A genetic analysis of variable number of tandem repeats (VNTR) polymorphism in the horse

Summary - Restriction fragment length polymorphism detected with the 33.6 minisatel-lite probe (Jeffreys et al, 1985) was analysed in 3 horse families of paternal half-sibs and in a sample of stallions from 4 breeds. Among the bands detected on Hae III genomic DNA digests, it was found that strongly hybridizing fragments behaved as alleles at 2 different loci. These 2 loci, showing 7 and 3 detectable alleles, were not closely linked to each other nor to informative blood groups and protein markers. No neo-mutation in allele length was observed at these 2 loci ip the 3 families. In the stallion sample, 8 alleles were detected at the first locus and no differences were found between their frequencies in the 4 breeds. Heterozygosity and polymorphism information content (PIC) values estimated in the 4 breeds show that the Thoroughbred is the least variable breed and the Arab the most. In the whole population the PIC values were 0.73 and 0.70 respectively.


INTRODUCTION
It is now well established that the genome of a number of animal and plant species contains regions that are highly variable, among which are the minisatellites. These regions are detected by probes from a different origin containing sequences several base-pairs long, tandemly repeated. Initially, highly polymorphic patterns displaying a large number of fragments on Southern blots resulting from variation in the number of tandem repeats and called fingerprints were obtained in man by Jeffreys et al (1985). These authors used a repeat found in the second intron of the myoglobin gene to design a probe that revealed related human sequences, some of which were themselves later used as probes. This was the case for the 33.6 and 33.15 regions that revealed extensive polymorphism in a number of organisms. Other probes containing repeated sequences such as a sequence from the M13 bacteriophage (Vassart et al, 1987), a sequence homologous to the Drosophila Per gene (Shin et al, 1985), in the 3' region of the a globin gene (Jarman et al, 1986) and a number of others, cloned from the human genome (see for example Nakamura et al, 1987), were found to detect very high levels of polymorphism.
In pets and domestic farm animals, polymorphisms were found in dogs and cats (Jeffreys and Morton 1987), in cattle (Georges et al, 1988), in poultry (Hillel et al, 1989), in sheep (Drinkwater et al, 1990) and in pigs (Coppieters et al, 1990) using either human, viral or homologous probes.
In horses, restriction fragment length polymorphisms (RFLP) have been detected using different probes such as M13, Jeffreys' minisatellites or core sequence, Per, human VNTR, (TG)n, a homologous horse probe and synthetic tandem repeats (Georges et al, 1988;Massina et al, 1989;Troyer et al, 1989;Bernoco and Byrns, 1991;Broad et al, 1991;Hopkins et al, 1991;Ellegren et al, 1992;Giulotto et al, 1992;Mariat et al, 1992). However, very little or no information about segregation of the fragments and linkage analysis has been reported as limited families were typed.
In this report, we show the segregation of DNA fragments hybridized with a human minisatellite probe in 3 large horse pedigrees and give some genetic characteristics of the regions detected in 4 horse breeds.

Animals
Three horse families consisting of a stallion and half or full paternal sibs, along with their mothers, were collected. The number of mother (74)offspring (87) combinations were 30, 27 and 30 for the 3 stallions, Iris Landais, Artichaut and Matador du Bois, respectively. All animals were of Selle Fran!ais origin. A sample of stallions from 4 breeds, Arabs (AR, N = 47), Thoroughbred (TB, N = 43), Selle Fran!ais (SF, N = 40) and Trotteur Fran!ais (TF, N = 43) were also obtained from different state stud-farms. All animals were blood typed by the Laboratoire des Groupes Sanguins des Chevaux, INRA-CRJ, Jouy-en-Josas, for the blood group systems (A, C, D, K, P, Q, U) and the protein systems (Alb, Tf, A1B, Es, Pi, Gpi, Pgd, Pgm) and all offspring qualified.
Purification of DNA Blood samples were drawn in tubes containing tripotassium EDTA (Vacutainer) and DNA was extracted according to the method of Jeanpierre (1987). Blood was treated twice in a solution of 10 mM NaCI, 10 mM EDTA, pH 8, for 30 min at +4°C. Pellets containing leucocytes were homogenized in a solution of 5 M guanidine HCI, 0.5 M ammonium acetate, 1.25% N-lauroylsarcosine containing 0.1 mg/ml proteinase K for 1 h at 60°C. DNA was then precipitated with 2 vol ethanol and washed twice with 70% ethanol.

Restriction endonuclease digestion and electrophoresis
Seven pg samples of horse DNA were digested for 20 h at 37°C by restriction endonucleases HaeIII or HinfI (BRL or Boehringer) according to the manufacturer's recommendations. Complete digestion was achieved with a total addition of 5 units per Ag DNA. Restriction endonuclease DNA fragments were subjected to horizontal electrophoresis in 1% agarose gels 25 cm in length in Tris-acetate buffer (40 mM Tris, 20 mM AcOH, 2 mM EDTA; pH 8.3). Electrophoresis was performed with circulating buffer at 55 V (HaeIII) or 60 V (Hinfl) for 18 h. Calibration was achieved by adding a sample of BstEII-KpnI-SmaI digested lambda phage DNA.

Transfer
Transfer was performed on Biodyne B membrane (Pall Industry, SA). The DNA digested fragments in the gels were depurinated in 0.15 N HCI for 10 min, denatured in a 0.4 N NaOH solution for 30 min and transferred using the method of Southern (1975) at room temperature for 18 h with 0.4 N NaOH.

Hybridization probes
The 2 human VNTR probes, 33.6 and 33.15 were kindly provided by AJ Jeffreys. They were labelled by the primer extension method according to Jeffreys (1985), except that the radiolabelled insert was not extracted from the plasmid.

Prehybridization and hybridization
Prehybridization was carried out for at least 4 h at 65°C in a mixture containing except that 0.05% sodium pyrophosphate was added and sufficient radiolabelled probe to provide !! 1 x 10 6 cpm/ml. Membranes were then washed twice for 15 min at room temperature in 2 x SSC, 0.1% SDS then in the same solution for 30 min at 65°C (in more stringent solution at the same temperature if necessary) and finally rinsed in 2 x SSC at room temperature. The restriction endonuclease fragments were vizualized by autoradiography in a cassette with a Kodak XAR film. and Cronex LI-PLUS intensifying screens at -70°C.

Linkage analysas
Calculations were perforined according to the Lod score method of Morton (1955).

Probes and enzymes
DNA extracted from our group of horses was digested with HaeIII and with Hinfl. Whatever the enzyme used, the 33.15 probe gave a large number of fragments with rather homogeneous intensities and no clear polymorphism (results not shown).
Conversely, the 33.6 probe detected fewer bands with different hybridization signal intensities. With Hinfl, all strongly hybridizing bands were detected in the range of small fragment lengths (less than 3.5 kb) (fig la). HaellI gave more evenly distributed bands spread over the autoradiograms (fig lb). Since, with DNA digested with Hinfl, the different levels of band migration appeared to be more heterogeneous than with HaeIII, it was decided to take into account only results obtained with the 33.6-HaeIII probe-enzyme combination.

Fragment identification and polymorphism
Blots usually consisted of DNA digests from 1 sire and its offspring along with DNA digests from their mothers. The 3 sires were also run on the same gel to compare the respective mobilities of their bands (fig 2). Two major groups of bands with a strong radioactive signal appeared in the range of medium (2.5 to 5 kb) and small fragment length inferior to 2 kb (fig 2) on HaeIII blots probed with 33.6. A total of 15 fragments were given a letter as an identification code (A to 0) according to their mobility and intensity while the rest of the bands giving low signals were not included in our interpretation. All strongly hybridizing fragments were found to be polymorphic.

Segregation of the fragments
Transmission of single paternal fragments was followed in the 3 sire progenies. Only informative dam(s)-offspring combinations, that is dam negative for each fragment, were taken into account. In this way, the presence or absence of a band in the offspring could be unambiguously interpreted. It was observed that a given fragment present in the sire was transmitted to half of his progeny (table I) suggesting a Mendelian inheritance of the corresponding allele.

Locus identification
Two approaches were consecutively used to determine the number of loci controlling the fragment polymorphism. One is based on the analysis of the relative segregation of paternal fragments in their progeny, the other considers that, in the same animal, any fragment should not triplicate with others already assigned to a locus. In the medium size range of fragments (2.5 to 5 kb), sires exhibited 2 bands that were found to behave as alleles at one locus. As no dam or offspring possesses more than 2 of these bands, it was concluded that these fragments were alleles at a single locus. Six alleles present in the sires were thus identified (B, C, D, E, F, I). In addition, one fragment (H) which was only present in some mares and part of their offspring, but did not triplicate with the 6 previously defined alleles, was also assigned to this first locus. The same arguments were employed for the smaller fragments which were shown to segregate as 3 alleles M, N and 0 at a second locus but the existence of negative animals in this band category suggests that there are other fragments generated by this locus not detected in our conditions. Some other fragments were not allocated to any specific other loci mainly because insufficient genetic information could be obtained (bands A, G, J, K and L).

Mutation rate in allele length
Each fragment belonging to 1 of 2 loci present in an offspring was also present in one of its parents which strongly suggests that no mutation or mutation-like event had occurred in these 3 families. The upper limit of the 5% confidence interval of the mutation rate based on the Poisson distribution is thus 1.8% at the first locus and 4.4% at the second locus according to the number of observed fragments transmitted to the progeny (168 and 66 respectively).

Linkage analysis
No close linkage was found between these 2 loci and between them and any of the other markers that were heterozygous in at least 1 of the 3 sires (A, D, P, Q and U blood groups and Al, Tf, Es, A1B, Pi, Pgd and Pgm for the proteins). No Lod scores exceeded 1.0 for recombination values ranging between 0.01 and 0.5.

Polymorphism in the 4 breeds
As for the families, all strongly hybridizing fragments were polymorphic. The F fragment, detected in families, was split into Fl and F2 fragments that were considered to be alleles. Conversely, H and I were difficult to distinguish as these 2 fragments are not frequent and migrate very close to each other and thus were pooled into a single fragment (H). No interpretation is given for the second locus because band intensities were sometimes variable and could become a source of error. The 4 populations were found to be in Hardy-Weinberg equilibrium based on the albumin locus (mean value X 2 = 1.28, df = 3). Allelic frequencies for the first minisatellite locus are given in table II and no differences were detected between the populations ( X 2 = 0.94, df = 21). Heterozygosities and polymorphism information content (PIC) values (Botstein et al, 1980) are given in table III.

DISCUSSION
In an initial screening, horse genomic DNA was digested with HaeIII or Hinfl and Southern blots were subjected to hybridization with either the 33.6 or 33.15 probe. It appeared that the 33.15 probe gave about 25 hybridizing bands per individual with a low degree of polymorphism and some background signal. On the contrary, 33.6, in the same experimental conditions allowed the detection of a smaller number of more distinct bands per individual. Comparison of horse DNA digested with HaeIII or Hinfl and probed with 33.6 revealed that the 2 enzymes detected roughly similar banding patterns with a greater number of fragment migration levels with HaeIII than with HinfI. For example in figure 1, the HaeIII-B fragment corresponds exactly to the Hinfl-C fragment while the HaeIII-E is included in the Hinfl-J, which means that the J fragment is heterogeneous (the second locus was not observable on Hinfl DNA digests). This observation confirms the prediction that the polymorphism detected with VNTR originates from the number of tandem repeats and not from mutation in the DNA restriction site of the enzyme. The banding pattern of the 33.6-HaeIII probe-enzyme combination resembles more a single-locus VNTR situation than a fingerprint and shows that at least 2 loci are predominantly recognized due to a great similarity between the horse minisatellite core sequence and the human probe sequence. On the contrary, the 33.15 probe detects a family of DNA fragments, the sequences of which are more similar to each other and are most probably generated by a greater number of loci.
The 33.6-HaeIII combination allowed an analysis of the segregation of strongly hybridizing fragments in sire progenies, an approach which has proved to be very tentative or even impossible with true fingerprints. Two loci were identified with 7 and 3 detectable alleles in these 3 families. No close linkage was detected between any of the red cell antigen, protein and DNA loci detected in this study and no mutation event could be observed which indicates that these loci are rather stable.
In the stallion populations, 8 alleles were detected at the first locus whose frequencies did not significantly differ between the 4 breeds. However, this observation has to be confirmed on a larger number of animals. On the other hand, some bands at this locus migrated very close to each other and we have certainly underestimated the real level of polymorphism of this system. It can be observed that the lowest PIC value is found in the Thoroughbred, which confirms the well-known homogeneity of this breed (Bowling and Clark 1985;Gu6rin and M6riaux, 1986;Ellegren et al, 1992). Conversely, the Arab sample is the most heterogeneous of the 4 breeds. At the second locus in the population sample, identification of alleles was rendered somewhat hazardous due to the low power of resolution in this part of the gel. Bands were found to be less distinct with variable intensities and levels of migration heterogeneous. These characteristics were much less important in the families probably because animals were interrelated, thus reducing the variability observed in unrelated animals of the same breed.
In spite of some limits imposed by the resolution of the technique, the high polymorphism detected with the 33.6 probe makes it a very useful marker for horse identification and linkage analysis in this species.