Rapid Typing of Transmissible Spongiform Encephalopathy Strains with Differential ELISA

A strain-typing ELISA distinguishes bovine spongiform encephalopathy from other scrapie strains in small ruminants.

sitive to PK degradation at its amino-terminus than PrPsc from scrapie. Recently "atypical" forms of scrapie sensitive to PK, such as the Nor98 strain, have been described (19). Since 2002, large-scale epidemiologic studies have led to the identifi cation of hundreds of atypical cases in several European countries (20); these cases included an unusual occurrence in sheep with genotypes previously associated with high resistance to scrapie (including the ARR/ARR genotype) (21). Converging data show that most atypical cases share identical biochemical and biologic features with Nor-98 (22).
Glycoform ratios and the molecular migration pattern of PK-treated PrPsc, as obtained in Western blot techniques, enable characterization of scrapie strains and BSE. (These topics have been reviewed by Groschup et al. [23]) This method identifi ed similarities between BSE in cattle and experimental BSE in sheep (24,25) and similarities between vCJD and BSE in cattle (26,27). However, these typing immunoblotting techniques are not easily applied to test huge numbers of fi eld samples. To increase specifi city, a differential immunoblotting technique involving 2 specifi c antibodies, directed against the amino-and carboxytermini of PrP, has been developed (10,28). Comparison of the signals obtained for both antibodies provides an easy and fast identifi cation of the BSE strain because binding of the anti-N terminal antibody is almost completely suppressed. This approach has been used to characterize unusual BSE cases in cattle (29) and naturally infected scrapie sheep (14,30).
We describe an ELISA designed to distinguish BSE and scrapie strains, on the basis of their differential resistance to PK. The design of this method is similar to that of a rapid test widely used in Europe (TeSeE, Bio-Rad, Hercules, CA, USA) and allows rapid screening of a large number of sheep samples previously shown to contain PKresistant prion protein (PrPres) by a conventional test.

Homogenization of Nervous Tissues
Nervous tissues were homogenized and calibrated according to the Bio-Rad purifi cation protocol. Homogenates (20% w/v) were diluted 10-fold in a negative sheep brain homogenate or tested undiluted if absorbance measurement in A reagent was <0.5.

Processing of Samples (Scrapie-associated Fibrils Preparation)
Two batches (A and A′ series) of 21 samples (250 μL (20% homogenate) of each sample per tube, manual procedure) or 29 samples (250 μL of each sample per well in 2 deep-well plates, automated protocol) were analyzed together with 2 scrapie controls (nerve tissue from 2 scrapie sheep, 99-1316 and PG1259 [manual] or 99-1316 and 99-1487 [automated]) and an experimental BSE-infected sheep (397BS, AFSSA). Each batch was treated in 1 set of conditions of PK treatment. For the manual protocol, 250 μL of A reagent (TeSeE purifi cation kit) containing PK (10 μg) is distributed in each tube of the A series, and 250 μL of A′ reagent/PK (5% N-lauroylsarcosine sodium salt [w/v], 5% [w/v] sodium dodecyl sulfate [SDS] containing 27.5 μg of PK) in each tube of the A′ series. All the tubes were then homogenized by 10 inversions and incubated at 37°C for 15 min. Then, 250 μL of B reagent (Bio-Rad purifi cation kit)/phenylmethylsulfonyl fl uoride (PMSF) (fi nal concentration 4 mmol/L) was added, before homogenization and centrifugation for 5 min at 20,000 × g at 20°C.
For the automated protocol, the deep-well plates were successively incubated for 12 min at 37°C in the TeSeE NSP automated system. Each well of the fi rst plate was processed with 250 μL of A reagent/PK before 15 min of incubation at 37°C. Then 250 μL of the B reagent containing PMSF (12 mmol/L) was added and incubated 5 min at 37°C. The deep-well plate was centrifuged for 10 min at 2,000× g, 4°C. The second plate (A′ series) was processed similarly with the A′ reagent/PK. For both protocols, supernatants were discarded and the tubes (or the plates) dried by inversion on absorbent paper for 5 min. Each pellet was denatured for 5 min at 100°C with 25 μL of C reagent (Bio-Rad purifi cation kit). We then added 350 μL of R6 buffer containing 4 mmol/L AEBSF. For the fi eld isolates, serial dilutions (3-and 10fold in R6 buffer/AEBSF) were performed to ensure an optimal ELISA signal.

Immunometric Assay
All the reagents were provided by the Bio-Rad TeSeE Sheep & Goat detection kit. We distributed, in duplicate, 100 μL of samples (undiluted, 3-and 10-fold diluted) and controls in microtiter plates coated with the fi rst anti-PrP antibody. The plate reacted for 2 h at room temperature (RT). After 3 washing cycles (R2 buffer), 100 μL/well of the enzyme conjugate was added for 2-h reaction at RT. After 3 washing cycles, 100 μL of substrate was added for 30 min in the dark at RT, before blocking the reaction with 100 μL of stopping solution and reading the absorbance at 450/620 nm. The mean ratio of the absorbances obtained in the 2 conditions (A and A′) was calculated for each sample by selecting a range of appropriate dilutions providing absorbance measurements ranging from 0.5 to 2.5 for A reagent.

Principle of the Typing Test
The typing ELISA is based on the screening test for the postmortem diagnosis of BSE initially developed by the Commissariat à l'Energie Atomique (CEA) (32,33). After selective purifi cation of PrPres, the denatured PrPres is measured by using a sandwich assay in microtiter plates.
The capture antibody recognizes an amino terminal epitope, while the tracer antibody binds to the core of the protein ( Figure 1, panel A). In the fi rst set of conditions, the PK digestion is performed in a control medium (mixture of detergents and chaotropic agents) to preserve the Nterminal epitope (Figure 1, panels A and B, PK treatment in A conditions). By varying the conditions of PK treatment, 610 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 14, No. 4, April 2008 Figure 1. Principle of the 2-site immunometric typing assay. In the screening conditions (Bio-Rad, Hercules, CA, USA), after mild proteinase K (PK) digestion, denatured PK-resistant prion protein (PrPres) is captured by the solid phase-immobilized antibody SAF34, recognizing the octarepeat region, and shown by the Bar224 tracer antibody, directed against the core of the protein (A).
In the typing test, the PrPres-containing sample is treated in 2 sets of PK conditions (B). In the fi rst set of conditions (PK treatment in A reagent, Bio-Rad screening condition), the octarepeat region is maintained in scrapie-and bovine spongiform encephalopathy (BSE)-associated PrPres; in the second set of conditions (high PK concentration in A′ reagent), BSE-and labile strain-associated PrPres is more sensitive to PK digestion than the classic PrPres associated with scrapie strains. Calculation of the ratio in the 2 conditions differentiates BSE and labile strains from classic scrapie strains. ab, antibody.
PrPsc associated with the BSE strain appears more sensitive to PK than most of the other prion strains. Conditions of PK treatment can thus be defi ned for selectively deleting the epitope recognized by the capture antibody in BSE strain while preserving it in most scrapie strains ( Figure 1, panel B, PK in A′ conditions), in agreement with previous reports (5,10,23,28).
The fi rst step of the Bio-Rad test (purifi cation and concentration of PrPres) was performed either manually or by using the NSP automated system and adapted for the typing test. Determination of the ratio between the 2 measurements (R = A/A′) allows differential detection of the BSE strain. In the current conditions, this ratio is <2 for most of the scrapie strains (manual or automated protocol) and close to 6 and 10 for the automated and manual protocols, respectively, with the BSE strain (Table 1). During the current study, we made 2 major changes with regards to the Bio-Rad test: 1) we added PK inhibitors (PMSF and AEBSF) at 2 stages of the purifi cation procedure for better control of PK digestion, and 2) we performed a series of dilutions to determine the optimal range of absorbance, which allowed a precise determination of the A/A′ ratio (ranging between 0.5 and 2.5 absorbance unit, Figure 2, panels A and B). To minimize the interassay variations, the ratio obtained for each sample was further normalized by dividing it by the ratio obtained for the experimental ovine BSE control ( Figure 2, panel C) To evaluate this typing test, 37 brain samples from 25 ARQ/ARQ sheep experimentally infected with BSE (fi rst or second passage) were compared with the brain samples of 3 controls. All samples analyzed in 3 independent experiments provide A/A′ ratios >7 with a mean of 11 0.2 ± 1.4 (online Appendix Figure 1, panel A, available from www.cdc.gov/EID/content/14/4/608-appG1.htm). To further investigate the possible infl uence of the genotype, we tested 7 spinal cord samples from experimental BSE bearing the ARR/ARR genotype, in at least 4 independent experiments. A/A′ ratios ranged from 5.5 to 6.9 (mean 6.0 ± 0.4) and were statistically different from the experimental ARQ/ARQ ovine BSE used as control (mean 8.8 ± 0.9, p<0.001 for all samples except no. 38, p<0.05) (online Appendix Figure 1, panel B).

Specifi city of Typing Test
We analyzed a large series of fi eld isolates identifi ed as TSE infected, by a rapid test, in the framework of the French active surveillance network. Of these 270 samples (153 in 2002 and 117 in 2003), 42 samples, which were initially identifi ed by using the TeSeE test, were categorized by AFSSA as atypical due to lack of confi rmation by other rapid tests or World Organization for Animal Health-modifi ed SAF immunoblot (AFSSA, no. 2004-SA-0045) (34).
When tested with the automated typing test, 10 of the 270 samples had PrPres levels below the detection limit, even undiluted (optical density <0.5 in A conditions). Each series of 32 samples included 3 internal controls (see Table  l for results obtained for 20 different series of tests). Among the 2 scrapie controls, 1 (99-1316) was classifi ed as classical as it has a PK resistance similar to most of the fi eld isolates; the second one (PG1259 or 99-1487) was classifi ed as intermediate because its PK resistance was intermediate between classic scrapie and experimental ovine BSE. The distribution of the normalized ratio recorded with the fi eld isolates is shown in Figure 3. The absorbance ratio under the 2 conditions of treatment was remarkably reproducible for the classic scrapie control (mean 1.6 ± 0.2), while larger variations were observed for experimental ovine BSE and the intermediate scrapie control (mean 6.4 ± 0.9 and 3.8 ± 0.5, respectively). These results demonstrate that the present test clearly discriminates classic scrapie strain from the 2 other controls. Because of interassay variations, experimental ovine BSE and intermediate scrapie may slightly overlap (Table 1), further justifying the use of normalized ratio ( Figure 2, panel C, and Figure 3).
As shown in Figure 3, only 10 samples (3.8%) provided ratios compatible with experimental BSE (normalized ratio 0.7-1.3), and 28 provided ratios superior to experimental BSE in sheep (normalized ratio >1.3-10.5, Figure  3). Twenty-nine samples, as well as 4 of the 10 samples compatible with experimental BSE and 9 samples with  Figure 3). Among the samples that yielded ratios compatible with experimental BSE in sheep, 1 goat isolate, Ch636, was extensively studied because of its BSE-like profi le in 3 different Western blot techniques (1). As shown in Figure 3, this sample had a normalized A/A′ ratio close to 1, very similar to that for experimental goat BSE (Figure 4, panel A, Table 2). This result is confi rmed by the migration pattern of the nonglycosylated band ( Figure 4,

Analysis of Nor98 Isolates
The typing test was used to analyze 18 sheep isolates from Norway (Table 3). Ratios were almost impossible to calculate because of the large decrease in signal in A' conditions, as shown in Figure 5, panel A for 3 isolates. Only 1 sample (Lavik) showed characteristics of a conventional scrapie isolate, providing an A/A' ratio of 0.84 ( Figure 5, panel A), a normalized ratio of 0.11, and a Western blot profi le close to that of a French scrapie isolate (Figure 4, panel B, lanes 3 and 9; Figure 5, panel B, lanes 2 and 4). Other samples had a pattern that included a 12-kDa band ( Figure 5, panel B) (19,22,34), characteristic of the Nor-98 strain.
After adapting the conditions of the PK treatment in the second set of measurements (A' conditions), we observed (see legend, Figure 6) a much lower A/A' ratio for those Nor-98, which enables discrimination of highly sensitive PK samples (nos. 24 and 26, online Appendix Figure  2 [available from www.cdc.gov/EID/content/14/4/608-appG2.htm] and Table 3) to mildly sensitive PK samples (nos. 8, 11, 16, and 22).

Discussion
When this study was initiated, no case of natural BSE in small ruminants was recorded, and only a few experimental ovine BSE samples were available, all belonging to the same PrP genotype (ARQ/ARQ), and mainly from a fi rst passage. The possible impact of the genotype, the route of infection, and the number of passages on the biochemical properties of PrPres associated with the BSE strain are poorly understood. Now, further data suggest that, at least during the second passage in sheep, the biochemical properties (glycoform pattern in brain) of the BSE agent are unchanged (35,36). In this study using our ELISA, small ruminant BSE samples clearly behaved differently from conventional scrapie samples. However, slight differences may exist (see ARQ/ARQ vs. ARR/ARR genotype in on-

612
Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 14, No. 4, April 2008 Figure 2. Determination of the A/A' ratio. A dilution series was assayed for each analyzed sample to determine the optimal range that would permit precise determination of the A/A′ ratio (absorbance ranging from 0.5 to 2.5 absorbance units in A reagent). A) A/A′ ratio is close to 1 for PK-resistant prion protein (PrPres) associated with classic scrapie strains (manual protocol, see Experimental Procedures) and B) close to 10 for experimental ovine bovine spongiform encephalopathy (BSE)-associated PrPres. C) To minimize interassay variations, the ratio obtained for each sample is thus normalized by dividing by the ratio obtained for the ovine BSE sample. line Appendix Figure 1). We do not know whether these fi ndings refl ect differences in the PK sensitivity of PrPres associated with these genotypes or the infl uence of different tissues.
The main diffi culty encountered for the development of a typing test is evaluation of its specifi city and sensitivity. In the current study, we unambiguously identifi ed all 37 experimental ovine BSE samples from 25 sheep, including 10 from a second passage. There are few data describing the molecular features of PrPres associated with experimental BSE in goats (37,38). In the framework of the French scrapie strain-typing network, 18 goats were analyzed by this ELISA, and 2 appeared compatible with experimental ovine and caprine BSE. One of them (Ch636), when analyzed with other molecular typing tests, appeared indistinguishable from experimental BSE and was later confi rmed as the fi rst natural case of BSE in a goat (1), after experimental transmission in wild-type and transgenic mice. The second BSE compatible sample (TR041528) was later clearly identifi ed as a case of atypical scrapie as defi ned by its migration pattern (34). All these data suggest a good sensitivity for our test, which unambiguously identifi ed all cases of experimental BSE in the sheep and goats tested, as well as the only natural case identifi ed to date in a goat.
Another key point during the development of this test was to ensure good reproducibility because this parameter clearly infl uences both sensitivity and specifi city. Ratios  obtained for the classicalal scrapie control were highly reproducible, whereas ratios measured for the experimental BSE in sheep and the intermediate scrapie control varied much more, leading to an overlap of the 95% confi dence interval (Table 1). To minimize interassay variations, the ratio obtained for each unknown sample was thus normalized by taking as reference the ratio measured for the ovine BSE sample (Figure 2, panel C, and Figure 3) in the same experiment. This enabled us to defi ne the range of normalized ratio compatible with BSE as the mean of experimental ovine BSE ± 2σ on the basis of reproducibility experiments recorded in Table 1 (1). This result indicates that the specifi city of this test is not that good because 9 false-positive results were recorded in 260 samples (specifi city 96.5%). However, the test appears useful since it excluded the presence of BSE for most fi eld samples, thus restricting the use of more specifi c but timeconsuming methods, like experimental transmission in mice, to a small number of isolates. Moreover, in a single screening, this test classifi ed all TSE-infected isolates as a function of their PK resistance and thus provided a rapid  . Proteinase K (PK) sensitivity of Nor98 isolates in stringent and mild detergent conditions. The ELISA typing test was performed on Nor98 isolates, with 5 concentrations (0.4-1.1 μg per mg of tissue) in the stringent A′ reagent (A) or in the mild A′′ reagent, adapted for PK-sensitive strains (B) (see Experimental Procedures). A/A′ (or A/A′′) ratios were calculated for each PK concentration, and normalized by dividing by the A/A' ratio (or A/A′′) obtained for the experimental ovine bovine spongiform encephalopathy (BSE) sample at the maximal PK concentration. In the A' reagent, even at the lowest PK concentration (PK 0.4 μg/mg tissue), the normalized ratios (using the experimental ovine BSE A/A′ PK1.1 ratio) obtained for the Nor98 isolates are >1, thus being 3× more sensitive than experimental ovine BSE. To evaluate possible differences in PK sensitivity among Nor98 isolates, this experiment was reproduced with the A′′ reagent (panel B), which is 3-to 6-fold more protective than the A′ reagent, as shown by the corresponding normalized ratios (A′ or A′′ reagent) for the same PK concentration (1.1 μg PK/ mg of tissue). classifi cation of sheep isolates according to this criterion. The test could also be modifi ed, by adjusting the range of PK sensitivity, to classify Nor-98 isolates.
All these data demonstrate that this ELISA-based typing test is suitable for a routine analysis of fi eld samples, as assessed by the positive evaluation from the European Commission as one of the tests recommended to identify the possible presence of BSE in small ruminant fl ocks (http:// eur-lex.europa.eu/LexUriServ/site/en/oj/2005/l_010/ l_01020050113en00090017.pdf). These typing tests are mainly designed to identify the BSE strain in small ruminant fl ocks. They are performed exclusively in national reference laboratories and based on Western blot techniques. In this context, the present ELISA is one of the secondary tests to be used to confi rm BSE suspicion. We believe it will help clarify the status of these unusual isolates.