Discriminating scrapie and bovine spongiform encephalopathy isolates by infrared spectroscopy of pathological prion protein.

For the surveillance of transmissible spongiform encephalopathies (TSEs) in animals and humans, the discrimination of different TSE strains causing scrapie, BSE, or Creutzfeldt-Jakob disease constitutes a substantial challenge. We addressed this problem by Fourier transform-infrared (FT-IR) spectroscopy of pathological prion protein PrP27-30. Different isolates of hamster-adapted scrapie (263K, 22A-H, and ME7-H) and BSE (BSE-H) were passaged in Syrian hamsters. Two of these agents, 22A-H and ME7-H, caused TSEs with indistinguishable clinical symptoms, neuropathological changes, and electrophoretic mobilities and glycosylation patterns of PrP27-30. However, FT-IR spectroscopy revealed that PrP27-30 of all four isolates featured different characteristics in the secondary structure, allowing a clear distinction between the passaged TSE agents. FT-IR analysis showed that phenotypic information is mirrored in beta-sheet and other secondary structure elements of PrP27-30, also in cases where immunobiochemical typing failed to detect structural differences. If the findings of this study hold true for nonexperimental TSEs in animals and humans, FT-IR characterization of PrP27-30 may provide a versatile tool for molecular strain typing without antibodies and without restrictions to specific TSEs or mammalian species.

From the Robert Koch-Institut, ‡P26 and ¶P13, Nordufer 20, 13353 Berlin, Germany For the surveillance of transmissible spongiform encephalopathies (TSEs) in animals and humans, the discrimination of different TSE strains causing scrapie, BSE, or Creutzfeldt-Jakob disease constitutes a substantial challenge. We addressed this problem by Fourier transform-infrared (FT-IR) spectroscopy of pathological prion protein PrP27-30. Different isolates of hamsteradapted scrapie (263K, 22A-H, and ME7-H) and BSE (BSE-H) were passaged in Syrian hamsters. Two of these agents, 22A-H and ME7-H, caused TSEs with indistinguishable clinical symptoms, neuropathological changes, and electrophoretic mobilities and glycosylation patterns of PrP27-30. However, FT-IR spectroscopy revealed that PrP27-30 of all four isolates featured different characteristics in the secondary structure, allowing a clear distinction between the passaged TSE agents. FT-IR analysis showed that phenotypic information is mirrored in ␤-sheet and other secondary structure elements of PrP27-30, also in cases where immunobiochemical typing failed to detect structural differences. If the findings of this study hold true for nonexperimental TSEs in animals and humans, FT-IR characterization of PrP27-30 may provide a versatile tool for molecular strain typing without antibodies and without restrictions to specific TSEs or mammalian species.
Transmissible spongiform encephalopathies (TSEs) 1 such as scrapie in sheep, bovine spongiform encephalopathy (BSE) in cattle, and Creutzfeldt-Jakob disease (CJD) in humans are invariably fatal neurodegenerative disorders of the central nervous system.
After the initial reports on the emergence of BSE and variant Creutzfeldt-Jakob disease, in 1986 and 1996, respectively, compelling evidence has gradually accumulated that the latter can most likely be attributed to transmissions, presumably via contaminated food, of BSE agent from cattle to man (1)(2)(3)(4). Therefore, effective infection control measures for the containment and repression of BSE have become a matter of crucial importance to public health. According to the present state of knowledge, the countermeasures implemented in response to the BSE epidemic are expected to minimize or even eliminate the risk of new primary variant Creutzfeldt-Jakob disease infections of humans directly originating from bovines (5).
However, further challenges in the area of infection control arise from the hypothetical risk that the BSE agent might have spread via contaminated feed such as meat and bone meal to sheep (5) and that BSE, like scrapie, might now be sustained in the ovine population. The clinical symptoms of scrapie, which has been endemic in sheep for centuries without any apparent association with human disease, cannot be reliably distinguished from those exhibited by experimentally challenged BSE-infected ovines. "While it is possible to demonstrate the presence of a TSE by several laboratory techniques using microscopy, electron microscopy, or immunological methods which detect the abnormal form of the prion protein, distinguishing between one strain of scrapie and another, and between BSE and scrapie, is not straightforward" (6). So far, reliable differentiation of BSE and scrapie in sheep has required time-consuming and expensive strain-typing in mice using lesion profiles (7). Therefore, the development of new methods for an inexpensive, robust, and rapid discrimination between BSE and scrapie constitutes a topical challenge in the surveillance of ovine TSEs addressed in a variety of studies (8,9). During the past few years, considerable progress has been achieved in this field of TSE research, predominantly by using immunobiochemical techniques (10 -13). However, apart from having some practical intricacies (14), these approaches require specific adjustments for each individual combination of TSE strain and host species. Therefore, alternative methods of strain differentiation, which do not require specific antibodies and can also be applied to a broad spectrum of TSEs and host species, would substantially improve our means for the molecular typing of TSE agents not only in sheep but also potentially in cattle and humans.
The causative agent of TSEs is widely considered to represent a new biological principle of infection. The prion hypothesis (15) holds that TSE agents ("prions") consist essentially if not entirely of misfolded prion protein PrP Sc . The normal cellular isoform of this protein (PrP C ) is expressed in neurons, lymphoid cells, and other tissues of mammals. According to the "protein only" model of the prion hypothesis, TSE agents replicate through a molecular mechanism in which abnormally folded PrP Sc acts as a catalyst or template nucleus, which recruits cellular prion protein and transforms it into its own "infectious" spatial structure (for a review see Ref. 16). This etiological model implies that the characteristic phenotypic features of different scrapie and other TSE agents, of which more than 20 have been isolated in different breeds of mice alone (17), must be coded in the secondary, tertiary, or quaternary structure of PrP Sc or in its specific glycosylation. While PrP C contains about 42% ␣-helical and only 3% ␤-sheet struc-ture, PrP Sc is substantially made up of ␤-sheets and exhibits a markedly reduced ␣-helical proportion (18 -20). Therefore, within the framework of the "protein only" model of the prion hypothesis, the secondary structure and, more specifically, the ␤-sheets of PrP Sc molecules appear as key candidates for coding the phenotypic characteristics of TSE strains.
During the past few years it has been conclusively shown that several different TSE strains can be distinguished by immunobiochemical typing of the electrophoretic mobilities and glycosylation characteristics of PrP Sc (or its protease-resistant core, PrP27-30) in the Western blot (21)(22)(23)(24). When comparing two strains of hamster-adapted transmissible mink encephalopathy (TME), termed hyper (HY) and drowsy (DY), the unglycosylated fraction of PrP27-30 derived from HY-and DY-PrP Sc by digestion with proteinase K exhibited different apparent molecular masses of 21 and 19 kDa, respectively (25). This experimental evidence for the presence of differentially accessible cleavage sites in HY-and DY-PrP Sc could most plausibly be accounted for by differences in the conformation of the misfolded prion protein molecules that were derived from PrP C with an identical amino acid sequence. Consistent with these findings, a conformation-dependent immunoassay has provided further indirect evidence that PrP Sc molecules from HYand DY-TME as well as from six other hamster-adapted TSE strains have differences in the three-dimensional structure of their polypeptide chains (26). When Fourier transform-infrared (FT-IR) spectroscopy was used to directly investigate the structure of prion proteins associated with different TSE strains, the conclusions described above were confirmed and substantially expanded by showing that PrP Sc extracts from HY-and DY-TME in hamsters exhibit different ␤-sheet structures (27).
Thus, conceptual as well as experimental clues strongly point to strain-specific phenotypic information of TSE agents either being encoded or at least mirrored in the secondary structure of PrP Sc . This suggests FT-IR spectroscopic structural profiling of PrP Sc or PrP27-30, a technique that should efficiently resolve differences in the secondary structure of the protein, as a promising tool for the molecular typing and differentiation of TSE agents.
With this rationale, the experiments described in this report pursued two aims. First, we intended to identify in laboratory animals TSE agents that cause similarly presenting TSEs and thereby mirror the diagnostic challenge of distinguishing infectious isolates that can neither be differentiated by the clinical symptoms or neuropathological changes they produce nor by the immunobiochemical properties of their associated PrP Sc . For this purpose, Syrian hamsters were chosen as model animals because, as outlined above, they have provided key insights into the relationship between phenotypically different TSE strains and PrP Sc structure. Second, upon successful identification and passage of such isolates in the hamster model, we investigated whether it is possible to reliably distinguish them and other TSE agents by FT-IR spectroscopic structural characterization of PrP27-30 extracted from the diseased individuals.
In the following, we report on an experimental proof-ofconcept that FT-IR profiling of PrP27-30 potentially provides a biophysical method for the swift differentiation of TSE agents, including those that are difficult or even impossible to discriminate by a variety of approaches for fast strain typing. Because the model animals used in our study have frequently provided base-line information about natural TSEs, this proof-of-concept may well be indicative of a diagnostic approach that allows the rapid and reliable discrimination of strains in nonexperimental TSEs of animals and humans.

EXPERIMENTAL PROCEDURES
TSE Agents and Animal Experiments-Serial passaging of hamsteradapted scrapie strains 263K, ME7-H, and 22A-H and of a new hamster-adapted BSE isolate, BSE-H, was performed by intracerebral infection of outbred Syrian hamsters with 50-l aliquots of 1% (w/v) hamster-brain homogenates in TBS (10 mM Tris-HCl, 133 mM NaCl, pH 7.4) from terminally ill donors. Hamster scrapie strain 263K (28) was originally provided by R. H. Kimberlin and has been serially passaged for more than 20 years in our laboratory.
Strains ME7-H and 22A-H generated by R.H. Kimberlin after 4 and 7 passages in hamsters, respectively (29), were kindly provided by the Institute for Animal Health, Edinburgh, UK. Isolates "RIV/8" of ME7-H and "RIV/12" of 22A-H were used for the first passage of these TSE agents at the Robert Koch-Institute.
BSE-H was newly derived in our laboratory after one passage of BSE agent from cattle in mice and subsequent transmission to hamsters. 2 Briefly, 20-l aliquots of 1% (w/v) brain tissue homogenate in TBS from a German BSE case in a cow imported from the UK were injected intracerebrally into C57Bl/10 mice; between 529 and 697 days postinfection (dpi), 4/9 recipients succumbed to fatal neurological disease with clinical symptoms strongly reminiscent of murine transmissible spongiform encephalopathy. 50-l aliquots of 1% (w/v) brain homogenate in TBS from a diseased C57Bl/10 mouse sacrificed at 606 dpi were then intracerebrally inoculated into hamsters. This produced terminal TSElike symptoms between 360 and 467 dpi in 10/10 animals.
For the present study, all examinations were performed on the 2nd serial passages of BSE-H, ME7-H, and 22A-H in our laboratory, which showed incubation times of 287 Ϯ 28, 331 Ϯ 16, and 206 Ϯ 8 days (expressed as mean Ϯ S.D.) until the occurrence of terminal disease, respectively, and on a passage of 263K with an incubation time of 83 Ϯ 5 days. These incubation times remained constant in further serial transmissions (data not shown) indicating stable propagation of BSE-H, ME7-H, and 22A-H after the first passage in our hamsters.
Animals were regularly observed for clinical symptoms and humanely sacrificed by CO 2 euthanasia at the terminal stage of disease. After sacrificing, the brains were immediately removed and further processed as outlined below.
Lesion Profiles-Brains were fixed in 4% formalin for 48 h, treated with formic acid for 60 min at room temperature to reduce infectivity, postfixed in formalin for 24 h, and trimmed coronally into 2-3-mm-thick slices. After processing in an enclosed tissue processor, samples were embedded in paraffin wax. 6-m microtome slices from the relevant regions were mounted onto slides and stained in ethyl eosin for 5-10 min. The scoring of vacuolar lesions was performed as described elsewhere (7).
PET Blot Mapping of Cerebral PrP Sc Distribution-PrP Sc -PET blotting (31) of hamster brain slices was performed as described elsewhere (30). In brief, hamster brains were fixed in 4% formalin for 48 h, treated with formic acid for 60 min at room temperature to reduce infectivity, postfixed in 4% formalin for 24 h, and trimmed coronally into 2-3-mmthick slices using a brain slicing mold. After processing in an enclosed tissue processor, samples were embedded in paraffin wax. 6-m microtome slices were mounted on nitrocellulose membranes (0.45 m pore size; Bio-Rad) and dried flat at 50°C for 16 h. For PET blotting according to Schulz-Schaeffer et al. (31), specimens were washed in TBS, pH 7.8, with 0.05% Tween 20 (TBST), and digested with 25 g/ml proteinase K (Roche Applied Science) in a buffer containing 10 mM Tris-HCl, pH 7.8, 100 mM NaCl, and 0.1% (w/v) Brij 35 for 2 h at 55°C. Sections were denatured in 3 M guanidine isothiocyanate, blocked in 0.2% (w/v) casein in TBST, and incubated with mouse ␣-hamster PrP antibody 3F4 (mAb 3F4; 1:2500 (32)) overnight at 4°C. After incubation with the secondary antibody (alkaline phosphatase-labeled rabbit antimouse IgG (1:2000; Dako, Denmark)) for 60 min at room temperature, sections were stained with nitro blue tetrazolium (Sigma) and 5-bromo-4-chloro-3-indolylphosphate (Sigma) to visualize the reaction product. The immunostained PET blots were assessed for cerebral PrP Sc distribution using a stereo microscope. For negative controls, brain slices were incubated in normal mouse serum (1:25,000) instead of mAb 3F4 prior to incubation with secondary antibody.
Western Blot Typing of PrP27-30 -50 l of 10% (w/v) brain homogenates in TBS, pH 7.4, were mixed with 5 l of 13% (w/v) Sarkosyl and 10 l of proteinase K stock solution (1 mg/ml; Roche Applied Science) and subsequently digested for 60 min at 37°C (33). The digestion was stopped by adding 435 l of 2ϫ sample buffer, i.e. 4% (w/v) SDS, 10% (v/v) 2-mercaptoethanol in 120 mM Tris-HCl, pH 6.8, containing 20% (w/v) glycerol and 0.05% (w/v) bromphenol blue, and boiling for 5 min. 1-5 l of the solution (corresponding to 1-5 ϫ 10 Ϫ5 g of homogenized brain tissue) were separated in a 15% SDS-PAGE (34) or in Tris glycine gels (NOVEX, Invitrogen) and subsequently blotted onto polyvinylidene difluoride (PVDF) membranes (Millipore) using the semidry method (Fast-Blot; Biometra). The PVDF membranes were blocked with 5% (w/v) low-fat milk powder in TBS for 30 min and incubated overnight at 4°C with mAb 3F4 (1:2000) in 3% bovine serum albumin in TBS. After washing in TBS and incubation for 60 min at room temperature with the secondary antibody (biotinylated goat anti-mouse IgG (1:2000) in 3% bovine serum albumin in TBS), a biotin-streptavidin kit (Dako, Denmark) for signal enhancement was used (30 min; at room temperature). After washing the membranes in TBS, antibody binding to PrP was visualized using a mixture of nitro blue tetrazolium (Sigma) and 5-bromo-4-chloro-3-indolylphosphate (Sigma) as substrate. The relative staining intensities of di-, mono-and nonglycosylated PrP in each sample were determined by densitometry for a graphical representation of the glycosylation profiles; densitometric measurements were performed in triplicate for each blotting membrane with samples from at least four donor animals infected with 263K-, ME7-H-, 22A-H, and BSE-H agent.
Assaying the Sensitivity of PrP Sc to Proteinase K at Different pH Values-25 l of 10% (w/v) brain homogenates in TBS from hamsters infected with the four different TSE strains (containing approximately the same amounts of PrP Sc ) were diluted 1:20 in 50 mM Tris adjusted to pH values of 4.0, 5.0, 6.0, 6.5, 7.0, 7.4, 8.0, 8.5, 9.0, and 10.0. Samples were digested with proteinase K as described above at a protease concentration of 40 g/ml for 60 min at 37°C. After stopping the digestion by adding 2ϫ sample buffer and boiling for 5 min, sample aliquots were subjected to SDS-PAGE, Western blotting, and PrP immunostaining with mAb 3F4 as described above.
FT-IR Spectroscopic Characterization of PrP27-30 -263K-, ME7-H-, 22A-H-, and BSE-H-associated proteinase K-resistant prion protein PrP27-30 was purified from the brains of intracerebrally infected Syrian hamsters at the terminal stage of disease, using a protocol published previously by Diringer et al. (35) with some modifications: (i) five hamster brains were used instead of 20 brains as starting material. (ii) At each step of the purification procedure that required centrifugation of samples, new polycarbonate vials were used (26.3-ml polycarbonate tubes; Beckman). (iii) Resuspending of pellets by ultrasonification was carried out in glass instead of polycarbonate tubes in order to avoid contamination of the protein with finely dispersed polycarbonate disintegrated from the tube walls. (iv) The final pellet was resuspended in 1 ml of distilled water adjusted to pH 8.5 with 1 M Tris solution and divided into volumes representing 2, 1, and 0.5 "brain equivalents" (BE), i.e. 2, 1, and 0.5 g of starting brain material, and centrifuged in a Beckman TL-100 ultracentrifuge using a TLA-45 rotor at 45,000 rpm for 1 h at 4°C. The supernatant was discarded, and the protein pellets were stored at Ϫ20°C until use.
For determination of the total protein content, 0.5 BE were mixed with 50 l of sample buffer and boiled for 5 min. The total protein content was determined by staining with Amido Black as described elsewhere (33). The purity of the extracted protein was monitored by SDS-PAGE and subsequent silver staining as described previously (33), and PrP27-30 was also visualized by Western blotting using mAb 3F4 as described above (not shown). The amount of total protein in the final fraction was usually in the range of 50 -150 g/5 BEs.
FT-IR Spectroscopic Measurements-FT-IR spectra of PrP27-30 were collected with a Bruker IFS28B FT-IR spectrometer. Spectral resolution applied was 4 cm Ϫ1 ; for apodization a Blackman-Harris 3-Term function was applied, and a zero filling factor of 4 was used yielding an encoding interval of ϳ1 data point per wave number. For each spectrum 128 scans were co-added and averaged. Transmission/ absorption FT-IR spectra were collected and electronically stored between 4000 and 400 wave numbers, while the instrument was continuously purged with dry air to reduce water vapor absorption. The FT-IR spectra were collected in the front measurement channel of an IFS28B FT-IR spectrometer (Bruker Optics GmbH, Germany) equipped with a DTGS detector.
For FT-IR measurements of samples in D 2 O, the protein from 2 BEs (20 -60 g) was resuspended and centrifuged twice using a TLA-45 rotor at 45,000 rpm for 10 min at 4°C in 1 ml of 0.1% Zwittergent 3-14 in D 2 O (Z/D 2 O) to obtain a finely dispersed protein suspension. Hydrogen-deuterium (H/D) exchange to equilibrium was obtained after 60 min of incubation at room temperature. Aliquots from the supernatant of the second centrifugation were used to collect reference spectra for digital subtraction from the protein Z/D 2 O suspension. The spectra of PrP27-30 were obtained from samples resuspended in Z/D 2 O with a final concentration of ϳ10 g/l. 1.8 l were transferred to an IR cell constructed from two CaF 2 windows, in one of which a cylindrical cavity with a 6-m path length was engraved. PrP27-30 samples from three independent purification runs of 263K, ME7-H, 22A-H, and BSE-H were each examined.
For comparison and graphical representation, the spectra were vector-normalized between 1600 and 1750 wave numbers to get similar intensities of the bands. Second derivatives were calculated using a 13-point smoothing function.

RESULTS
With the study background outlined above, we focused on approaches for the characterization and discrimination of TSE agents that, in principle, would also be applicable for a swift strain differentiation under field conditions. This excluded parameters such as the incubation time or techniques such as strain typing by lesion profiling in mice (7,17) as diagnostic options and produced the following results.
Clinical Examination for Neurological and Behavioral Symptoms-Hamsters diseased with 263K scrapie showed head bobbing, generalized tremor, and ataxia of gait. These animals were frequently and persistently in motion, easily irritated by touch and noise, upon which they often twitch, and had difficulties maintaining balance and rising from a supine position.
In contrast, clinically ill animals challenged with ME7-H or 22A-H scrapie agent exhibited phlegmatic sluggishness with bradykinesia and kyphosis. Unlike 263K scrapie hamsters, ME7-H-and 22A-H-infected animals were frequently and persistently resting, not obviously irritated by touch or noise, and well able to slowly rise from a supine position until very advanced stages of disease. Head bobbing, generalized tremor, or ataxic gait as in 263K scrapie were not observed, but the animals showed signs of hind limb paralysis.
Hamsters diseased with BSE-H were also found to rest frequently and continuously in a kyphotic position and then appeared lethargic. However, in contrast to the phlegmatic sluggishness and apparent unresponsiveness to irritation by touch and noise as observed in ME7-H and 22A-H scrapie, BSE-H hamsters exhibited spontaneous convulsions from a resting position and had an extreme sensibility to touch (especially to the hindlimbs) which easily triggered tetanic responses. When moving, BSE-H hamsters showed ataxia of gait and hind limb paralysis without bradykinesia. Unlike the 263K scrapie hamsters, they were fairly well able to rise from a supine position and did not show head bobbing and generalized tremor.
Neurological and behavioral symptoms allowed us to distinguish 263K scrapie and BSE-H from each other as well as both from ME7-H and 22A-H. However, a symptomatic discrimination between the latter two was not possible. Neuropathological Lesion Profiling-Following clinical examination, we analyzed the cerebral lesion profiles caused by 263K, ME7-H, 22A-H, and BSE-H agent. Lesion profiles provide a well established tool for the typing of TSE strains in mice by mapping the targeting and measuring the severity of vacuolation in selected brain areas (7). For hamsters infected with 263K or BSE-H lesion profiles showed slight but significant differences between each other and with respect to ME7-H and 22A-H (Fig. 1). As compared with the latter two scrapie strains, the most prominent differences were found in the dorsal medulla and in the thalamus (scoring areas 1 and 5) for BSE-H or in the superior colliculus, the hypothalamus, and the septum (scoring areas 3, 4 and 7) for 263K. On the other hand, scrapie isolates ME7-H and 22A-H produced similar patterns of vacuolation in the brains of infected hamsters which are virtually indistinguishable.
PET Blot Mapping of Cerebral PrP Sc Distribution-In addition to the histological analysis for vacuolation, brain specimens from our model animals were also examined for the pattern of cerebral PrP Sc distribution. This approach for strain differentiation (36) was pursued using the PET blot technique (31). For each of the four TSE isolates passaged in our hamsters, we scrutinized sets of brain specimens from six terminally ill donors that revealed consistently reproducible cerebral staining patterns for 263K, ME7-H, 22A-H, and BSE-H. The most prominent differences in the distribution and intensity of PrP Sc deposition were found at the four coronal cutting planes displayed in Fig. 2. PET blots with negative controls in which mAb 3F4 was replaced by normal mouse serum did not produce any PrP or unspecific immunostaining for the four different TSE isolates.
In light of the findings summarized in Table I, PET blot visualization of cerebral PrP Sc distribution appears as a useful tool for the phenotypic characterization of TSE agents. In our experiments this approach allowed us to clearly make a distinction between 263K and BSE-H and to discriminate both of these isolates from ME7-H and 22A-H. However, as with lesion profiles, PET blot differences that would enable a diagnostic separation of ME7-H and 22A-H could not be discerned.
Western Blot Typing of PrP27-30 -In order to compare the electrophoretic mobilities and glycosylation characteristics of PrP27-30, brain homogenates from hamsters infected with the four different TSE isolates were digested with proteinase K, subjected to SDS-PAGE, and immunoblotted using the mAb 3F4. This technique, which is commonly in use for PrP-based molecular differentiation of TSE strains (21)(22)(23), revealed similar glycosylation patterns of PrP27-30 for 263K and BSE-H (Fig. 3A, lanes 1 and 2). For these isolates, the bulk of the pathologically misfolded protein was diglycosylated (Fig. 3B). Although PrP27-30 associated with 263K and BSE-H agent cannot be clearly differentiated by the glycosylation pattern, or "glycotype," it can be easily distinguished by the different electrophoretic mobility of the unglycosylated band (Fig. 3, lanes 1  and 2). This indicates differential processing of 263K-and BSE-H-associated PrP Sc by proteinase K due to differences in the protein structure and/or protein aggregation. Conspicuously, PrP27-30 from BSE-H-infected hamsters exhibited biochemical characteristics (i.e. a preponderance of the diglycosylated band and a relatively low molecular mass of the unglycosylated proportion) previously observed (3,21,23,37) for BSE-derived infections in a variety of other species.
PrP27-30 fractions from ME7-H-and 22A-H-infected hamsters showed more or less overlapping glycotypes that, however, clearly differed from those observed for 263K and BSE-H (Fig. 3, A, lanes 3 and 4, and B). Furthermore, they exhibited identical electrophoretic migration patterns for all three PrP27-30 bands (Fig. 3A, lanes 3 and 4). Thus, conventional immunobiochemical Western blot typing of pathological prion protein did not allow us to make a distinction between ME7-H and 22A-H.
Assaying the Sensitivity of PrP Sc to Proteinase K at Different pH Values-Assaying the sensitivity of PrP Sc to degradation by proteinase K at various pH values provided a further potential approach for the distinction of TSE agents (38,39). With this rationale, brain homogenates from hamsters infected with 263K, BSE-H, ME7-H, and 22A-H were exposed to proteinase K at different pH values between pH 4.0 and 10.0 prior to SDS-PAGE and immunoblotting using mAb 3F4. This revealed profiles for the pH-dependent sensitivity of PrP Sc to proteinase K as shown in Fig. 4, A-D. On this basis it was possible to distinguish 263K and BSE-H from each other, as well as both of them from ME7-H and 22A-H scrapie.
In the assay, 263K scrapie showed a maximum of proteinase K resistance at about pH 7.0 -7.4, with gradually increasing proteolytic sensitivity toward acidic and strongly increasing proteolytic sensitivity toward alkaline conditions (Fig. 4A). For PrP Sc from the other examined TSE agents, the most prominent resistance to proteinase K was observed in the pH range of 4.0 -6.5 (Fig. 4, B-D, lanes 1-4). Most interesting, PrP Sc from BSE-H showed already strong or complete degradation by proteinase K at pH values of 7.4 and 8.0, respectively (Fig. 4B,  lanes 6 and 7). As compared with BSE-H, PrP Sc from ME7-H and 22A-H displayed higher proteolytic resistance at pH 7.4 -8.0, but at pH values Ͼ8.0, it was also degraded more rapidly than that from hamsters infected with 263K scrapie (Fig. 4, C  and D, lanes 6 -8). However, also in this assay no significant differences could be observed for ME7-H-and 22A-H scrapie (Fig. 4, C and D). FT-IR Spectroscopic Characterization of PrP27-30 -In order to get information about the secondary structure of PrP Sc derived from the different TSE agents on a molecular level, we analyzed PrP27-30 extracts from each isolate by FT-IR spectroscopy. Fig. 5 shows the second derivative spectra obtained from independent PrP27-30 preparations of 263K, ME7-H, 22A-H, and BSE-H in the secondary structure-sensitive amide I region between 1700 -1600 wave numbers (cm Ϫ1 ).
The method used for the extraction of pathological prion protein from hamster brain homogenates has been shown previously for 263K scrapie to deliver highly purified PrP27-30 fractions in which the level of contamination with other proteins is not higher than ϳ10 -15% (35). Examination by SDS-PAGE and subsequent silver staining revealed only minor, relatively uniform contaminations such as ferritin (35) and no significant differences in the degree of purity for the PrP27-30 extracts of 263K, ME7-H, 22A-H, and BSE-H (Fig. 6). Although some variation was observed in independent extraction runs with respect to the silver staining intensity of unspecific bands, this did not substantially  affect the FT-IR spectra obtained for the four different TSE isolates. Thus, several different lines of evidence strongly suggest that the FT-IR spectra obtained from our extracts mirror in their amide I region primarily structural features of PrP27-30 rather than those of different contaminants. The amide I band originates essentially from CϭO stretching vibrations of amide groups of the protein backbone and is the most useful infrared absorption band for secondary structure studies of proteins by FT-IR spectroscopy (40,41). Within this spectral region PrP27-30 from the four different TSE isolates exhibited specific infrared patterns as demonstrated by variations in the width and relative intensities of the band components and by differences in individual peak positions (Table II).
All strains investigated in D 2 O suspension showed two intense amide I band components in the range of 1616 -1640 cm Ϫ1 , which can be attributed to different ␤-sheet structures (42)(43)(44). A common characteristic for all scrutinized TSE agents was the peak at 1620 cm Ϫ1 , but the second higher frequency ␤-sheet band was distinct for each of the individual isolates (at 1637 cm Ϫ1 for 263K, 1634 cm Ϫ1 for ME7-H, 1632 cm Ϫ1 for BSE-H, and at 1630 cm Ϫ1 for 22A-H; Table II). Furthermore, the two ␤-sheet absorption bands were clearly separated in the case of 263K and ME7-H, although for BSE-H the component peak at 1632 cm Ϫ1 is only weakly resolved, and, conversely for the isolate 22A-H, the peak at 1620 cm Ϫ1 appeared as a shoulder of the ␤-sheet band component at 1630 cm Ϫ1 . Taken together, these results suggest that the PrP27-30 extracts from the different TSE agents feature significant differences in their ␤-sheet structure.
At 1671 and 1670 cm Ϫ1 , respectively, three TSE agents (263K, BSE-H, and 22A-H) produced pronounced high frequency amide I absorption bands, which we tentatively assign to turn structures. PrP27-30 from ME7-H and BSE-H, however, showed absorption bands also at 1679 and 1677 cm Ϫ1 , respectively, which might be attributed to the high frequency component band of ␤-sheet structures (41,42). All strains analyzed showed a distinct absorption band with varying relative intensity at 1656 -1659 cm Ϫ1 , which is tentatively assigned to ␣-helical structures (45,46).
Additionally, small amide I band components at 1642 and 1647 cm Ϫ1 were observed for 22A-H and BSE-H, respectively. No such features could be found in the spectra of ME7-H and 263K. The origin of this component remains unclear. Absorp-tion features at both frequencies have been described to originate from disordered structure in several proteins (42). However, the same components might also indicate an ␣-helical structure shifted to lower frequencies by unusual interactions between amide groups and solvent or by the accessibility of amide groups to H/D exchange (40,45). Although with the FT-IR data collected in our study highly resolved secondary structure assignments are not possible, our examinations revealed consistent differences in the spectral patterns of PrP27-30 from the four passaged TSE agents. This provided strong evidence that PrP27-30 extracted from 263K-, ME7-H,-22A-H-, and BSE-H-infected hamsters featured distinct conformations in ␤-sheet and other secondary structure elements and allowed us to clearly distinguish all passaged TSE agents from each other. DISCUSSION For our typing experiments we used three different scrapie strains passaged in hamsters (263K, 22A-H, and ME7-H) and an isolate of hamster-adapted BSE agent newly derived in our laboratory. Two of the passaged agents, 22A-H and ME7-H, turned out to cause TSEs with indistinguishable neurological and behavioral clinical symptoms, indistinguishable lesion profiles, and indistinguishable electrophoretic mobilities or glycosylation patterns of PrP27-30. Even when additional methods for their neuropathological or biochemical differentiation, such as PET blot analysis of the cerebral PrP Sc distribution or Western blot typing of pathological prion protein after proteinase K digestion at different pH values, respectively, were applied, a reliable discrimination between ME7-H and 22A-H was not possible. Phenomenologically, the latter two strains differed only with respect to their incubation times (331 Ϯ 16 versus 206 Ϯ 8 days), but this parameter would not provide a diagnostic option under field conditions.
Unexpectedly, in our model animals ME7-H and 22A-H produced somewhat longer incubation periods upon intracerebral passage than originally reported by Kimberlin et al. (29). The reason for this is yet unknown but might possibly be accounted for by adaptation of the agents to our particular breed of Syrian hamsters.
Other than with the clinical, neuropathological, or immunobiochemical diagnostic methods described above, all four isolates, including ME7-H and 22A-H, could be clearly differ-  1-10). For each of the four TSE agents brain homogenates from four different donor animals were analyzed and gave consistent results. entiated by FT-IR spectroscopic characterization of their pathological prion protein. This was possibly based on the observation that PrP27-30 extracts derived from 263K-, 22A-H-, ME7-H-, or BSE-H-infected hamster brains showed conformational diversity, attributed mainly to differences in ␤-sheet structure but also to variations in other secondary structure components.
Phenotypically Distinct TSE Agents Correlate with Conformational Differences in PrP27-30 -During the past few years a growing body of experimental data has provided several lines of evidence showing that PrP Sc molecules associated with phenotypically distinct TSE agents in the same species and hosts with identical PrP C display different conformations (21,22,25,26). This notion has also been substantiated by FT-IR spectroscopic data on PrP27-30 from three different hamster-passaged TSE strains (27).
In contrast to the study by Caughey et al. (27), we examined the second derivative spectra of PrP27-30 obtained from D 2 O suspensions. In our experience this has the advantage of providing more spectral contrast, although the assignment of spectral features to defined secondary structure may be somewhat complicated by the uncertainty of incomplete H/D exchange (18,46). With our approach we found absorption characteristics in the amide I region for PrP27-30 from 263K scrapie that were very similar to FT-IR data published previously (18). The turn feature observed at 1671 cm Ϫ1 , the ␣-helical component at 1656 cm Ϫ1 , and the ␤-sheet components at 1637 and 1620 cm Ϫ1 were detected at nearly identical peak positions and comparable band intensities, demonstrating major congruence in the observed secondary structure of PrP27-30, which was extracted by different methods in two laboratories from 263K-infected hamsters. However, the low frequency ␤-sheet component observed at 1620 cm Ϫ1 in our spectra from PrP27-30 of 263K in D 2 O suspension is not exactly matched by the two distinct ␤-sheet components displayed at 1627 and 1619 cm Ϫ1 in the report by Caughey et al. (18). This dissimilarity might be caused by differences in the exposure of samples to ultrasonic treatment.
In our study we were able to confirm the pioneering FT-IR approach by Caughey et al. (27) with a different set of scrapie strains and a hamster-adapted BSE isolate. Furthermore, this approach could be substantially extended by identifying, for the first time, differences in the FT-IR spectra of PrP27-30 from two TSE strains, ME7-H and 22A-H, which showed indistinguishable electrophoretic mobilities or glycosylation patterns (Figs. 3-5 and Table II). Possibly, this progress could not have been achieved without obtaining FT-IR spectra from suspensions of PrP27-30 in D 2 O, which produced a particularly strong spectral contrast. In any case, our findings show that phenotypic information of different hamster-adapted TSE agents is consistently mirrored in ␤-sheet and other secondary structure elements of PrP Sc /PrP27-30, also when immunobiochemical typing fails to detect structural differences in the pathological prion protein.
Whether the conformational diversities of PrP Sc /PrP27-30, also observed in this study with 263K, ME7-H, 22A-H, and BSE-H, provide the molecular basis for coding distinct phenotypic characteristics of different TSE strains (25,47) or merely reflect the effect of an as yet unknown informational molecule or factor in the infectious agent remains to be established.
Diagnostic Implications-The results of this study strongly corroborate the diagnostic potential of FT-IR spectroscopy for TSE strain differentiation. In particular, they show that infrared spectroscopic structural characterization of pathological prion protein is able to resolve conformational differences of PrP27-30 molecules from different TSE agents, which can be missed by immunobiochemical typing. The diagnostic potential and limitations of FT-IR spectroscopy in this regard will have to be determined by application of the technique to a broader spectrum of TSE agents, host species, and PrP genotypes than was possible in this "feasibility study." In addition to scrapie and BSE agents from sheep, such future studies should also include isolates from patients with different forms of sporadic, hereditary, or acquired human TSEs. Recently, the emergence of new BSE phenotypes in cattle has been reported from France and Italy (48,49) and alarmed the scientific community as well as public health authorities. FT-IR profiling of pathological prion protein provides a diagnostic tool that could not only contribute to the further elucidation of the infectious isolates underlying these conspicuous BSE cases but would potentially improve the epidemiological surveillance of TSE agents in general, as well as their sometimes poor standardization.