Support for single major genes influencing fat androstenone level and development of bulbo-urethral glands in young boars

Cette etude concerne une analyse genetique, en deux etapes, des donnees recueillies dans les quatre generations d'une experience de selection comportant une lignee temoin et une lignee selectionnee dans une population croisee a base de Large White et de Landrace. L'indice a deux caracteres utilise pour le choix des verrats de la lignee selectionnee combinait la teneur en androstenone du gras et l'epaisseur moyenne des glandes bulbo-urethrales droite et gauche, ce dernier caractere etant un indicateur du statut de maturite sexuelle des jeunes verrats. La teneur en androstenone a ete mesuree sur une biopsie de gras dorsal prelevee a 118 kg de poids vif alors que le developpement des glandes bulbo-urethrales a ete mesure par echotomographie, a l'aide d'une sonde rectale, a 99 kg de poids vif. Au total, 949 mâles entiers, issus de 57 peres et 353 meres, ont ete mesures pour l'un et l'autre caractere. Dans la premiere etape de l'analyse, les parametres genetiques des deux caracteres ont ete estimes a l'aide d'une procedure REML appliquee a un modele animal bicaractere. Les estimees d'heritabilite (± erreur standard) sont 0,55 ± 0,07 pour la teneur en androstenone du gras et 0,63 ± 0, 05 pour l'epaisseur moyenne des glandes bulbo-urethrales. Une correlation genetique relativement elevee (0,68 ± 0,05) a ete trouvee entre les deux caracteres. Une seconde etape de l'etude a consiste a tester, a l'aide de methodes d'analyse de segregation, l'hypothese d'un determinisme genetique mixte (polygenes + un gene majeur) pour chaque caractere. Un gene a effet majeur sur la teneur en androstenone du gras a ete mis en evidence : selon le modele genetique le plus explicatif, l'allele « faible » (L) est completement dominant sur l'allele «fort» (H), avec une difference entre les genotypes HH et LL (ou HL) voisine de trois ecarts types phenotypiques du caractere. Un gene a effet majeur sur l'epaisseur moyenne des glandes bulbo-urethrales (difference de l'ordre de deux ecarts types phenotypiques entre les genotypes homozygotes) a egalement ete mis en evidence avec une dominance probablement incomplete de l'allele «faible». Ces deux genes ne semblent pas etre lies avec le systeme d'histocompatibilite majeur (SLA). La possible identite des deux genes majeurs postules est discutee.

As reviewed by Willeke (1993), fat androstenone level in boars of a given weight or age is a highly heritable trait and has been successfully selected for. However, the extent to which genetic variation in this trait is associated with genetic factors influencing the sexual development of the boar or is dependent on the proper genetic capacity for C19D16 steroid synthesis in the puberal boar remains to be evaluated.
A selection experiment has been carried out in an attempt to answer the question as to whether it is possible or not to decrease fat androstenone levels in a population of boars without affecting the sexual maturity status of these boars, the latter being appraised by the size of bulbo-urethral glands at a fixed liveweight.
The purpose of this note is to report estimates of genetic parameters for the two traits included in the selection index as well as segregation analysis results providing evidence for major genes influencing the same two traits. Moreover, a linkage between a QTL having large effects on fat androstenone level and the major histocompatibility complex of the pig (SLA) was recently reported by Bidanel et al (1996) in the F 2 generation of the Meishan by Large White cross, whereas significant associations of some SLA haplotypes with male genital tract development had earlier been found by Rothschild et al (1986). Information on serologically defined SLA haplotypes was available for part of the animals of the present experiment and was taken into account in further segregation analyses.

Animals and traits
The present data set deals with a four-generation selection experiment conducted over the years 1988-1993 at the INRA experimental farm of RouiII6 (Vienne, France). From a foundation stock (Go) consisting of 141 Large White x Landrace (F d entire males sired by 18 different boars, two lines of boars have been established: an unselected control line (C) and a selected line (S). In the line S, replacement boars (one generation per year) were chosen among animals showing the highest values of a selection index including the following two traits: -the average thickness of right and left bulbo-urethral glands (B in mm), measured by echotomography using a rectal probe on the live animal at 99.1 ! 2.7 kg liveweight (average age: 162 days). The correlation of that live thickness measurement with the weight of bulbo-urethral glands measured after slaughter a few days later was 0.86 in a sample of 175 boars (Sellier et al, 1993); -the androstenone content of a small biopsy sample of backfat removed from the neck region at 117.6 ! 2.4 kg liveweight (average age: 185 days). Androstenone content (in ppm) was determined using the radioimmunoassay procedure described by Bonneau and Russeil (1985). A logarithmic transformation was used so that the variable included in the selection index was A = log (androstenone content x10 2 ).
The two-trait selection index (I = 100 -63A + 4B) had been established with the objective of decreasing A while maintaining a normal sexual maturity status (ie, no expected genetic change in B). The line C comprised four or five randomly chosen sires and, on average, 23 F l dams in each generation. The corresponding numbers for the line S were five and 48, respectively. For more details on the design of this experiment, see Sellier et al (1993).
A total of 949 candidate boars, from 57 sires and 353 dams, were recorded for both traits A and B throughout the experiment. In addition, serological SLA class I typing was performed on boars belonging to the foundation stock Go and to generations 1 and 2. The SLA class I antigens were defined by microlymphocytotoxicity testing as described by Renard et al (1988). Animals were typed by means of panels of 100 alloantisera, which allowed identification of 18 internationally defined SLA specificities (Bl, B3, Bll, W2, W4-W10, W12-W18) and 15 locally defined specificities (FJ1, D4, FJ4, FJ13, DB19, FJ20, SB22, FJ23, FJ25, FJ27, FJ28, DC31, FJ33, FJ35 and FJ39). For investigating the linkage of SLA with the postulated major genes, the number of informative meioses was 441.

Statistical analyses
In the first step of the analyses, heritabilities and phenotypic and genetic correlations for fat androstenone level and bulbo-urethral gland size were estimated by a bivariate REML animal model procedure using the VCE (variance component estimation) package (Groeneveld and Kovac, 1990). The mixed model of analysis included the fixed effect of year, the live weight at measurement as a covariate, the random effects of litter and animal additive genetic value and the random residual effect. Pedigree information on males, starting from Go, was used to calculate the inverse of the numerator relationship matrix. The distribution of fat androstenone levels was far from being normal, mainly owing to its marked skewness, and a suitable Box-Cox power transformation (MacLean et al, 1976) was used for removing deviation from normality prior to REML analysis (fig 1).
In a second step, segregation analyses (Elston and Stewart, 1971) were performed for both traits. Considering the results of some preliminary analyses (Fouilloux, 1995), the presence of major genes with large effects could indeed be suspected for several reasons such as the very high realized heritability for size of bulbourethral glands (mainly owing to a surprisingly sharp increase in the line S between generations 2 and 3) and some evidence for mixture of distributions for fat androstenone level. Analyses were carried out separately for each trait as a multivariate approach was precluded owing to computing cost considerations.
Several tests were performed. First, the hypotheses of mixed inheritance (H i , polygenes and one two-allele major gene) and polygenic inheritance (H o ) were compared. The conclusion of this test (ie, the existence of a major gene) led us to consider other hypotheses of mixed inheritance: a model in which one allele of the major gene is completely dominant over the second one (hypothesis H!) and then models assuming linkage between the major gene and the SLA locus (hypotheses H 2 and H2). The latter hypotheses were tested by use of the method described by Le Roy and Elsen (1992). This method implies that the recombination rate between SLA and the major gene is a parameter to be estimated as all other parameters of the model. All segregation analyses of fat androstenone level were carried out with a simultaneous Box-Cox transformation of the residues of the model (Morton et al, 1978). Using a transformation to remove skewness induced by a major gene is expected to increase the robustness of the segregation analysis, but this also leads to a loss of power (Demenais et al, 1986). Moreover, the pedigree submitted to analysis was simplified, as it was considered as a set of independent sire families.
The present h 2 estimate for fat androstenone level is higher than the value of 0.25 reported by Jonsson and Andresen (1979) and is close to the lower bound of the range of values (0.54-0.88) found for the same trait Willeke et al, 1987;Sellier and Bonneau, 1988) or for boar taint intensity score (Jonsson and Wismer-Pedersen, 1974 (1979), Toelle et al (1984), Bonneau and Sellier (1986), Sellier and Bonneau (1988) and predicted testis weight at 150 days of age were carried out in a Large White-Landrace composite population, and the realized heritability of testis weight was only 0.35 (t 0.02).
Regarding the relationships between fat androstenone level and bulbo-urethral gland size, the phenotypic correlation found here is quite similar to that reported by Bonneau and Sellier (1986). The present estimate of genetic correlation falls within the range of values (0.25-0.91) found in young boars for various genital tract measurements by Bonneau and Sellier (1986) and Sellier and Bonneau (1988). The latter authors reported a genetic correlation of 0.60 between testis size at 104 kg liveweight and fat androstenone level at 124 kg liveweight, which is in good agreement with the estimate found here with comparable stages of measurement for the respective traits.
The present estimates of genetic parameters thus differ from those of previous studies only by the higher heritability of the sexual maturity indicator.

Segregation analysis
The results of various segregation analyses are given in table I for fat androstenone level and bulbo-urethral gland size. For both traits, the likelihood ratio referring to the H l versus H o comparison is significant at the P < 0.001 level. The polygenic inheritance hypothesis must be rejected, showing that, whatever the trait, a major single gene was also segregating in the population studied. For fat androstenone level, the genetic model that best explained the situation under the mixed inheritance hypothesis appears to be the model H' in which the 'low androstenone' allele (L) is completely dominant over the 'high androstenone' allele (H). There is no indication of a linkage between this postulated major gene and the SLA system. The likelihood ratios referring to the H' versus H' or H 2 versus H I comparisons are not significant (0.5 and 0.7, respectively), indicating that the estimate of recombination rate (0.36 for H!) does not differ from 0.5.
For bulbo-urethral gland size, the mixed inheritance model in which the complete dominance of the 'small size' allele is assumed (hypothesis H!) is also the most acceptable, but the likelihood ratio referring to the model in which no restriction is imposed on the dominance parameter for the major gene (H i versus H' comparison) is marginally significant, at the P < 0.10 level (3.4 Fat androstenone levels are determined by the balance between the intensities of testicular androstenone synthesis and androstenone degradation (Bonneau et al, 1982). The intensity of androstenone biosynthesis depends on sexual maturity and on the activity of andien-b-synthase, the enzyme system leading to the C19D16 steroids biosynthesis pathway (Gower, 1972). It may be speculated that there is evidence for a major gene in the present study, which could regulate the activity of andien-b-synthase, whereas the remaining polygenic variation could correspond to differences between animals in their capacity to catabolize androstenone. The existence of such differences has been postulated from previous results (Claus, 1976;Bonneau et al, 1982). Alternatively, the fairly close genetic correlation between fat androstenone level and bulbo-urethral gland size could result from the presence of a unique major gene influencing androstenone synthesis through sexual maturity. Considering the predicted genotypes of individuals at the two postulated loci would suggest that the respective genotypes do not agree in most cases, but the number of individuals whose genotypes at both major loci are known with a reasonably high probability (more than 0.80) is quite limited. More generally, the poor accuracy of the prediction of individual genotypes at each postulated major locus does not make it possible to verify the coherence between parent and offspring genotypes nor to compare the time trends of allelic frequencies in control and selected lines. Only a specific experiment relying on planned matings between animals of well-determined genotype would permit the confirmation of the results of this study. The finding of a major gene affecting fat androstenone level gives rise to new prospects for reducing incidence of boar taint by means of breeding. The average level of individuals carrying the 'low androstenone' allele at the postulated major locus is indeed well below the critical threshold value of 1 ppm. Avoiding boar taint in a pig population would hopefully be obtained by fixing this allele. However, the expected consequences on sexual maturity status of boars remain uncertain as the possible identity of the two major loci detected independently for each trait could not be accurately tested at this stage of the study.

ACKNOWLEDGMENTS
Thanks are expressed to F Bariteau and J Bussière (Station exp6rimentale d'ins6mination artificielle, RouiII6, France) for their efficient contribution to the achievement of this selection experiment. The expert technical assistance of J Portanguen and F Giovanni, who performed the androstenone determinations, is gratefully acknowledged. This experiment was supported by grants from AIP 'Constitution de lign6es g6n6tiquement int6ressantes' and from INRA 'Direction Scientifique des Productions Animales'.