Genetic Linkage In

In the present study an extensive amount of data, comprising more than 30,000 offspring in total, was analyzed to evaluate the influence of age and sex on the recombination frequency in the K-PGD segment of the equine linkage group (LG) I and the influence of age, breed and sex on recombination in the ALES segment of LG 11. A highly significant sex difference is reported for both segments. Male and female recombination values in the K-PGD segment were estimated at 25.8 f 0.8 and 33.3 f 2.5%, respectively. Similarly, recombination was less frequent in the male (36.6 0.7%) than in the female (46.6 & 1.2%) in the AI-Es segment. Comparison of data from two Swedish horse breeds revealed no significant breed differences in either sex for recombination in the Al-Es segment. N o evidence of an age effect was found in any segment or sex. The distribution of individual male recombination estimates was also investigated, and a significant heterogeneity among stallions was revealed in the K-PGD segment. The results are discussed in relation to previous studies on factors affecting recombination in mammals. ENETIC recombination is known to be affected by a number of environG mental and genetic factors (cf. MORGAN, BRIDGES and STURTEVANT 1925; SIMCHEN and STAMBERG 1969). The effect of a given factor may involve the entire genome or be limited to certain chromosomal segments. Most of the knowledge on this subject has been derived from experimental organisms such as Drosophila and Neurospora. In mammals, information on factors influencing recombination is quite limited. Although the biochemical and genetic basis for recombination is likely to be similar among all eukaryotes, we expect the influence of environmental agents (e.g., temperature) to be less important in mammals than in lower eukaryotes due to the higher degree of homeostasis in the former group. Increased knowledge on the most important factors affecting recombination and on the distribution of recombination fractions within populations is of great importance for genetic studies in general and for gene mapping work in particular. Sex differences in the frequency of recombination in animals are well documented in a number of species, including man and mouse (DUNN and BENNETT Genetics 106 109-122 January, 1984. 110 L. ANDERSON AND K. SANDBERG 1967; WEITKAMP 1973; CALLAN and PERRY 1977). There is a general tendency for recombination to be less frequent in the heterogametic sex; this effect is extreme in Drosophila males (XY) and silk-worm females (20) in which no recombination is detected. The most extensive information on sex differences in recombination in a mammalian species is provided by data from the mouse; DUNN and BENNETT (1 967) tabulated male and female recombination estimates for 53 linkage pairs in this species. For 30 pairs there was no significant difference between sexes, whereas a significantly higher recombination estimate was found for females in 18 cases and for males in five cases. The pairs with a higher female rate were distributed among ten linkage groups. Four of the pairs with a higher male rate were found in one linkage group. Thus, in the mouse there appears to be a region-specific effect of sex on the frequency of recombination. Although based on more limited data, the same conclusion seems to be justified in man (WEITKAMP 1973, 1976). BRIDGES reported as early as 19 15 that crossing over in Drosophila varied with the age of the mother. The effect of parental age on recombination appears to be less obvious in mammalian species. WALLACE, MACSWINEY and EDWARDS (1976) examined data covering 18 segments over eight chromosomes in the mouse and found no consistent age-related trend in either sex when two-point recombination and multiple-point interference ratios were investigated. From their own and previously published work in mice (FISHER 1949; BODMER 1961; REID and PARSONS 1963) they concluded that the number of significant agerelated changes observed is small compared with the number of cases in which no significant age effect has been detected. The results suggest that age effects are not common and, if they exist, may be specific to chromosome region, sex and strain. In man, the very limited data published are compatible with these conclusions, because indications of age-related changes have been found in some studies (WEITKAMP 1973) but not in others (RENWICK and SCHULZE 1965; WEITKAMP et al. 1973; ELSTON, LANCE and NAMBOODIRI 1976). Most studies on this subject have been based on an assumption of a linear trend with age, but the possibility of a curvilinear relationship has been suggested on the basis of recombination data from the mouse (WALLACE 1957) and also by age trends in chiasma frequencies in various species (MAYO 1974). Genetic variation with respect to the frequency of recombination has been disclosed in Drosophila and other experimental organisms, because it has been found possible to alter the recombination frequency by artificial selection [see MAYNARD SMITH (1978) and TUCIC, AYALA and MARINKOVIC (1981) for review]. Genetic variation for recombination has also been postulated to be of common occurrence in natural populations (SIMCHEN and STAMBERG 1969; MAYNARD SMITH 1978). The presence of genes controlling recombination may be reflected in population data as a recombination difference ( 1 ) between individuals, (2) between groups of related individuals or (3) between populations or races. Such genes may also be revealed as a difference in recombination associated with a specific marker allele; the specific allele effect may be due to the expression of the allele itself or to a chromosomal rearrangement (e.g., inversion) or some other genetic factor affecting recombination in gametic association with the marker allele. Detecting genetic variation for recombination GENETIC LINKAGE IN THE HORSE 111 from population data is evidently a laborious enterprise and it is not surprising that existing evidence from mammalian species is very sparse. In man, data clearly indicating differences in recombination between populations of blacks and whites from two regions on chromosome 1 and one region on chromosome 6 have been reported (WEITKAMP 1974, 1976). GEDDE-DAHL et al. (1 975, 198 1) found an allele-specific difference of the recombination frequency between the Pi and Gm loci in man. The accumulation of data from the routine blood-typing service in our laboratory has created a possibility for an extensive study on genetic linkage in the horse. The first paper in this ongoing study examined the linkage relationships among 15 blood marker loci (SANDBERG and ANDERSSON 1984). Linkage of the loci for serum albumin (AZ) and serum esterase (Es) was demonstrated, and additional data on the previously described linkage (SANDBERG 1974a) between the loci for blood group K (K) and 6-phosphogluconate dehydrogenase (PGD) were reported; the K-PGD linkage and the AZ-Es linkage belongs to equine linkage groups I and 11, respectively. This second paper reports on the influence of age, breed and sex on recombination in the AI-Es segment and on the influence of age and sex on recombination in the K-PGD segment. The distribution of individual male recombination estimates for these two segments is also analyzed to provide information on the within-population variability of recombination frequencies. The latter analysis is of particular interest for the K-PGD segment in which a significant heterogeneity among stallions was previously detected (SANDBERG and ANDERSON 1984). MATERIALS AND METHODS Aniwcils: The horses used in the present study belonged to the Swedish Trotter (ST) breed and the North-Swedish Trotter (NST) breed. The material consisted of all horses blood typed in our laboratory in the period 1970-1979. Altogether 28,652 and 5617 complete families (sire, dam and offspring tested) were available for the ST and NST breed, respectively. The age of parents and offspring was obtained from registration records. Genetic markers: The Al, Es and PGD loci all determine simple electrophoretic systems involving multiple codominant alleles. The blood group system K is a simple one-factor (Ka), two-allele system. The methods for the analysis of these genetic markers are given by SANDBERG and ANDERSON (1984). LinRage analysis: Paternal and maternal half-sib groups were examined. The genetic contribution of a heterozygous parent with regard to the Al, E5 and PGD loci could be determined in all matings except those in which sire, dam and offspring had the same genotype. However, double intercrosses in the ALES combination were excluded when sire and dam had the same heterozygous genotype at both loci. On examination of half-sib groups, such matings may introduce a dilocus segregation bias for linked loci (ANDERSON 1983). With regard to blood group locus K, only matings between parents heterozygous for the blood group factor Ka and parents lacking the factor were included. The heterozygosity of a dam or sire was inferred from the occurrence of at least one offspring lacking the factor. When the family size is small, this selection of parents leads to a segregation bias. The data on recombination in the K-PGD segment were, nevertheless, unbiased, because parents were selected through the offspring with regard to only one of the characters under study (6 MORTON 1955). The linkage phase of stallions was inferred from the segregation data. All stallions with at least 50 informative offspring (i .e., offspring for which the gametic contribution of the sire could be determined) for the Al-Es segment and all stallions with at least 15 informative offspring for the K-PGD segment were used. These numbers were chosen on the basis of previously reported recombination 112 L. ANDERSSON AND K. SANDBERG estimates (SANDBERG and ANDERSON 1984) and represent the limits at which the linkage phase of a stallion could be determined with about 95% confidence. This method of selecting families may bias the recombination estimates slightly in the direction of close linkage because the wrong linkage phase will be assumed for those stallions for which the number of recombinants by chance is greater than the number of nonrecombinants. This possible bias is negligible in the present study because for the majority of stallions the number of offspring by far exceeded the limits set to determine the linkage phase with 95% confidence. The data on female recombination were based on all mares for which the linkage phase could be determined from the genotype of their parents. After the linkage phase of the parents had been determined, each informative offspring could be classified as recombinant or nonrecombinant. The possible occurrence of double or multiple crossovers in the material should be noted. Statistical analysis: Heterogeneity of recombination fractions in any comparison was tested by conventional contingency chi-square analyses. Curvilinear regression was applied to examine the influence of age on recombination, as we had no a priori reason to expect a linear relationship. For this analysis, recombination fractions were transformed using the angular transformation and weighted by the sample size (HARVEY 1982). The curvilinear regression was performed as described by SOKAL and ROHLF (198 1 , p. 67 1) in a stepwise manner by fitting consecutively first a linear regression to the data and then adding higher order terms, the procedure initially being continued up to adding the cubic power of age. In the final analysis, higher order terms were dropped if their effect was insignificant compared with the effect of lower order terms. The General Linear Model procedure of the Statistical Analysis System (GOODNIGHT, SALL and SARLE 1982) was utilized for the calculations. The possible effect of stallion and age on recombination was analyzed simultaneously using a loglinear model as described by BISHOP, FIENBERG and HOLLAND (1975). This method is designed to be used for tests of interactions in multiway tables of attribute data; age was treated as a qualitative variable in the analysis. In the present study it was of interest to test for three-factor interactions between stallion, age and recombination and to test for two-factor interactions between stallion and recombination, and between age and recombination. The test for two-factor interactions was made within each level of the third factor, i t . , conditional independence was tested for. The computer program BMDP4F (BROWN 1981) was utilized for the calculations of the G-statistic (log likelihood ratio test) which is distributed as approximately chi-square.