| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* INRA, UR339 Laboratoire de Génétique Biochimique et Cytogénétique, F-78352 Jouy-en-Josas, France
Union Nationale des Coopératives agricoles d’Elevage et d’Insémination Animale, F-75595 Paris, France
INRA, UR337 Station de Génétique Quantitative et Appliquée, F-78350 Jouy-en-Josas, France
1 Corresponding author: mathieu.gautier{at}jouy.inra.fr
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Key Words: quantitative trait locus • dairy trait • acyl-CoA:diacylglycerol acyltransferase gene (DGAT1) • variable number of tandem repeats
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Other studies have shown that the 2 alleles at the K232A polymorphism segregate in several breeds from different countries (Kaupe et al., 2004). The K232A mutation probably occurred early in the history of cattle domestication or even before domestication (Winter et al., 2002; Kaupe et al., 2004). Nevertheless, the estimated frequency of the wild-type lysine variant (the K allele) varies greatly according to the population considered, which might reflect different breeding objectives regarding milk composition in different countries and breeds. The effect of the mutation on milk production traits has been extensively characterized and the K allele is always associated with an increase in fat content, fat yield, and protein content, and a decrease in protein and milk yields.
Evidence for the existence of additional mutations contributing to the observed QTL effect has been reported recently (Bennewitz et al., 2004). In particular, one allele at a variable number of tandem repeat (VNTR) locus located in the 5' noncoding region of DGAT1 has been proposed as a putative causative variant (Kuhn et al., 2004).
In this paper, we report the characterization of both the VNTR and the K232A polymorphisms in the DGAT1 gene in the 3 main French dairy cattle breeds (Holstein, Normande, and Montbéliarde). We also investigate the existence of putative causative alleles at the VNTR locus.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
, the animals considered in this study are organized according to 3 breed-specific granddaughter designs (GDD; Weller et al., 1990) including 6, 9, and 26 half-sib families for the Montbéliarde, Normande, and Holstein breeds, respectively. A total of 3,025 progeny-tested sons were used. In addition to the 14 initial families (Boichard et al., 2003), 27 new families (4, 6, and 17 in the Montbéliarde, Normande, and Holstein breeds, respectively) derived from the French marker-assisted selection (MAS) program (Boichard et al., 2002) were integrated in this design.
|
Genetic Markers.
Genotypes for 2 bovine microsatellite markers CSSM066 and ILSTS039 were available from the MAS and initial QTL programs (Boichard et al., 2002, 2003). However, in the 2003 QTL program, animals had been genotyped only for CSSM066; thus, 2,941 (97%) and 1,514 (50%) sons were genotyped for CSSM066 and ILSTS039, respectively.
For the VNTR, the genotyping procedure consisted of a fluorescent PCR amplification with the following primers VNTR_F: TCAGGATCCAGAGGTACCAG and VNTR_R: CAATGAGAAGGCACGGACTGTGAA. For a 5-repeat allele, the length of the resulting PCR product was 228 bp. The PCR reactions were performed using the Go-Taq Flexi (Promega, Paris, France) according to the manufacturer’s recommendations on a PTC-100 thermocycler (MJ Research, Paris, France) in a final volume of 10 µL. Samples were preheated for 5 min at 95°C, subjected to 40 cycles (95°C for 30 s, 58°C for 30 s, and 72°C for 30 s), and a final extension step of 5 min at 72°C. The resulting 228-bp PCR products were purified on a Sephadex G50 column (Amersham, Paris, France) before running on a Megabace 96 capillary sequencer (Molecular Dynamics, Paris, France). Raw data were then analyzed using Genetic Profiler v1.5 (Molecular Dynamics). Thereafter, alleles were named according to the number of repetitions (R) of the 18-bp motif. Some genotyped individuals were sequenced to confirm the expected number of repeats. In total, 2,958 (98%) progeny-tested sons were genotyped for this marker.
The K232A polymorphism was analyzed using the TaqMan allelic discrimination technology according to the standard genotyping protocol. The TaqMan probes (CCGTTGGCCGCCTT and CCGTTGGCCTTCTTA specific for the A and K alleles, respectively) and primers (DGAT1-F: CGCTTGCTCGTAGCTTTGG and DGAT1-R: CCGCGGTAGGTCAGGTTGT) were provided by Applied Biosystems (Paris, France). This procedure genotyped 3,020 (100%) progeny-tested sons for this marker.
For some nongenotyped sires from the GDD, marker genotypes were reconstructed based on the genotypes of their sons. This provided 24 additional genotyped individuals for the VNTR, the K232A polymorphism, and CSSM066, and 14 more for ILSTS039.
Pedigrees.
Extended pedigrees were used for mixed model analyses. In addition to the genotyped animals, their ancestors over 3 generations were included in the pedigree. Noninformative ancestors; that is, ancestors with no parents and at most one progeny, were deleted iteratively. The size of these pedigree files was 1,006, 1,427, and 4,833 animals for the Montbéliarde, Normande, and Holstein breeds, respectively. The data files contained 422, 578, and 2,259 animals with records, respectively.
Statistical Analyses
Estimation of Allelic Frequencies.
Estimation of allelic frequencies was done on the founder alleles. In the 3 different breed-specific GDD, these corresponded to the 2 sire alleles and each son’s maternally inherited alleles.
Linkage Map Construction.
Marker order and map distances were estimated using CRIMAP 2.4 software (Green et al., 1990). The CHROMPIC option was subsequently used to identify unlikely double crossovers. Final map distances were computed based on Haldane’s mapping function.
QTL Mapping Analysis.
Linkage analysis was performed using a regression interval analysis (Haley and Knott, 1992) with the web-based version of the QTL Express software (Seaton et al., 2002) fitting a 1-QTL model every centimorgan along the considered region. Records used were equal to twice the DYD of the bulls, weighted by their respective reliability. A 95% chromosome-wise significance threshold was computed based on 10,000 permutations. The P-value associated with the null hypothesis of a QTL homozygosity was calculated based on the t-test value at the maximum peak position. The rejection threshold was set to 5%.
VNTR and K232A Association Tests.
Association tests were performed considering the 2 markers simultaneously with the following model:
 |
where yij is the DYD of son j of sire i, si is the fixed effect of sire i, nkj is the number of copies of the K allele carried by son j at the K232A polymorphism, nvj is the number of copies of the VNTR allele considered, bk and bv are, respectively, half the allele substitution effect of the K allele at the K232A polymorphism and the VNTR allele, and eij is the residual effect associated to this record. The weight of each record was computed as already described (Kuhn et al., 2004). Because more than 2 alleles were segregating for the VNTR, all alleles with a frequency greater than 0.05 were successively tested using the above model.
Estimation of the Effect of the K232A Polymorphism.
To estimate the effect of the K232A polymorphism, the traits were modeled separately using the following linear mixed model:
 |
where y is a vector containing twice the DYD for bulls, µ is a vector with the fixed effect of the mean, nk is a vector containing the number of copies of the K allele (0, 1, or 2) corresponding to the DYD, bk is the allele substitution effect, u is a vector of random polygenic effects, and e is a vector of random residual effects; Z is a design matrix relating the DYD of the individuals to their corresponding polygenic effect.
The (co)variance structure was:
 |
is the variance associated to the random polygenic effect and 
the residual variance; A is the additive relationship matrix and R is a diagonal matrix containing the inverse of the weight of the corresponding DYD. These weights were daughter equivalents and were estimated as previously described (VanRaden and Wiggans, 1991) with a correction for the number of cows per herd.
These analyses were performed using the ASREML package (Gilmour et al., 2000).
Haplotype Analysis and Linkage Disequilibrium Estimation.
The most likely haplotype phases of each individual from the different breed-specific pedigrees were computed using the deterministic algorithm already described (Boichard et al., 2003). Estimation of haplotype frequencies and linkage disequilibrium (LD) was performed on the founder haplotypes (see above).
Pairwise LD measures were derived from the standard measure of LD between 2 alleles at 2 different loci: Dij = p(AiBj) – p(Ai)p(Bj), where p(Ai) is the frequency of allele Ai at locus A, p(Bj) the frequency of allele Bj at locus B and p(AiBj) the frequency of haplotype AiBj as estimated from the population. Three multiallelic LD measures derived from this standard measure were considered in the study.
First, multiallelic D' was measured as (Hedrick, 1987):
 |
where k and l are the number of alleles for marker A and B, respectively, and
 |
Second, multiallelic r2 was measured as (Hill and Robertson, 1968)
 |
Finally, the multiallelic 
2 statistic is defined as (Hill, 1975)
 |
where N is the number of haplotypes considered to compute the corresponding pairwise measure. 2 is asymptotically distributed according to a 2 distribution with (k – 1)(l – 1) degrees of freedom. The threshold for significance of LD was set to 0.001.
To limit bias introduced by rare alleles in the estimation of the different measures and to preserve power to reject the null hypothesis of absence of LD, marker alleles with frequency <0.05 were pooled.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
). Distributions of genotypes are presented in Table 2 for the 3 breeds. Only 14 out of the 384 Montbéliarde bulls were heterozygous for K232A and none was homozygous for the lysine variant (KK). Indeed, most animals were homozygous for the alanine variant (AA) in both the Montbéliarde and Normande breeds. In the Holstein breed, where the K allele is much more frequent, approximately 10% of the bulls were homozygous (KK) and 43% were heterozygous.
|
). The 2R allele segregated both in the Montbéliarde and Normande breeds and was not reported in previous studies (Kuhn et al., 2004; Sanders et al., 2006). For Holsteins, the VNTR allele frequencies varied within a similar range in Germany and France (Table 3).
|
The estimated 2-point Haldane genetic distances for each pair of markers are given in Table 4. It is worth noting that the VNTR and K232A polymorphisms were less than 10 kb apart. Thus, distances appeared strongly inflated for markers linked to the VNTR. Respectively, 1 recombinant out of 289 coinformative meioses for the VNTR/K232A marker pair was observed, 16 out of 522 for the VNTR/ILSTS039 marker pair, and 114 out of 1,155 for the VNTR/CSSM066 marker pair (131 in total). A small fraction of these recombinant gametes might have been created by mutational events. Only paternal gametes are informative in our design. Consequently, a mutation can create a recombinant gamete only if the maternally inherited allele is identical in state to one of the paternal alleles, or if the mutated allele is identical in state to the alternative paternal allele. Therefore, the observed inflation in genetic distances was probably the result of a recombination hot spot in this centromeric region and an underestimation of genetic distances when considering the K232A polymorphism due to a lack of informative meioses.
|
, all the Montbéliarde sires were homozygous AA at the K232A polymorphism and 5 were heterozygous at the QTL. In the Normande breed, only sire #1611 (out of the 3 sires that were heterozygous at the QTL) was also heterozygous at the K232A polymorphism. In the Holstein breed, the 11 sires heterozygous at the K232A polymorphism were also heterozygous at the QTL and 2 sires homozygous at the mutation were heterozygous at the QTL.
|
VNTR and K232A Association Tests.
Both the VNTR and K232A polymorphisms were previously suggested as causative. Thus, we performed association tests considering the 2 markers simultaneously to test the effect of the different VNTR alleles after correction for the K232A effect. As expected, the allele substitution effect for the K232A polymorphism was highly significant (P < 10–4) for fat content in the 3 breeds. In the Montbéliarde breed it was also significant (P < 0.05) for milk yield and protein content. In the Normande and Holstein breeds, the effect was significant (P < 10–4) for all traits except for protein yield in the Normande breed.
No allele at the VNTR was associated with trait variation across all 3 breeds. Nevertheless, in the Montbéliarde breed, the 4R allele was associated with milk and protein yield (P < 0.01), fat content (P < 10–4), and protein content (P < 0.05) variations, whereas in the Normande breed, only the 3R allele was associated with protein yield variation (P < 0.05). Finally, as observed in the German Holstein population (Kuhn et al., 2004), the 7R allele in the French Holstein breed was significantly associated with an increase in fat yield (P < 0.05), protein content (P < 0.05), and fat content (P < 10–4).
Characterization of the Effect of the K232A Polymorphism.
The average allelic substitution effect of the K232A polymorphism for dairy traits is presented in Table 6. The allelic substitution effect was estimated using twice the DYD. Therefore, it indicates the difference in terms of EBV and not of PTA. In the Holstein breed, the K allele increased fat yield (+14.90 kg) and fat (+0.34 %) and protein (+0.08%) contents, whereas it decreased milk (–351.2 kg) and protein (–4.6 kg) yields. These results are consistent for the 3 breeds but the magnitude of substitution effects varied. These estimated average allelic substitution effects were in agreement with previous studies (Grisart et al., 2002; Spelman et al., 2002; Thaller et al., 2003; Weller et al., 2003).
|
. The variances range from 2.4 to 42.5% of the total genetic variance for protein yield and fat content, respectively.
Haplotype Analysis and D' Among Markers.
As shown in Table 7, all markers were found to be in significant LD with each other among the 3 different populations considered. As expected, irrespective of the multiallelic measure considered (r2 or D'), LD decreases when the genetic distance separating the markers increases. Interestingly, r2 values for the 3 marker pairs involving the K232A polymorphism are all far from 1, indicating that no allele at the 3 other markers (even the VNTR located only 10 kb away) behaved as a perfect surrogate for the K232A alleles. For instance, as shown in Table 8, in the Montbéliarde breed the K allele at the K232A polymorphism was always associated with the 5R allele at the VNTR (the D' value is thus equal to 1), but an almost equal fraction of this latter allele at the VNTR was associated with the A allele at the K232A polymorphism (the r2 value is thus far from 1). Similarly, in French Holsteins, as previously observed in German Holsteins (Kuhn et al., 2004), the K allele was almost always associated with the 5R allele at the VNTR (D' = 0.98). In the Normande breed, the K allele was associated mainly with the 4R allele at the VNTR, whereas a much smaller fraction was also associated with the 5R allele (D' = 0.68).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
The effects of the K232A polymorphism were estimated on an EBV scale but with twice the DYD as records (which were not regressed as EBV). For the Holstein breed, the results were close to those previously published (Grisart et al., 2002; Spelman et al., 2002; Thaller et al., 2003), which were computed on a PTA scale with DYD as records. The K allele improved fat yield and fat and protein contents and had a negative effect on the other dairy traits. The effects in the 3 breeds were consistent with respect to their direction but the magnitude of the estimated effects changed. Similar tendencies have been described in Fleckvieh and Braunvieh cattle and Ayrshire and Jersey breeds, respectively (Spelman et al., 2002; Thaller et al., 2003). However, because the K allele is rare in the Montbéliarde and Normande breeds, the standard error of the estimators of these effects is large in these breeds.
The combination of allele frequency and allele substitution effects gives an indication on genetic variance due to this polymorphism. As expected, this variance was very large for fat content in the Holstein breed (more than 40% of the genetic variation). The proportion of explained genetic variance estimated in this study for the different traits was close to that originally reported (Grisart et al., 2002). In addition, using a variance component approach on a larger sample of the French Holstein population, the proportion of genetic variance explained by the QTL traced by closely linked markers has recently been found to be very similar (Druet et al., 2006). Nevertheless, because of a lower frequency of the K allele in the Israeli Holstein population, lower values have been observed (Weller et al., 2003). Interestingly, the estimated variance for protein yield was relatively small, indicating that the use of DGAT1 to improve this trait (the most important in the French breeding goal) is less efficient.
Although there is no doubt that the K232A polymorphism had a strong effect, at least in the Holstein breed, it seemed not solely responsible for the QTL as previously postulated (Bennewitz et al., 2004; Kuhn et al., 2004). Indeed, 2 Holstein sires appeared to be heterozygous at the QTL but were not heterozygous at the K232A polymorphism. A similar trend was observed in the Normande breed. Moreover, in the Montbéliarde breed, the K232A polymorphism was clearly not the underlying QTN. These results confirm that another mutation must contribute to the observed QTL effect. Recently, a VNTR in the 5' noncoding region of DGAT1 has been proposed to explain the additional part of the QTL variance, and one allele (7R in our study) was suggested as a putative causal allele (Kuhn et al., 2004; Sanders et al., 2006). Our study showed little evidence for this hypothesis. Indeed, among the 6 Holstein sires homozygous at the K232A polymorphism and carrying 1 copy of the 7R allele, only one was heterozygous at the QTL. In the Normande and Montbéliarde breeds, this allele was only present at a low frequency (<0.005) and its effect cannot be estimated. Recently, a functional analysis (Furbass et al., 2006) established that the Sp1 transcription factor could bind to the VNTR and demonstrated experimentally its functional relevance regarding induction of the transduction of a reporter gene. Nevertheless, no significant difference in the stimulating effect of the 3 different alleles compared (3R, 6R, and 7R) was observed. These different elements do not support a direct effect of the VNTR. For each breed considered, a different allele (3R in Normande breed, 4R in Montbéliarde breed, and 7R in Holstein breed) was found to be associated with a significant effect after correction for the K232A effect. This association of some VNTR alleles may be the result of residual LD existing with at least one additional causal polymorphism other than the K232A polymorphism underlying the QTL. Yet, a comprehensive screen for mutations in the DGAT1 gene (in particular in the 5' noncoding region) did not identify any significant association with the trait for any of the polymorphisms found except the VNTR (Winter et al., 2002; Kuhn et al., 2004); therefore, the additional causal variants might have to be searched in the surrounding intergenic regions.
The mutational event creating the K232A polymorphism from the K allele (the ancestral state) probably occurred early in the history of domestication of European cattle (Winter et al., 2002) more than 10,000 yr ago. In our study, the most important allelic diversity at the ILSTS039 microsatellite and the VNTR markers was associated with the A allele at the K232A polymorphism. The high mutation rate at the VNTR (0.8% per generation and paternal gamete from our study) and microsatellite loci (known to range from 10–3 to 10–4) can be considered as the primary factor generating allelic diversity in the haplotypes carrying the A allele, given that the expected numbers of alleles observed at these 2 markers was in almost perfect agreement with a stepwise mutation model (Ohta and Kimura, 1973).
Additionally, the frequency of the ancestral K allele at the K232A polymorphism has increased only recently, particularly in the Holstein breed, possibly due to more pronounced selection favoring fat content. This might have contributed to markedly reduced allelic diversity at surrounding markers by a genetic hitchhiking phenomenon (Grisart et al., 2004). Alternatively, local and recent introgression of the K allele in some European breeds might explain the low level of within-breed diversity associated with the K-carrying haplotypes (Kaupe et al., 2004). Indeed, some ancestral K alleles, probably from different origins, still segregate in the different cattle populations. In our study for instance, within the Normande and Holstein populations, the haplotypes carrying the K allele were mainly associated with different alleles at both the ILSTS039 and VNTR markers. On the other hand, the few K alleles segregating in the Montbéliarde breed might have been introduced recently during the known limited crossbreeding events with Red Holstein, which occurred in the 1970s. Indeed, in the Montbéliarde breed, the K alleles were all associated with the same ILSTS039 and VNTR alleles (218 and 5R, respectively) whereas in the Holstein breed, 85 and 99% of the K alleles were associated with the same 2 alleles.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Received for publication October 26, 2006. Accepted for publication February 28, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. Banos, J. A. Woolliams, B. W. Woodward, A. B. Forbes, and M. P. Coffey Impact of Single Nucleotide Polymorphisms in Leptin, Leptin Receptor, Growth Hormone Receptor, and Diacylglycerol Acyltransferase (DGAT1) Gene Loci on Milk Production, Feed, and Body Energy Traits of UK Dairy Cows J Dairy Sci, August 1, 2008; 91(8): 3190 - 3200. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Naslund, W. F. Fikse, G. R. Pielberg, and A. Lunden Frequency and Effect of the Bovine Acyl-CoA:Diacylglycerol Acyltransferase 1 (DGAT1) K232A Polymorphism in Swedish Dairy Cattle J Dairy Sci, May 1, 2008; 91(5): 2127 - 2134. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
