Combined detection of the polymorphisms at codon 136, 154 and 171 suggests that VRQ and ARR are antagonist in determining disease susceptibility: VRQ is associated with a high incidence of natural scrapie and ARR is associated with a low incidence of natural scrapie. Data obtained by screening both scrapie-affected sheep and control sheep of Manech Tête Rousse breed in France (Table 1) show that the VRQ allelic variant is present in 32.29% of all scrapie cases and in only 3.50% of control sheep. This allele is associated with a high incidence of disease with a χ² ratio of 62.20 (p = 3.1×10^–15). The ARR allelic variant is present in 30.77% of control sheep, but in 0% of scrapie cases: it is associated with a low incidence of disease with a χ² ratio of 38.38 (p = 5.8×10^–10). The AHQ variant, which generally has a very low occurrence in sheep breeds, can be tentatively associated with a low incidence of disease similar to that of the ARR variant (presently available data are insufficient to suggest a different incidence value). The ARQ variant does not show association with disease incidence (χ² = 2.18, p = 0.14), neither does the ARH variant (which is rarely distinguished from the ARQ variant). Therefore, the five alleles can be classified into three allelic groups:

R = resistant = ARR or AHQ

S = susceptible = ARQ or ARH

H = hypersusceptible = VRQ

Following this classification, appurtenance to the S class (the ARQ and ARH alleles) has no effect on the disease incidence; appurtenance to the R class (the ARR and AHQ alleles) has a negative effect on the disease incidence; appurtenance to the H class (the VRQ allele) has a positive effect on the disease incidence.

A genotype is obtained by pairing two alleles chosen among five possible variants; a three-class allelic model allows us to define six genotypic groups:

• • G1 genotypes result from two alleles both belonging to the R class (corresponding to the ARR/ARR, ARR/AHQ and AHQ/AHQ genotypes);
• • G2 genotypes result from two alleles belonging to the R and S classes respectively (corresponding to the ARR/ARQ, ARR/ARH, AHQ/ARQ and AHQ/ARH genotypes);
• • G3 genotypes result from two alleles belonging to the R and H classes respectively (corresponding to the ARR/VRQ and AHQ/VRQ genotypes);
• • G4 genotypes result from two alleles both belonging to the S class (corresponding to the ARQ/ARQ, ARQ/ARH and ARH/ARH genotypes);
• • G5 genotypes result from two alleles belonging to the S and H classes respectively (corresponding to the ARQ/VRQ and ARH/VRQ genotypes);
• • G6 genotypes result from two alleles both belonging to the H class (corresponding to the VRQ/VRQ genotype).

Table 2 shows the frequencies of genotypes in scrapie-affected sheep and in control sheep.

For each one of the genotypic groups Gi that are represented both in the scrapie-affected and in the control groups, we compute an observed genetic risk R(Gi)_obs as the ratio of the proportion of this group in control sheep, by the proportion of this group in scrapie-affected sheep. It is remarkable, in Table 3, that genotypes of the G4 group are found in the scrapie-affected group at the same frequency as in the control group (67.71%, 64.69%): the observed genetic risk, R(G4)_obs, is 1.05 ≈ 1. The same calculation for the G5 group gives an observed genetic risk R(G5)_obs of 9.93 ≈ 10. Because there is no scrapie-affected sheep belonging to the G2 genotypic group, the observed genetic risk R(G2)_obs is 0, which, given the size of the control group, can be estimated as ≤ 2×10^–2 (conversely, as there is no G6 sheep present in the control group, the genetic risk R(G6) cannot be estimated).

From the data above, it is tempting to speculate about the quantitative effect of pairing PrP allelic variants on scrapie susceptibility. Let us suppose that the risk factor R(Gi) associated with the genotype Gi (G1 to G6) is the produce of elementary factors r_j and r_k, with the indices j and k chosen among the three allelic susceptibility classes R, S, H defined above. Data from Table 3 allow us to calculate the values of the elementary risk factors r_S, r_H and r_R: the value R(_G4) for the susceptible–susceptible genotypic group G4 suggests that r_S ≈ 1. Hence, the relative risk factor associated to the hypersusceptible–susceptible genotypic group G5 suggests that r_H ≈ 10. In the same way, the relative risk factor associated with the resistant–susceptible genotypic group G2 suggests that r_R ≈ 2×10^–2. For the other genotypic groups, the calculated relative genetic risk assuming a multiplicative effect is consistent with the published data 〚17, 18〛. These studies compare the PrP genotypes at codon 136, 154, and 171 of natural scrapie-affected sheep with those of control sheep of the same age range from the same flock. Table 4 presents all the data collected from the different monitored flocks: Texel, Halfbred, Herdwick, Merino, Shetland, Suffolk, Poll Dorset, Soay, Manech Tête Rousse.

Table 4 shows, for each genotypic group, the observed relative genetic risk and the calculated relative genetic risk. Sheep distributions according to genotypes are significantly different in the scrapie-affected population and in the control population (χ² = 302.8913, p = 2.3935×10^–63). Most of the scrapie cases are in the G5 group, and the frequency of G5 animals in affected population (46.81%) is about ten times higher in this population than in the control population (5.17%): R(G5)_obs = 9.05. The calculated relative risk has a good predictive value (r_S r_H = 10). Many scrapie cases are also in the G4 group, and the frequency of G4 animals in the affected population (40.43%) is approximately the same in the control population (48.62%). The genetic risk R(G4)_obs equals 0.83. Again, the calculated relative risk has a good predictive value (r_S r_S = 1). The frequency of G6 animals in the affected population (11.17%) is nearly a hundred times higher than in the control population (0.17%): R(G6)_obs = 64.79. This is in accordance with the calculated value of the relative risk (r_H r_H = 10²). The relative frequency of G2 and G3 sheep in affected population (0 and 1.60%), compared with the same frequencies in the control population (26.72 and 6.90%), are in accordance with the calculated values (r_R r_S = 2×10^–2 and r_R r_H = 0.2). Finally, in the most resistant genotypic group, G1, no case of scrapie have been found in sheep neither in Europe nor in the USA. Only one affected Suffolk sheep with the genotype ARR/ARR has been reported in Japan 〚19〛. Our multiplicative calculation method takes into account a risk of four affected G1 sheep for ten thousand healthy sheep (rr = r_R r_R = 4×10^-4). It is to be noted that in a recent case-control study involving 18 British flocks of various breeds, Hoek and al. 〚20〛 obtained very similar results: the risks associated with different genotypes, compared to the risk associated with the ARQ/ARQ ancestral genotype gave odds-ratios similar to our R(Gi)_cal values: 0.23 for the AHQ/VRQ genotype and 0.51 for the ARR/VRQ genotype (G3 group), 15.9 for the ARH/VRQ genotype and 14.6 for the ARQ/VRQ genotype (G5 group), 0.04 for the ARR/ARQ genotype (G2 group), and finally 65.0 for the VRQ/VRQ genotype (G6 group).

In the outbreaks of natural scrapie described here, the primary risk factor associated with the occurrence of the disease is unambiguously the PrP genotype. Recent analysis of the human PrP gene in cases of sporadic and familial Creutzfeldt–Jakob disease (CJD) and Gerstmann–Sträussler syndrome have revealed a number of mutant alleles of this gene that appear to control the incidence of these diseases 〚21〛. Similar association of a normal allele with disease onset has been proposed for Huntington disease 〚22〛. But scrapie cannot be entirely genetic in origin, as proposed by Parry 〚23〛: this disease is also transmissible. Experimental challenges of sheep of different genotypes, with different scrapie isolates, are consistent with these observations of natural outbreaks 〚12〛. {ARR, AHQ}/VRQ are antagonist: the first is associated with low (r_R ≈ 2×10^-2) and the second with high (r_H ≈ 10) disease susceptibility. This fact suggests that the relative dominance of these allelic variants account for all scrapie sources (‘strains’). Only one reversed association between the SSBP/1 isolates and the BSE isolates is described by Goldmann et al. 〚9〛 for sheep heterozygous at both codon 136 and 171 (VQ/AR), but this inversion has been obtained in a very specific intracerebral challenge.

Differentiating a genetic origin and a genetic susceptibility in a contaminated environment (infectious dose) is very difficult. It is not clearly decidable whether resistant sheep cannot become infected, or whether the incubation period in these sheep is so long in relation to sheep life expectancy that clinical signs are never observed. In the first case, a resistant sheep might be able to limit multiplication of the infectious agent, and to tolerate a low level of scrapie infection. In the second case, resistant genotype sheep might act as carriers: without the appearance of the disease, a resistant sheep could act as a source of infection of more susceptible sheep. Unable to limit multiplication and spread of scrapie infection, susceptible sheep become more quickly ill 〚24〛. Hunter 〚25〛 says that it is possible that the rate-limiting step is neuro-invasion without which infection, from peripheral tissues, could not spread to the brain to cause illness and death. In this context, the multiplicative effect should be linked to a serial process determining firstly the multiplication and secondly the neuro-invasion of the infectious agent. Additional data will help refining the risk factors associated with each allele, and will probably allow a clearer understanding of the processes controlled by the genotypic factors. This evidence of a multiplicative effect of scrapie genetic susceptibility is very likely to exist also in other neurodegenerative diseases.

La tremblante du mouton est une encéphalopathie spongiforme transmissible, dont l’agent infectieux n’a pas été totalement caractérisé. L’infection se traduit par une accumulation au niveau du cerveau d’un isomorphe de la protéine prion (PrP) codée par l’hôte. Malgré l’évidence expérimentale d’une origine infectieuse de la maladie, l’existence d’une variabilité importante de la forme naturelle de la maladie au sein des élevages suggère une influence des facteurs génétiques. Il a été établi que le polymorphisme du gène de la PrP pouvait être corrélé à l’incidence de la tremblante ainsi qu’à la durée de survie des individus infectés. Cinq formes alléliques déterminantes ont ainsi été décrites à partir des variations des séquences d’acides aminés aux codons 136, 154 et 171. Deux formes alléliques ont été associées à une « résistance à la maladie », un à une « hypersensibilité à la maladie » et deux à un impact intermédiaire. L’objectif de ce travail est de proposer une méthode de calcul explicite du risque de tremblante. Nous étudions l’effet de l’appariement des cinq variants alléliques sur la sensibilité génétique à la tremblante. Nous montrons que la sensibilité d’un individu donné à la tremblante est le produit des effets élémentaires de chacun de ses deux allèles. Les facteurs de risque associés aux six catégories génétiques rencontrées sont les produits des facteurs de risques associés aux catégories alléliques correspondantes. Cette propriété multiplicative des facteurs de risques alléliques de la maladie constitue une piste pour comprendre les différentes cinétiques du processus dégénératif des maladies à prion.