in individual mouse profiles (15).
Previous studies indicate that subpassage is a prerequisite for strain characterization of classical scrapie (16, 17), primary isolation data arising from ovine scrapie sources are predominantly considered only for discrimination from BSE, for which RIII mice play a crucial role (16, 18); recently, however, this view has been challenged (19). The aim of the present study was to investigate lesion profiles and IHC patterns after primary transmission of classical scrapie sources of various ovine PrP genotypes to RIII, C57BL, and VM mice. The results presented here are a retrospective analysis of the primary isolation data of a large number of natural scrapie sources.
Materials and Methods
The animal selection procedure is described in detail elsewhere (19). In brief, between 1996 and 1999, clinical scrapie suspects from commercial farms throughout England and Wales were transported live to a Veterinary Laboratories Agency regional laboratory. The principle of selection was for each farm to contribute either a single case or more than 1 case, but of different PrP genotypes, in an effort to include as many different farms as possible and to avoid over-representation of farms with a high incidence of scrapie. After death, the medulla at the level of the obex was fixed for definitive diagnosis, and the remaining brain was frozen for biochemical analysis and bioassay studies. Clinical cases were confirmed positive for classical scrapie by statutory tests, including histopathology, IHC, and Western blot. The biochemical properties of each isolate were also analyzed in detail to exclude the possibility of BSE (19). Each case was also blood sampled for PrP genotyping.
Mouse Inoculation and Sample Preparation
All selected cases were inoculated into groups of RIII, C57BL, and VM mice; each panel consisted of 20 mice aged 6 to 10 weeks. Inoculations were made during a period of 11.5 months. Each mouse was inoculated intracerebrally (20 μL) and intraperitoneally (100 μL) with 10% (wt/vol) medulla homogenate prepared in sterile isotonic sodium chloride solution. Mice were monitored from 250 days postinoculation for clinical signs of TSE infection, including weight loss, incontinence, abnormal gait, vacant stare, and a rough coat. Mice were killed using carbon dioxide when a predetermined clinical end point had been reached, defined as having received positive clinical scores of TSE in 2 consecutive weeks, or having received scores of “definitely affected” in 2 of 3 consecutive weeks. Mice were also killed if there was significant deterioration, lack of mobility, or inability to eat or drink at any time. These mice may also have exhibited clinical signs of TSE but not consistently enough to be scored as clinically positive. Therefore, they likely had been killed at an early clinical stage. All work was carried out in accordance with the Animals (Scientific Procedures) Act 1986. Brains were removed and fixed in 10% neutral buffered formalin for at least 3 days at room temperature before being cut into 5 coronal levels to reveal caudal medulla, rostral medulla, midbrain, thalamic, and frontal levels. Tissues were processed and embedded in paraffin using routine histological methods. All 5 coronal levels were embedded in the same block, and 3-μm-thick sections were subsequently mounted on slides.
Histopathologic and Immunohistochemical Analyses
Sections were stained with hematoxylin and eosin, as previously detailed (18). Postmortem TSE diagnosis was confirmed based on the presence of characteristic neuropil vacuolation. Lesion severity in positive samples was further quantified as previously described (14, 19). In brief, scores were assigned, on a scale of 0 to 5, for specific neuroanatomical gray matter areas of the brain. Mean scores for each brain area were calculated for each inoculum and then plotted to produce lesion profiles. Because of variations between mice, profiles were only constructed for inocula that included 5 or more mice with positive clinical scores, in accordance with established methodology (20).
It is our experience that a mouse found to be pathologically positive will contain sufficient PrPSc to analyze the deposition pattern irrespective of its clinical status. Therefore, to include the full spectrum of IHC variability from each inoculum, we examined 5 slides representing the minimum, maximum, 25th, 50th, and 75th percentiles for incubation period based on pathologically positive mice. Samples were labeled with rabbit polyclonal antibody Rb486, which recognizes amino acids 221 to 233 of the bovine PrP, using a standard method (21). A description of deposition type was limited to general neuropil deposition (GND), where at low magnification (10×), no specific formations were observed or characterized as plaques or aggregates. We used a qualitative method whereby the presence of PrPSc types in neuroanatomical areas was recorded onto brain maps, as described elsewhere (22).
Readers were blind to sample source when lesion scores were assessed. On the basis of lesion scores, the isolates were then separated by cluster analysis that enabled objective identification of clusters. We then characterized the clusters by comparing lesion profile and PrPSc IHC deposition pattern. Lesion profiles for each isolate were plotted and grouped according to their assigned cluster. The IHC slides were sorted by isolate and by cluster, and subsequent deposition patterns were recorded to identify the predominant patterns associated with each cluster and the mice that presented alternative patterns ("outliers") within each isolate. All results were subsequently considered in relation to the genotypes of the ovine sources.
Average lesion profiles produced from different TSE isolates were compared using cluster analysis (Statistica 8.0 statistical software, StatSoft, Inc, Tulsa, OK). This approach identifies structure in data by identifying natural groupings (clusters). A cluster is a collection of data (lesion profiles) that are more similar to each other than they are to data in other clusters, as defined by a designated list of characteristics or indicators. Here, the indicators were the mean vacuolation scores for each of the 9 neuroanatomical gray matter areas scored (23). Linkage distance denotes the distance at which respective isolates were linked to form a new cluster. We used this method previously to compare the outcome of cluster analysis with manual separation of lesion profiles and showed that they were well correlated (19). Results were visualized as a dendrogram.
Incubation Period Analysis
Mean incubation periods were calculated for each inoculum including only clinically and pathologically positive mice and then plotted as box blots.
Cluster Analysis Gives Rise to Distinct Groups of Lesion Profiles on Primary Isolation of Classical Scrapie
Lesion profiling was applied to 31 inocula where 5 or more mice had been diagnosed as positive for TSE based on clinical signs and histology (Table 1). None of these cases originated from the same flock. They included 30 inocula for RIII mice and 17 for C57BL mice, of which 16 gave profiles in both RIII and C57BL mice. The remaining inocula transmitted adequately in either the RIII or the C57BL line. Only 3 of the inocula caused clinical disease in 5 or more VM mice. Data from 31 transmissions that did not satisfy the selection criteria of the study are presented in Table, Supplemental Digital Content 1, http://www.jnen.oxfordjournals.org/lookup/suppl/doi:10.1097/NEN.0b013e3181db2497/-/DC1. To establish whether profiles arising from primary isolation formed common groups, cluster analysis was used for lesion scores from RIII and C57BL mice. Data from the VM mice were not included in the cluster analysis.
For both mouse lines, distinct clusters of lesion profiles arose on primary isolation (Fig. 1). There were 4 clusters each for RIII and C57BL mice, termed R1-R4, C1-C3, and C5 for RIII and C57BL mouse lines, respectively.
Lesion profiles arising from primary isolation of classical scrapie form distinct clusters. Tree diagrams after cluster analysis of the vacuolation scores in selected neuroanatomical brain areas demonstrate distinct groups, as measured by Euclidean distance for RIII (A) and C57BL (B) mice.
The allocation of inocula in RIII-derived clusters corresponded to their distribution in C57BL-derived clusters, although there were fewer inocula with adequate data to produce lesion profiles through the C57BL mouse line. All inocula in Cluster C1 were included in Cluster R1, all but 1 inoculum in Cluster C2 were also integrated together in Cluster R2, and a single inoculum was the only member of Clusters R3 and C3. With the exception of a single inoculum, Cluster R4 consisted of sources that did not transmit efficiently to C57BL mice, so there was no corresponding C cluster. Cluster C5 consisted of a single inoculum that did not transmit efficiently to RIII mice, so there was no corresponding R cluster.
Cluster analysis of different scrapie sources based on their lesion profiles in RIII or C57BL mice arranged the sources into groups where a predominant PrP genotype could be correlated. This association between the different clusters and PrP genotype was absolute for Clusters R2, R3, C2, and C3, which were composed solely of ARQ/ARQ scrapie sources. The remaining clusters, R1, R4, C1, and C5, consisted mainly of inocula that were either homozygous or heterozygous for the VRQ allele, suggesting that the presence of valine at codon 136 may have a dominant effect on the underlying pathology of the disease on primary transmission in WT mice. There were 3 ARQ/ARQ sources that were exceptional because they were allocated in clusters associated with the presence of the VRQ allele (Fig. 1).
Lesion profiles for RIII, C57BL, and VM mice were plotted according to their assigned clusters in Figures 2-4, respectively. Profiles for R2 and R3 were markedly similar. This data subset was the focus of a previous study because the shape of this profile resembles that of BSE in RIII mice (19). Profiles for R1 and R4 were markedly different from Clusters R2 and R3 and also differed between each other, primarily in lesion intensity in G6 and G7 (hippocampus and septal nuclei, respectively) (Fig. 2). Similarly, the C2 profile shared a degree of similarity with the C3 cluster, and both were distinctly different from the profiles of C1 and C5 (Fig. 3). The general trend of the profile of C3 shares greater similarities with C2 than with C1 or C5. In VM mice, 2 sources generated profiles of similar shape but of different lesion intensity; the third source generated an alternate profile (Fig. 4). All of the VM sources derived from ARQ/ARQ sheep.
Average lesion profiles in RIII mice grouped according to the clusters determined by cluster analysis. Profiles were obtained after quantification of specific vacuolation in 9 neuroanatomical gray matter areas as follows: G1, dorsal medulla nuclei; G2, folia of cerebellar cortex including the granular layer adjacent to the fourth ventricle; G3, cortex of the superior colliculus; G4, hypothalamus; G5, thalamus; G6, hippocampus; G7, septal nuclei of the paraterminal body; G8, cerebral cortex (at the level of G4 and G5); G9, cerebral cortex (at the level of G7). Solid lines indicate ARQ/ARQ sources, widely dashed lines indicate VRQ/VRQ sources, and narrow dashed lines indicate the ARQ/VRQ and ARR/VRQ inocula.
Average lesion profiles in C57BL mice grouped according to the clusters determined by cluster analysis. Profiles were obtained after quantification of specific vacuolation as in Figure 2. Solid lines indicate ARQ/ARQ sources, widely dashed lines indicate VRQ/VRQ sources, and narrow dashed lines indicate the ARQ/VRQ and ARR/VRQ inocula.
Average lesion profiles in VM mice. Profiles were obtained after quantification of specific vacuolation as in Figure 2. Solid lines indicate ARQ/ARQ sources.
Analysis of PrPSc Deposition Pattern
In RIII mice, Cluster R1 which was composed predominantly of VRQ/VRQ ovine sources, gave mild diffuse GND throughout the medulla, with some aggregated GND localized to the cerebellar granular layer. Additional mild GND was mostly seen in the substantia nigra, red nuclei, midbrain raphe, superior colliculus, hypothalamus, lateral thalamus, habenular bodies, dentate gyrus molecular layer, vertical limb of the diagonal band, and septohippocampal nuclei (Fig. 5). Cluster R4 gave 2 distinct deposition patterns. The first, R4/P1, had a particularly widespread GND distribution, with many aggregates of varying size. Notably, hippocampal deposition was targeted to the dentate gyrus polymorph layer and to the pyramidal layer. The PrPSc deposition associated with this pattern was indistinguishable from that associated with the passage of the fully characterized ME7 strain bioassayed at the Veterinary Laboratories Agency in RIII mice. In addition to the hippocampus, deposition was comparable to that of ME7 across the 5 coronal sections analyzed. Small aggregates throughout, including the granular layer of the cerebellum, the cortex, and the septal nuclei, are characteristic features of ME7. The second PrPSc deposition pattern, R4/P2, seemed similar to that shown by R1 but with increased intensity and targeting of additional neuroanatomical areas, namely the interpeduncular nucleus, lateral hypothalamus, caudate putamen, and the internal pyramidal layer of the cortex (Layer V). With the exception of Scrapie 80, all slides from the R4 cluster showed one of these patterns.
The PrPSc deposition patterns show genotype associations in RIII mice. Left panels: brain maps depicting predominant immunohistochemistry (IHC) patterns. From the bottom left in a counterclockwise direction, the maps represent medulla, midbrain, thalamic, and frontal sections; blue shading represents general neuropil deposition; blue circles are aggregated PrPSc; red stars indicate plaques. R4 gave rise to 2 distinct deposition patterns denoted P1 (Pattern 1) and P2 (Pattern 2). Right panels: Representative photographs of characteristic deposition patterns. The R1 shows hippocampal targeting to the molecular layer of the dentate gyrus (MolDG). For R2 and R3, cortex plaques are shown. The R4/P1 shows hippocampal targeting to the polymorph layer of the dentate gyrus (PoDG) and the pyramidal layer (Py) and aggregated PrPSc. The R4/P2 shows hippocampal targeting to the MolDG. Scale bars = (large panels) 500 μm; (insets) 100 μm.
Clusters R2 and R3, which were composed exclusively of ARQ/ARQ sources, gave a highly distinctive deposition pattern characterized by widespread GND including many plaques and aggregates of PrPSc, particularly throughout the hypothalamus and thalamus, habenular bodies, cortex, and caudate putamen. Hippocampal GND was targeted to the molecular layer of the dentate gyrus.
The predominant patterns that arose from C57BL mice clusters are shown in Figure 6. Cluster C1 was characterized by a PrPSc distribution pattern that was predominantly GND with small aggregates of PrPSc. In the medulla, GND was minimal and localized mainly to the locus coeruleus. The GND and aggregates targeted to specific neuroanatomical areas were seen throughout the midbrain, thalamus, and frontal levels. In the hippocampus, GND was targeted to the molecular layer of the dentate gyrus. The cortex contained GND and aggregates at all levels, either targeted to the internal pyramidal layer of the cortex (Layer V) (Fig. 6) or diffused throughout the cortex. Cluster C5 consisted of a single VRQ/VRQ case where 2 slides showed a PrPSc deposition pattern that shared some similarities with Cluster C1 (C5/P1). The second deposition pattern (C5/P2) in this cluster was similar to that of ME7 passaged in C57BL mice (Fig. 7) (ME7 strain passaged through RIII and C57BL mouse lines are indistinguishable [not shown]). This pattern is characterized by widespread GND and aggregates throughout the brain, with characteristic targeting to the polymorph layer of the dentate gyrus and to the pyramidal layer of the hippocampus.
The PrPSc deposition patterns show genotype association in C57BL mice. Left panels: brain maps depicting predominant immunohistochemistry (IHC) deposition patterns are shown, as in Figure 5. C5 (Scrapie 79) animals produced 2 patterns, that is, the first is comparable to C1/P1 (termed C5/P1) and the second is a unique pattern, C5/P2. Right panels: Representative photographs of characteristic deposition patterns. C1 and C5/P1 show hippocampal targeting to the molecular layer of the dentate gyrus (MolDG). For C2, thalamic plaques/MolDG targeting is shown. The C5/P2 shows hippocampal targeting to the polymorph layer of the dentate gyrus (PoDG) and the pyramidal layer (Py). Scales bars = (large panels) 500 μm; (inset) 100 μm.
R4/P1 mice and C5/P2 mice show the same PrPSc pattern as ME7 passaged through mice with Prnpa genotype. Representative photographs are shown for comparison of the cerebellum (left panels) and the hippocampus (right panels) of ME7 in RIII, R4/P1, and C5/P2 mice. Scale bars = 500 μm.
The PrPSc type and distribution in Cluster C2 were similar to that of R2, with numerous plaques and aggregates evident throughout the thalamic nuclei, caudate putamen, and cortex. Additional GND was mostly widespread. In the hippocampus, GND was targeted to the dentate gyrus molecular layer. A double layer of deposition was frequently evident in the cortex affecting Layers I (molecular layer) and V (internal pyramidal layer). Clusters R2 and C2 were exclusively composed of ARQ/ARQ-derived inocula, and Cluster C3 consisted of a single inoculum that was also ARQ/ARQ. The PrPSc deposition pattern on 3 slides was consistent with Cluster C2. The pattern in one of the other slides was similar to Cluster C1, and the other was similar to Cluster C5/P2.
The PrPSc type and distribution were similar in all VM mice. The GND was widespread, but the most distinctive characteristic was multiple plaques and aggregates located across all 5 coronal sections analyzed (not shown).
IHC Identifies Outlier Mice
Whereas brain maps show the predominant pattern of PrPSc associated with each cluster (Figs. 5 and 6), IHC analysis provided a way to identify any outlier mice. There were instances when a small number of inoculated mice did not conform to the predominantly expressed IHC pattern for the associated cluster (Table 2). For example, Scrapie 80, an ARQ/ARQ source, was allocated in Cluster R4 (a VRQ-associated cluster). Immunohistochemistry revealed that although two of the mice showed a PrPSc pattern compatible with Cluster R4 (Fig. 8A), 2 mice shared an IHC pattern compatible with Cluster R2, an ARQ/ARQ cluster (Fig. 8B), and a single mouse showed a combination of R2 and R4/P1 patterns. In C57BL mice, this source was assigned into Cluster C2, which consists of only ARQ/ARQ inocula. Two mice showed C2-compatible patterns and in each of the remaining 3 mice, there was coexistence of C2 and C5/P2 patterns (Table 2). For the remaining 2 ARQ/ARQ inocula (Scrapie 14 and Scrapie 67), which were grouped in VRQ-associated clusters, the PrPSc deposition patterns were also consistent with the patterns produced by VRQ-associated ovine sources.
Different PrPSc patterns within specific inocula. (A,B) Individual mice inoculated with the same source (Scrapie 80) show distinct patterns. A representative of R4/P1 shows characteristic deposition in the polymorph layer of the dentate gyrus (PoDG) and the pyramidal layer of the hippocampus (Py) (A). A representative of R2 shows deposition in the molecular layer of the dentate gyrus (MolDG) and plaques (B). (C) Mouse inoculated with Scrapie 82 that has a mixture of patterns associated with C2 and C5/P2 as indicated by deposition in the MolDG, PolDG, and plaques. Scale bars = 500 μm.
The most consistent IHC distribution patterns were derived from ARQ/ARQ sources, but there were exceptions. C2 gave rise to a predominant IHC pattern as previously described, but 6 of the 55 slides analyzed (split between 3 of the inocula included in this cluster) showed definite features of C5/P2 to varying degrees in conjunction with the C2 pattern (Fig. 8C).
Incubation Period Differences Between Clusters
Mean incubation periods were calculated for each inoculum and compared in the clusters for RIII and C57BL mice. This analysis suggested a general trend in both RIII and C57BL mouse lines, whereby the incubation periods of clusters consisting exclusively of ARQ/ARQ inocula (R2 and C2) tended to be shorter than those associated with the VRQ allele (Fig. 9).
ARQ/ARQ scrapie clusters tended toward lower incubation periods on primary isolation. Mean incubation periods (days) plotted for each cluster as box plots for RIII mice (A) and C57BL mice (B); boxes indicate the maximum, 75th percentile, median, 25th percentile, and minimum incubation periods.
We demonstrate that primary isolation of classical scrapie in both RIII and C57BL mice can produce characteristic lesion profiles that share similar trends in discrete clusters. Our observations do not support earlier reports suggesting that lesion profiles at primary isolation of scrapie are disordered (16,17); however, those reports were based on limited data without any representation of ARQ/ARQ sources.
With 1 exception, inocula that produced lesion profiles in both the C57BL and the RIII mouse lines were grouped in similar clusters. The lesion profiles between RIII and C57BL corresponding clusters were dissimilar, but the inocula were allocated independently in equivalent clusters, suggesting that the factors that affect TSE histopathology within a specific mouse line at primary isolation are influenced by agent- or recipient-specific features or both. It also indicates that different mouse lines inoculated with the same source can produce different corresponding profiles even if they have the same PrP sequence, suggesting that other recipient genetic factors may affect TSE histopathology (24, 25). These factors may even determine proliferation of different strains in specific mouse lines, for example, as indicated by Scrapie 80, which belongs to noncorresponding RIII and C57BL clusters; however, serial subpassage and strain typing are required to substantiate this interpretation.
Our data demonstrate that the clusters contain inocula, which consist of a predominant ovine PrP genotype that is mainly linked to codon 136. This indicates an association between donor PrP genotype and the murine lesion profile and demonstrates that this association can be observed at least on primary isolation. It remains to be seen whether this link persists on subsequent passages, which would indicate for the first time that there is an association between PrP genotype and strain. These experiments are currently ongoing in our laboratory. Inasmuch as PrPSc distribution patterns demonstrated that certain patterns were attributable to specific clusters, PrPSc distribution, lesion profiles, and donor PrP genotype may be interrelated.
Clusters were generally associated with a single principle distribution pattern, although R4, a predominantly VRQ-associated cluster, generated 2 distinct PrPSc deposition patterns. Because IHC permits the observation of patterns at the level of the individual animal, we were able to identify that these 2 patterns occurred both within and between inocula. It is significant that one of these IHC patterns, R4/P1, bore a striking resemblance to the mouse-passaged strain ME7, the features of which were consequently observed in outlier mice from all clusters, with the exception of R3 and C1. Even in clusters associated with a predominant PrPSc pattern, there were specific inocula where in addition to the mice that conformed to the PrPSc pattern associated with the cluster, we identified a number of mice that had alternative patterns (Table 2). Interestingly, these outlier mice emerged from transmission of ARQ/ARQ sources in C57BL mice, whereas the RIII mice were more homogeneous in transmission of phenotype. The reverse trend was observed for VRQ-associated inocula for which pattern homogeneity was constant in C57BL and diverse in RIII mice.
The observation of an ME7-like IHC pattern in distinct clusters of RIII and C57BL mice on primary isolation is previously unreported. Primary isolation in WT mice inherently requires the crossing of a species barrier and it would be premature to claim that we have isolated ME7. However, there is a strong correlation with ME7 based on PrPSc IHC pattern. It is believed that strains usually require a subpassage to adapt to the new host, and therefore the appearance of ME7-specific markers as early as primary isolation suggests that strain adaptation may have progressed further in these mice.
Although a link between lesion profile and genotype was observed, we identified 3 sources (all ARQ/ARQ) that, on primary isolation in RIII mice, clustered with VRQ/VRQ isolates based on lesion profile. One of these sources was characterized by heterogeneity with respect to the PrPSc patterns observed, whereas analysis of PrPSc IHC patterns and incubation period data from the other 2 sources confirmed that they were consistent with VRQ/VRQ sources. These cases are interesting because they infer that there are different parameter(s) that influence the phenotype of the disease and are more dominant than the genotype of the donor, for example, a strain effect. Similarly, a previous study in sheep investigating the influence of PrP genotype on PrPSc IHC patterns at the level of the obex identified a single farm where the genotype of all affected sheep was ARQ/ARQ, but the PrPSc deposition patterns were compatible with those observed in VRQ/VRQ animals (22). The study suggested that the association between IHC pattern and ovine genotype may be predominantly strain-driven but that PrP genotype can influence strain selection. The observation of genotype-associated clusters of lesion profiles in the current study would support such a hypothesis, although a subpassage is required to assess whether the genotype association persists and whether a particular genotype propagates a specific strain.
It is not possible to conclude whether the observed diversity in the PrPSc patterns reflects diversity of strains in the original host (particularly in the cases where more than 2 patterns were identified from a single inoculum) or if it is an effect of the species barrier without serial passages. The coexistence of multiple PrPSc types, as identified by Western blot in individuals with Creutzfeldt-Jakob disease, does indicate that more than 1 TSE strain can coexist in a single host (10). Our analysis showed that the selected samples that were subjected to IHC provided adequate evidence to substantiate our findings. Therefore, we did not consider it essential to examine all pathologically positive mice with IHC.
Attack rates and incubation periods on primary isolation are highly dependent on titer, and as such, we compared the incubation periods of each cluster using box plots, which indicate the spread of the data without inferring any degree of significance. Results showed a general trend whereby ARQ/ARQ ovine sources give shorter incubation periods and higher attack rates than VRQ/VRQ sources (Table 1; Fig. 9). This was apparent for both RIII and C57BL mouse lines. In contrast, attack rates in VM mice in all but 3 inocula were too low to allow selection for lesion profile analysis (Table 1). The attack rates in the 3 mouse lines in this study confirm a previous report on primary isolation of classical scrapie in WT mice (17).
In conclusion, we analyzed primary isolation data in RIII and C57BL mice arising from a large number of classical scrapie field cases and observed that the resultant lesion profiles shared similar shapes and can be grouped into clusters of similar lesion profiles. In addition, we showed that ovine PrP genotype seems to be one of the factors that affect both the murine lesion profile and the PrPSc distribution pattern. These observations are in agreement with previous data based on a small number of samples in which the deposition pattern of PrPSc was related to ovine genotype during primary transmission of classical scrapie into WT and transgenic mice (26). Further analyses, including serial passages and full strain typing, are required to determine whether the trends in the lesion profiles and the PrPSc patterns at primary isolation can be used as reliable indicators of identifying strains. The diversity in the PrPSc patterns on primary isolation that can be observed among mice challenged with a single source may be lost in the lesion profile or during serial passage for strain typing, which is highly selective. Therefore, IHC analysis based on individual mice during primary isolation may provide a powerful method for analyzing different aspects of the agent.
The authors thank their colleagues in histopathology for skilled technical support. The authors also thank Ian Dexter and Karen Hartley and their teams for technical expertise and support. The ME7 strain was kindly provided by Dr Nora Hunter of The Roslin Institute and R(D)SVS, University of Edinburgh.
This work was funded by the UK Department of the Environment, Food and Rural Affairs (Defra), under projects SE1919 and SE1849.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (http://jnen.oxfordjournals.org/).