J. Anim. Sci. 2003. 81:912-917
© 2003 American Society of Animal Science
Evaluation of melatonin receptor 1a as a candidate gene influencing reproduction in an autumn-lambing sheep flock
D. R. Notter*,1,
N. E. Cockett
and
T. S. Hadfield
* Department of Animal and Poultry Sciences, Virginia Polytechnic Institute and State University, Blacksburg 24061 and
and
Department of Animal, Dairy, and Veterinary Sciences, Utah State University, Logan 84322
1 Correspondence:
phone: 540-231-5135; fax: 540-231-3010; E-mail:
drnotter{at}vt.edu.
 |
Abstract
|
|---|
The effect of the melatonin receptor 1a (MTNR1A) gene on fertility and litter size in autumn lambing was studied with 373 ewes from a population of sheep selected for 10 yr for fertility in May and June matings. Animals were from a composite line of 50% Dorset, 25% Rambouillet, and 25% Finnsheep. Two restriction fragment length polymorphisms were present in the population, with allelic frequencies of 0.42 and 0.58 for an MnlI polymorphism (alleles M and m, respectively) and of 0.34 and 0.66 for an RsaI polymorphism (alleles R and r, respectively). Genotypic frequencies for the polymorphisms were not independent, suggesting an association between them in foundation animals. Effects of MTNR1A genotype on fertility and litter size were evaluated using mixed linear model or REML procedures, but were not significant for matings involving ewes of all ages. However, in adult ewes (3 yr old and older), fertility of ewes of genotype mm was 10.0 (mixed model; P < 0.09) to 11.2% (REML; P = 0.03) less than that of other ewes. Genotypic effects associated with MTNR1A in adult ewes accounted for 23.8% of the estimated total additive genetic variance in fertility. Litter size was not significantly associated with MTNR1A genotype; adult ewes of genotype mm had approximately 0.11 fewer lambs per ewe lambing than ewes of other genotypes. Fertility in autumn lambing is lowly heritable, expressed only in females, and manifested relatively late in life and only in some management systems. Access to genetic markers would thus be advantageous in selection programs.
Key Words: Genetic Markers • Melatonin • Receptors • Reproduction • Selection • Sheep
|
Introduction
|
|---|
Use of polymorphic QTL or associated genetic markers in selection programs is particularly advantageous for traits that are lowly heritable, expressed relatively late in life, expressed in only one sex, or manifested only in some environments or management systems. Fertility of ewes lambing in autumn meets all these criteria (Al-Shorepy and Notter, 1996; 1997), and efficiency of selection to reduce seasonality of breeding in sheep would be improved by identification of informative genetic markers. Annual fluctuations in timing and duration of the nocturnal elevation in circulating melatonin is known to be a key factor influencing seasonal reproduction in sheep (Malpaux et al., 1996). Discovery by Messer et al. (1997) of two polymorphic RFLP sites within the ovine melatonin receptor 1a gene (MTNR1A) provided opportunity to evaluate the influence of this gene on seasonal reproduction. Pelletier et al. (2000) subsequently reported that these two polymorphic sites are separated by only 5 bp and that neither is directly associated with an amino acid substitution in the melatonin receptor. Pelletier et al. (2000) also reported that the MTNR1A genotype was associated with seasonal reproduction in Merino d’Arles ewes. The current study was designed to determine if a similar relationship exists within a crossbred population developed through selection for fertility in autumn lambing.
|
Materials and Methods
|
|---|
Records of fertility (ewes lambing per ewe exposed) and litter size (number of lambs born per ewe lambing) in autumn lambing came from a flock selected for fertility and a contemporary unselected environmental control line (Figure 1
). Both lines were derived from a common base population of 50% Dorset, 25% Rambouillet, and 25% Finnsheep breeding (Fossceco and Notter, 1995). The selected line was maintained as a closed flock of approximately 125 ewes and 10 rams. Approximately one third of the ewes and one half of the rams were replaced each year; selected ewe lambs were first exposed to breeding at 7 to 8 mo of age. The 55 ewes in the environmental control line were born in a spring-lambing genetic control flock (Figure 1). Ewe lambs from the genetic control line were transferred to the autumn-lambing environmental control line at a rate that allowed maintenance of a comparable ewe age distribution within selected and environmental control lines. This population structure ensured that ewes in the environmental control line were both unselected and contemporary with selected ewes. In May and June of each year, ewes were exposed to 10 to 15 rams in single-sire breeding pastures for 60 d. Lambing therefore occurred in October and November.
View larger version (33K):
[in this window]
[in a new window]
|
Figure 1. Experimental design of the selection experiment. Arrows indicate the source of replacements for each flock.
|
|
Details of management and selection procedures were given by Al-Shorepy and Notter (1996; 1997), and selection responses through 1996 were summarized by Notter et al. (1998). The selection experiment ended after the autumn 1998 lambing. The genetic control line was terminated at that time, but the selected line was maintained, and evaluation of selected and remaining environmental control line ewes in autumn lambing continued through autumn 2000. Genotyping of animals began in 1997 following identification of polymorphic RFLP sites within MTNR1A (Messer et al., 1997). All breeding animals present in selected and environmental control flocks were genotyped at that time, and replacement animals were genotyped through 2000.
Genomic DNA was extracted from whole blood by salt precipitation (Miller et al., 1988), and amplification of an 824-bp fragment of ovine MTNR1A was performed as described by Messer et al. (1997). Polymorphisms in MTNR1A were detected after digestion of resulting amplicons with MnlI or RsaI (Messer et al., 1997). In addition to constant bands of 216, 137, 82, and 67 bp, MnlI also yielded polymorphic fragments of 236 and 50 bp when the cleavage site was present (allele M) or a 286-bp fragment if the site was absent (allele m). Digestion with RsaI yielded an RFLP with fragments of 290 and 5 bp when the cleavage site was present (allele R) or 295 bp when the site was absent (allele r). Genotypes were represented as MM, Mm, and mm for MnlI and RR, Rr, and rr for RsaI.
Data included results of 2,490 breeding opportunities and 1,157 autumn lambings by 903 ewes between 1988 and 2000. The pedigree file included these animals and their ancestors back to the foundation of the flocks and contained 1,503 animals. Genotypes for at least one polymorphism were available for 373 ewes by 86 sires with 1,215 breeding records and 692 lambings. Proportions of genotyped ewes by year of birth and year of mating are shown in Figure 2. Because of the very low mean fertility of young ewes, a supplemental set of analyses was performed on only matings of adult ewes (third and subsequent breeding opportunities). These data contained results of 524 breeding opportunities and 394 lambings by 206 ewes with genotypes for both polymorphisms.
View larger version (26K):
[in this window]
[in a new window]
|
Figure 2. Number of ewes born or mated in each year of the study (open bars) that were genotyped (dark bars). Note the difference in scale between the charts.
|
|
Data were analyzed with linear models using either the mixed model procedure of SAS (SAS Inst., Inc., Cary, NC) or REML procedures of Boldman et al. (1993). Genotypic effects were evaluated jointly by fitting effects of each of the six high-frequency haplotypes shown in Table 1. Samples from 11 of 373 ewes did not produce definitive genotypes for both sites and these animals were removed from the analysis. The mixed model analysis included fixed effects of genotype, ewe age (1, 2, or 
3 yr), and year of mating and random effects of ewe (nested within genotype) and breeding pasture (nested within year). Genotypic effects were tested with the between-ewe mean square, year effects were tested with the between-breeding pasture mean square, and ewe age effects were tested with the residual mean square.
The REML model included fixed effects of genotype, ewe age, and year and random animal additive, animal permanent environmental, breeding pasture, and residual effects. The full pedigree file was incorporated to account for relationships among animals. Variance components for random effects were initially estimated using records of both genotyped and nongenotyped ewes, and these estimates were subsequently used as input parameters in the analysis of performance of genotyped animals.
All analyses were applied to records from ewes of all ages and from adult ewes only. The analyses were likewise repeated for fertility and litter size. Based on results of Al-Shorepy and Notter (1996), effects of breeding pasture were not included in the model for litter size. Based on results of Pelletier et al. (2000) and Wright (2000), a contrast was used to compare individuals carrying at least one copy of the M allele (genotypes MMrr, MmRr, and Mmrr) to those that were homozygous for the m allele (genotypes mmRR, mmRr, and mmrr).
|
Results and Discussion
|
|---|
Genotypic Frequencies
Genotypic frequencies for the two polymorphisms within ovine MTNR1A are shown in Table 1
. Among ewes with records, allelic frequencies were 0.42 and 0.58 for the MnlI polymorphism and 0.34 and 0.66 at the RsaI polymorphism. If the two restriction sites were considered separately, genotypic frequencies did not differ from those expected under random mating for the MnlI (
2 = 0.15 with 1 df; P < 0.75) or RsaI (2 = 2.35; P < 0.15) polymorphism. However, the polymorphisms were not in joint equilibrium (2 = 102.7 with 4 df; P < 0.001). Animals of genotypes MMRR, MMRr, and MmRR occurred at low frequencies and were underrepresented relative to expectations based on allelic frequencies.
Values in Table 1 suggest the presence of three common parental gametic types (Mr, mR, and mr), whereas MR gametes were rare. Direct estimation of frequencies of parental gametes requires knowledge of the gametes that formed the MmRr individuals, which cannot be determined from these data. However, Hill (1974) provided an iterative method to estimate gametic frequencies in data of this sort. The resulting estimated frequencies of parental gametes were 0.036 for MR, 0.385 for Mr, 0.302 for mR, and 0.277 for mr. With these estimates of gametic frequencies, genotypic frequencies in Table 1 still differ from those expected under random mating (P < 0.001) conditions. Departures from expected random-mating equilibrium frequencies remained (P < 0.01) after removal of low-frequency genotypes containing MR and recalculation of the 2 statistic under a three-allele model. The mmrr and mmRR genotypes tended to be over-represented, whereas mmRr was under-represented.
Precise expression of random-mating equilibrium genotypic frequencies was not expected in these animals. Ewes were born over 11 yr and older genotyped ewes represented a selected group, especially for ewes born before 1994 (Figure 2). Ewes in selected and control lines had a common origin, but were maintained separately following 1988. The generation interval was about 2.4 yr in the selected line and 3.6 yr in the control line (Al-Shorepy and Notter, 1997). Thus, opportunities for divergence in allelic frequencies due to genetic drift were modest; large changes in allelic frequencies due to selection would be anticipated only within chromosome segments with major effects on fertility. These restriction sites are only 5 bp apart (Pelletier et al., 2000), so recombination is unlikely; associations among alleles likely represent associations present in foundation animals. The relatively high and uniform frequencies of Mr, mR, and mr in this crossbred composite line lead to speculation that frequencies of different allelic combinations may have been different in different breeds, but this hypothesis cannot be tested because genotypic information is not available for foundation animals.
Genotypic Effects on Fertility
In the mixed model analyses of fertility, repeatability of ewe effects averaged 0.16 for all ewe ages and 0.25 for adult matings. Breeding pasture effects accounted for an average of 6.8 and 10.1% of phenotypic variance for all ewe ages and adult matings, respectively.
Fertility increased (P < 0.001) with age (Table 2), and year effects were also significant (P < 0.02). Genotypic effects (Table 3) were evaluated for the six genotypes that occurred with frequencies greater than 0.10 (Table 1). Genotypic effects were not significant (P = 0.25) for all ewe ages or in adult matings. However, ewes carrying at least one copy of the M allele tended to be superior to others in adult matings (mean difference of 10.0 ± 5.7%; P = 0.09), but not for all ewe ages (mean difference of 5.7 ± 4.1%; P < 0.20).
View this table:
[in this window]
[in a new window]
|
Table 2. Least squares means and SE for ewe age effects on fertility (%) and number born from the mixed linear model
|
|
View this table:
[in this window]
[in a new window]
|
Table 3. Least squares means and SE for fertility and number born for each genotype from mixed linear models applied to records from all ewe ages or only adult matings
|
|
Variance component estimates for fertility from REML using the complete data set and ignoring genotypic effects (Table 4) yielded estimates of heritability of 0.09 for all ewe ages and 0.07 for adult matings. Repeatability estimates (calculated as the sum of ewe additive genetic and permanent environmental effects) were 0.15 for both all ewe ages and adult matings and were smaller than repeatability estimates from the mixed model analysis, especially for adult matings. The proportion of phenotypic variance attributed to effects of breeding pasture was 0.08 in all matings and 0.11 in adult matings.
View this table:
[in this window]
[in a new window]
|
Table 4. Variance component estimates from REML for fertility and number born for all ewe ages or only adult matingsa
|
|
Constants and SE for genotypic effects from REML (Table 5) revealed relatively large differences among genotypes, with a range in fertility of 13.3% for all ewe ages and 21.0% in adult matings. Homozygous mm ewes were inferior to the remaining genotypes by 5.7 ± 3.4% in all matings and by 11.2 ± 5.1% (P = 0.03) in adult matings.
View this table:
[in this window]
[in a new window]
|
Table 5. Constants and SE for fertility and number born for each genotype from REML analysis applied to all ewe ages or only adult matingsa
|
|
Results of the REML analysis were used to quantify the apparent contribution of this marked chromosome segment to genetic variation in fertility under a single-locus, three-allele model considering only the Mr, mR, and mr alleles. Genotypes under this model are represented as Mr/Mr, Mr/mR, and so on, to specify both allelic composition and linkage phase in uniting gametes. The mR/mr heterozygote was inferior in genotypic value to both parental homozygotes in both sets of data, suggesting overdominance for low fertility for this pair of alleles. The Mr/mR heterozygote was likewise inferior to either homozygote in all matings. In adult matings, Mr/mR was similar to the mR/mR homozygote and inferior to Mr/Mr, yielding an estimate of degree of dominance of 0.75 for mR relative to Mr. In contrast, the Mr/mr heterozygote was intermediate to Mr/Mr and mr/mr in both sets of data but more similar to Mr/Mr. The estimate of degree of dominance of Mr relative to mr was 0.75 in all matings and 0.59 in adult matings.
Estimates of breeding values and dominance deviations for fertility (Jacquard, 1974) were derived for each genotype at their predicted random-mating equilibrium frequencies to allow partition of total genotypic variance (
) at this locus into additive (
) and dominance () components. Across all ewe ages, estimates of , , and 
associated with this marked segment of ovine chromosome 26 were 20.9, 10.9, and 10.1%2, respectively. In adult matings, comparable estimates were 49.6, 35.7, and 13.9%2, respectively. Estimates of the total additive and ewe permanent environmental variances for fertility, ignoring the markers and using all records, were 167 and 107%2, respectively, for all ewe ages and 151 and 148%2, respectively, for adult matings. This marked chromosome segment thus accounted for 6.5 and 23.8% of the estimated total additive genetic variation in fertility for all ewe ages and in adult matings, respectively. Dominance effects of this chromosome segment accounted for 9.4 and 9.3% of ewe permanent environmental effects for all ewe ages and in adult matings, respectively.
An association between fertility in autumn lambing and MTNR1A genotype was detected only in adult matings and only from an orthogonal contrast in the REML analysis. Failure to detect genotypic effects in matings that included 1- and 2-yr-old ewes likely reflected the lower mean fertility of these ewes, especially at 1 yr of age (Table 2). Notter et al. (1998) reported that fertility of ewe lambs has not responded to selection, despite substantial selection response in older ewes. Younger ewes therefore appear more sensitive to seasonal effects on reproduction. Notter and Chemineau (2001) also reported that nocturnal circulating melatonin concentrations decreased with increasing ewe age. The direct role of the region surrounding MTNR1A on this age-specific reproductive pattern cannot be determined. However, the limited expression of genetic differences in fertility in young ewes accentuates the potential value of marker-assisted selection for this trait.
The REML analysis, which attempted to fit both genotypic effects and residual additive effect, and which accounted for relationships among animals, seemed to do a better job of detecting genotypic effects. However, this approach was somewhat ad hoc and was dependent on variance components estimated from the full dataset without simultaneous consideration of genotypic effects of MTNR1A. Limited numbers of observations, incomplete genotyping, and small family sizes precluded use of more comprehensive analyses.
Results of the current study support those of Pelletier et al. (2000), who reported that Merino d’Arles ewes that regularly cycled in spring had a higher frequency of allele M (indicating the presence of the Mln1 restriction site) than ewes that had not cycled (0.76 vs. 0.50; P < 0.001). Wright (2000) reported that the M allele had a positive effect on autumn lambing success in Columbia ewes, but not in Hampshires or a Targhee-Finsheep-Dorset composite line.
Genotypic Effects on Litter Size
Repeatability of ewe effects on litter size from mixed model analyses averaged 0.14 for all ewe ages and 0.16 for adult matings. Litter size increased with ewe age (P < 0.001; Table 3) and tended to differ among years (P < 0.10). Genotypic effects on litter size were small and not significant for all ewe ages (P = 0.70). Differences among genotypes were somewhat larger in adult matings (P = 0.20) and generally consistent with observed differences in fertility (Table 3). Ewes of genotype MMrr were superior in mean litter size (2.00 ± 0.09), whereas ewes of genotype mmRr were notably inferior (1.64 ± 0.14). Adult ewes that carried at least one copy of M had slightly larger litters than those that did not (1.87 ± 0.05 vs. 1.76 ± 0.08; P < 0.30).
Repeatability estimates for litter size from REML analysis of all records ignoring effects of MTNR1A genotype (Table 4) were 0.14 for all matings and 0.17 for adult matings and were thus similar to estimates from mixed model analyses. However, heritability estimates for litter size in these data were low (0.02 for all ewe ages and 0.01 in adult matings). This result was not consistent with the heritability estimate of 0.10 for litter size reported by Al-Shorepy and Notter (1996) using records from the early years of the selection study and may indicate that heritability of litter size has changed during the course of the selection. Differences among genotypes were small across all ewe ages. In adult matings, superiority of MMrr individuals and poor performance by mmRr were notable, but differences among other genotypes were very small, and consistent patterns of genotypic effects were not observed. Adult ewes carrying at least one copy of the M allele produced 0.11 more lambs than mm ewes. Estimates of genotypic effects from the mixed model and REML analyses were thus essentially identical for litter size, but SE from the REML analysis were larger. The contribution of MTNR1A to additive variance in litter size could not be calculated because of the low heritability observed for this trait.
Conclusions
Results of this study are consistent with those of Pelletier et al. (2000) and Wright (2000) and suggest that the region surrounding MTRN1A on ovine chromosome 26 influences autumn lambing performance. However, a specific role for MTRN1A cannot be confirmed. The two polymorphisms studied here do not result in amino acid substitutions in the melatonin receptor (Pelletier et al., 2000), and functional differences in the receptor therefore are not anticipated for different genotypes. However, both Pelletier et al. (2000) and Barrett et al. (1997) have documented the presence of other polymorphic sites in MTRN1A that do produce different forms of receptor protein. The polymorphisms studied here should thus be considered possible genetic markers that may be associated with other causal polymorphisms within the gene, in associated regulatory sequences, or in adjacent genes.
Results of this experiment highlight challenges involved in identification of QTL in selected populations, especially in species of low fecundity. Well-designed selection experiments with sheep and meat cattle commonly use relatively large numbers of founder sires and rely on preferential removal of inferior individuals within and among families over several overlapping generations to generate selection response.
Opportunities for genetic drift, founder effects, and differences in linkage phase between QTL and marker genes among families are therefore large. For foundation animals in projects initiated a decade or more ago, DNA is often not available, whereas DNA storage on all animals in selection lines should now be a standard procedure. In contrast to the experimental designs commonly used in selection experiments, experiments to detect QTL commonly rely on production of relatively large numbers of unselected, segregating progeny from very few families. These designs have reasonable power to identify associations between markers and QTL, but require that the small number of sampled families represent the population of interest. Selected populations are potentially rich sources of QTL for traits under selection, but will commonly require additional experimental work to confirm the existence and estimate the importance of such genes. Screening of selected populations using methods similar to those of the current study will be useful to identify genes or chromosome segments that merit further study.
|
Implications
|
|---|
The region of ovine chromosome 26 surrounding the melatonin receptor 1a gene seems to affect fertility in autumn lambing. Genetic markers in this region merit study for improvement of this lowly heritable and sex-limited reproductive trait.
Received for publication February 8, 2002.
Accepted for publication November 27, 2002.
|
Literature Cited
|
|---|
Al-Shorepy, S. A., and D. R. Notter. 1996. Genetic variation and covariation for ewe reproduction, lamb growth, and lamb scrotal circumference in a fall-lambing sheep flock. J. Anim. Sci. 74:1490–1498.[Abstract]
Al-Shorepy, S. A., and D. R. Notter. 1997. Response to selection for fertility in a fall-lambing sheep flock. J. Anim. Sci. 75:2033–2040.[Abstract/Free Full Text]
Barrett, P., S. Conway, R. Jockers, A. D. Strosberg, B. Guardiola-Lemaitre, P. Delagrange, and P. J. Morgan. 1997. Cloning and functional analysis of a polymorphic variant of the ovine Mel 1a melatonin receptor. Biochemica Biophysica Acta 1356:299–307.[Medline]
Boldman, K. G., L. A. Kriese, L. D. Van Vleck, and S. D. Kachman. 1993. A Manual for Use of MTDFREML. A Set of Programs to Obtain Estimates of Variances and Covariances [Draft]. ARS, USDA, Washington, D.C.
Fossceco, S. L., and D. R. Notter. 1995. Heritabilities and genetic correlations of body weight, testis growth and ewe lamb reproductive traits in crossbred sheep. Anim. Sci. 60:185–195.
Hill, W. G. 1974. Estimation of linkage disequilibrium in randomly mating populations. Heredity 33:229–339.[Medline]
Jacquard, A. 1974. The Genetic Structure of Populations. Springer-Verlag, New York.
Malpaux, B., C. Viguie, D. C. Skinner, J. C. Thiery, J. Pelletier, and P. Chemineau. 1996. Seasonal breeding in sheep: Mechanism of action of melatonin. Anim. Reprod. Sci. 42:109–117.
Messer, L. A., L. Wang, C. K. Tuggle, M. Yerle, P. Chardon, D. Pomp, J. E. Womack, W. Barendse, A. M. Crawford, D. R. Notter, and M. F. Rothschild. 1997. Mapping of the melatonin receptor 1a (MTNR1A) gene in pigs, sheep, and cattle. Mamm. Genome 8:368–370.[Medline]
Miller, S. A., D. D. Dykes, and H. F. Polesky. 1988. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res. 16:1215–1216.[Free Full Text]
Notter, D. R., S. A. Al-Shorepy, J. N. Vincent, and E. C. McQuown. 1998. Selection to improve fertility in fall lambing. Proc. 6th World Congr. Genet. Appl. Livest. Prod., Armidale, Australia 27:43–46.
Notter, D. R., and P. Chemineau. 2001. Nocturnal melatonin and prolactin plasma concentrations in sheep selected for fertility in autumn lambing. J. Anim. Sci. 79:2895–2901.[Abstract/Free Full Text]
Pelletier, J., L. Bodin, E. Hanocq, B. Malpaux, J. Teyssier, J. Thimonier, and P. Chemineau. 2000. Association between expression of reproductive seasonality and alleles of the gene for Mel1a receptor in the ewe. Biol. Reprod. 62:1096–1101.[Abstract/Free Full Text]
Wright, C. D. 2000. Polymorphisms at the melatonin (MTNR1A) gene and their association to reproductive performance in fall lambing ewes. M.S. Thesis, South Dakota State Univ., Brookings.
Top
Abstract
Introduction
