Intragenic haplotypes at the bovine CSN1S1 locus

of , alternative for the differentiation of CSN1S1*B and C the alleles and c . Genotyping of belonging to and frequency of the alleles. CSN1S1*B occurred in all breeds with frequencies varying from 0.50 in Blackup to 1.0 in e.g. Ayrshire. CSN1S1*b on the other from 0.63 in Jersey, 0.97 in Ayrshire to in in the mutations do not occur together. the four intragenic haplotypes ( B-b, B-c, C-c and C-b) is predominant in with frequencies varying from in to in Angler and Scottish Highland, two in Ayrshire, three in Asturian Valley and Turkish Grey Steppe to all four in the other 12 breeds. Correlation between allele frequencies and the geographic origin of the breeds was significant for the Mae III promoter polymorphism.


Introduction
Variants of different milk protein genes in cattle are discussed in the context of studies regarding quantitative and qualitative traits and are used within evolutionary and diversity studies. For αs 1 -casein (CSN1S1) 9 alleles (A, B, C, D, E Yak , E Bali , F, G, H) have been described within different cattle breeds, with CSN1S1*B being the predominant allele in Bos taurus and CSN1S1*C in Bos indicus and Bos grunniens breeds (EIGEL et al., 1984;FORMAGGIONI et al., 1999). The other alleles are rare in other breeds that have been studied. It has been postulated that CSN1S1*B has positive effects on milk yield (LIN et al., 1986), whereas higher milk protein content is found in animal heterozygous for CSN1S1*BC compared to BB homozygous animals (NG- KWAI-HANG et al., 1990;ALEANDRI et al., 1990, BOVENHUIS et al., 1992. In Nordic cattle breeds LIEN et al. (1999) identified an allele frequency gradient with low frequency of CSN1S1*B in native breeds to high frequency of CSN1S1*B and loss of CSN1S1*C in high selected dairy cattle. This is in agreement with the reported very high frequencies of CSN1S1*B in breeds selected for milk production, up to fixation or nearly-fixation in Ayrshire, Angler, or Holstein Friesian (ERHARDT, 1993;IKONEN et al., 1996). Development of several DNA tests for genotyping milk protein genes offered the possibility to type animals independent from age, sex, and lactation (LÉVEZIEL et al., 1988). SCHLEE & ROTTMANN (1992) and DAVID & DEUTCH (1992) developed allele specific PCR tests (ASPCR) for differentiation of CSN1S1*B and C, while LIEN et al. (1993) used an amplification created restriction site (ACRS) for HphI to discriminate between CSN1S1*B and CSN1S1*C. The latter test detects the causal nucleotide substitution inside exon 17 of the gene. KOCZAN et al. (1993) described a polymorphism in the promoter of the CSN1S1 gene affecting a MaeIII restriction site. The resulting PCR-RFLP test was suggested as an alternative to the ASPCR test for the differentiation of CSN1S1*B and C. However, TURECKOVÁ et al. (2001) recently reported that in Czech and German Red Cattle the MaeIII promoter polymorphism is not always linked to the HphI ACRS in exon 17. Single strand conformation polymorphism (SSCP) analysis is a rapid and sensitive screening technique, that allows mutation identification without restriction enzyme digests or special primers (ORITA et al., 1989). PCR-SSCP analysis for milk protein genotyping identified new alleles in both -endangered and production breeds -which were subsequently characterized by DNA sequencing (PRINZENBERG et al., 1999;CAROLI et al., 2001). The aim of this study was to determine the occurrence of the two CSN1S1 polymorphisms described in different cattle breeds and to define intragenic haplotypes as genetic markers using a PCR-SSCP based DNA test for typing the causal nucleotide substitution in exon 17.

DNA test (PCR-RFLP) for CSN1S1 promoter polymorphism:
A 310 bp fragment containing the first 274 bp of the promoter and parts of exon 1 was amplified, digested with MaeIII and separated on an agarose gel as described by KOCZAN et al. (1993). Uncut fragments of the 310 bp fragment (indicating G in nucleotide position 1957) were named c, fragments cut by MaeIII (indicating A in nucleotide position 1957) were named b. PCR-SSCP test for CSN1S1*B and C: For differentiation of the CSN1S1 alleles B and C a SSCP-based DNA test was developed using 58 DNA samples from animals with known CSN1S1 genotypes as standard samples. Genomic DNA was amplified by PCR to give a 223 bp-fragment containing exon 17 of the CSN1S1 gene (Position 17644-17867) of GenBank Acc. No. 59856). PCR was in a final volume of 25 µl, containing 100 ng genomic DNA, 15 pmol of each primer (CSN1S1-5: 5` CAC TGT TGC TTT TTC AAT GGT C 3` CSN1S1-3: 5` AAG GCA ACA ATA TGC AGT CAT TT 3`), 1 U Taq-polymerase (Peqlab Biotechnologie GmbH, Erlangen), 200 µM dNTP, 1.5 mM MgCl 2 , 10 mM Tris-HCl pH 8.8, 50 mM KCl with an initial denaturation step at 94°C for 5 min, followed by 30 cycles with a denaturation at 94°C for 1 min, annealing of 56°C for 1 min, elongation of 72°C for 1 min and a final elongation of 72°C for 5 min. Four microlitres of the PCR product were mixed with 6 µl formamide buffer (95% formamide, 0.025% bromphenolblue, 0.025% xylenecyanol FF, 20 mM EDTA), denatured at 91°C for 3 min and immediately chilled on ice. Samples were run 3 h at 200 V at 10°C on a 10% acrylamide:bisacrylamide gel (37:1) with 2% glycerol. DNA fragments were visualised by silver staining (BASSAM et al., 1991). Haplotype definition: Haplotypes were determined based on the results of PCR-RFLP analyses determining nucleotide position (nt) 1957 (b=nt1957: A, c=nt1957: G) and exon 17 PCR-SSCP discriminating alleles B and C (Table 1).  Data analysis: Allele frequencies, observed and expected genotype frequencies, and deviations from Hardy-Weinberg equilibrium were evaluated by GENEPOP Software (RAYMOND and ROUSSET, 1995). Expected and observed haplotype frequencies were calculated and compared (χ 2 -values) with EH software (XIE and OTT, 1993). Correlations between degree of latitude and allele frequencies and linear regressions of allele frequencies on geographic latitude were calculated.
Results PCR-SSCP analysis for CSN1S1 exon 17 showed two distinct fragment patterns for alleles B and C which was in agreement for all 58 DNA samples of known genotypes (Figure 1).

Genotype frequencies
The observed and expected genotype frequencies and the probability-values for Hardy-Weinberg equilibrium are shown in Table 2. All populations were in Hardy-Weinberg equilibrium at CSN1S1 locus for position 17807 and position 1957 except Pezzata Rossa and Piemontese for the first locus, where an excess of homozygous genotypes was observed. All of the expected six genotypes at the CSN1S1 locus were observed in eight of the 17 breeds analysed while in Scottish Highland and Angler only two genotypes were found.  Table 3 gives the allele and haplotype frequencies at the CSN1S1 locus and shows differences both in the occurrence and the frequencies of the alleles and the haplotypes in the cattle breeds studied. CSN1S1*B occurred in all breeds with frequencies varying from 0.50 in Anatolian Black to 1.0 in Ayrshire, Scottish Highland, and Angler. CSN1S1*b on the other hand varied from 0.63 in Jersey, 0.97 in Ayrshire to 1.0 in Angler and Scottish Highland. In most breeds frequencies for CSN1S1*b are higher than for CSN1S1*B. In contrast, in Aberdeen Angus, Ayrshire, Casta Navarra, Chianina, Piemontese, and Turkish Grey Steppe CSN1S1*b occurs in lower frequencies than B.  Table 3 Allele frequencies, expected, and observed haplotype frequencies of CSN1S1 in European cattle populations and corresponding χ 2 -value (Allelfrequenzen, erwartete und beobachtete Haplotypenfrequenzen von CSN1S1 in europäischen Rinderpopulationen und entsprechende χ 2 -Werte) (χ2-limit for 1% significance level is 11.34 and for 5% level is 7.81. In breeds marked with "-" no χ2-analysis has been done due to fixation of one or two of the loci.)

Correlations and regression analyses with geographic data
Frequencies for CSN1S1*B and b could be shown to be increasing from southern sampling area (Turkey) to northern Europe (Scotland). Correlation of geographic latitude in the range of the sampling area (37° -58°N) with CSN1S1*B allele frequencies was r=0.417 (n.s.), and r=0.556 (p<0.05) for CSN1S1*b. Regression analysis resulted in a regression equitation of y=0.00843*x+0.45308 for CSN1S1*B and of y=0.01103*x+0.31914 for CSN1S1*b allele frequencies and geographic latitude ( Figure 2).

Discussion
The PCR-SSCP test developed for the differentiation of CSN1S1*B and C offers a rapid and cost-effective alternative to the ASPCR described by SCHLEE and ROTTMANN (1992) and DAVID and DEUTCH (1992) and the ACRS method described by LIEN et al. (1993). Our genotyping results show, in agreement with TURECKOVÀ et al. (2001), that the promoter polymorphism described by KOCZAN et al. (1993) can not be used as a reliable genotyping method to infer CSN1S1*B and C-genotypes in all breeds. The small number of animals analysed by KOCZAN et al. (1993) belonged to Jersey, Holstein Friesian and German Simmental. Aberdeen Angus, Ayrshire, British Friesian, Charolais, Turkish Grey Steppe, Pezzata Rossa, and Piemontese show low frequencies for B-c and C-b haplotypes, while frequencies of same haplotypes are high in Chianina and Anatolian Black. Our results support the occurrence of different intragenic haplotypes in milk protein genes, that might contribute to variation of different milk production traits (SCHILD and GELDERMANN 1996;EHRMANN et al., 1997). Expected and observed genotype frequencies at CSN1S1 nt 17807 in Piemontese and Pezzata Rossa were not in Hardy-Weinberg equilibrium. Both breeds showed much lower frequencies of heterozygous animals than expected from the allele frequencies.
This may be a sampling artefact, however at position nt 1957 all breeds were in Hardy-Weinberg equilibrium. Despite of close genetic linkage χ 2 -test for linkage disequilibrium does not show significant linkage in all populations. Linkage disequilibrium declines with increasing generations (FALCONER, 1984), so the level and extent of disequilibrium diminish in older populations. The British breed Ayrshire is a long established breed, as is the Hereford. Casta Navarra and Fighting Bull have also been maintained over many centuries. Maremmana represents a low-selected historic genotype, and Chianina is regarded to be the oldest Italian breed. Anatolian Black and Turkish Grey Steppe are very heterogenous breeds that remained without specific selection pressure over centuries (PORTER, 1991). Thus the limited linkage disequilibrium in these breeds points to the two mutations being very old. On the other hand Pezzata Rossa is a newly founded population (herdbook established in 1957) originating in crosses of Simmental with Friulana cattle. Simmental was planned to substitute Friulana by backcrossing, however its introgression stopped after a few generations. Haplotypes derived from both parental breeds are likely to be still present in the Pezzata Rossa and may cause the lack of linkage disequilibrium observed. Frequencies of CSN1S1*B are in tendency higher and significantly higher for CSN1S1*b in northern than in southern European breeds, with highest values in dairy breeds of north western European origin. LIEN et al. (1999) reported an apparently contrary frequency gradient in Nordic breeds with high frequencies of CSN1S1*C in autochtone breeds of northern Scandinavia and lower frequencies to fixation in dairy breeds originated in southern Scandinavia. This indicates rather a selection gradient than a geographic gradient, supporting the suggestion made by LIN et al. (1986) that occurrence of CSN1S1*B variant is correlated with selection for improvement in milk production traits. Our study analysed northern European cattle breeds, that include a large proportion of dairy breeds, selection pressure thus is expected to lead to increasing fixation of alleles linked to production traits. Allele frequencies of CSN1S1 show that in Anatolian Black both alleles B and C occur in equal frequency. LOFTUS et al. (1999) described a high admixture proportion of Bos indicus with 30.6% in Anatolian Black and BAKER & MANWELL (1980) pointed out that CSN1S1*C occurs in very high frequencies in zebu cattle. Therefore the high frequency of CSN1S1*C in Anatolian Black may have occurred by introgression of zebu genes. However loss of haplotypes along a south-north gradient leading to fixation or nearly fixation of the CSN1S1 B-b haplotype in northern European cattle populations could be explained by drift. The loss of genetic diversity along a south-north gradient has already been described for a number of different loci. This may reflect distance from the center of domestication of cattle (MEDJUGORAC et al., 1994;TROY et al., 2001).