By combining a genome wide association study and a candidate gene approach in AI rams of the Norwegian White Sheep breed, we successfully identified a SNP causing an amino acid change in the mature GDF9 protein. This SNP showed a stronger association to EBVs for litter size than any of the SNPs present at the Illumina 50K SNP array (Figure ). This sequence variant has previously been reported to exist in Belclare and Cambride sheep (the G7 polymorphism) [5
], but these authors focused on polymorphisms that caused female sterility in the homozygous state. Animals homozygous for c.1111A were found to be fertile in that study, and no additional attempt to correlate this polymorphism with ovulation rate/litter size was performed. Also, the number of animals carrying this mutation was most likely too low to detect any genotype dependent variation in fertility.
Without any functional testing, we cannot conclude that the c.1111G>A is the causal mutation for the differences in EBVs for litter size observed in the NWS population, however our evidence is suggestive of a functional association for two reasons. Firstly, this is an amino acid change in the mature region (the bioactive part) of the GDF9 protein [13
], and secondly, valine is found in this position across 6 highly different mammalian species (sheep, cattle, pig, cat, human and mouse), while in chicken and zebrafish valine is substituted by another aliphatic amino acid, isoleucine (Figure ).
Figure 6 Amino acid conservation at the 371 – position. A sequence alignment comparing the protein sequence at ovine position 371 across Sheep (NP_001136360.1), Cattle (NP_777106.1), Pig (NP_001001909.1), Cat (NP_001159372.1), Human (NP_005251.1), Mouse (more ...)
GDF9 is mediating its effect by binding to transforming growth factor, beta receptor 1 (TGFBR1) [14
] and bone morphogenetic protein receptor, type II (BMPR2) [16
]. As pointed out by Hanrahan et al. [5
], the c.1111G>A polymorphism represents a relatively conservative change in the sense that one nonpolar amino acid (V) is substituted by another (M). However, the side chain of methionine is structurally different from that of valine, so depending on the location relative to the receptor binding region of GDF9, a reduced (but not lost) binding capacity can possibly be explained by this change.
Daughters of rams being homozygous for c.1111A gave birth to 0.46 - 0.57 additional lambs compared to daughters of c.1111G homozygous rams, while daughters of heterozygous rams gave 0.20 - 0.25 additional lambs. These figures can be considered as conservative, since they build on EBVs that are regressed towards the mean to compensate for a non-infinite number of progeny tested offspring. Based on the development of the c.1111A allele frequency in the AI rams over time (Figure ), we can assume a similar but lagging frequency development in the ewe population, and thus low overall frequency. Basically, it will then be the heterozygous effect that is contained in the contrast of GG with AA rams, while the GA group with 50% heterozygote daughters will be logically intermediate. Also, when using the animal model for calculating EBVs without modeling the allele effect of c.1111A, the sire will mainly provide the allele, while the effect will be shared with the dam and therefore be underestimated when the frequency is low among the ewes.
The EBVs were estimated based on the daughters’ performance in terms of number of lambs born, but only ewes that gave births were included. Since GDF9 is known to influence oocyte maturation and ovulation rate, the ewe’s genotype will be determinative for her fertility. Therefore, this study should be followed up by genotyping a large number of ewes with known reproductive performance to obtain a better estimate of the separate genotypic effects, and to confirm that homozygous ewes are fertile as reported for Cambridge and Belclare sheep [5
No single event in the breeding history of NWS has influenced litter size more than the introduction of Finnish landrace. Finnish landrace sheep are well known for their high fertility, and have been crossed to several breeds to investigate this trait [17
]. Also in Norway, Finnish landrace was imported in late 1960s and early 1970s to improve fertility [19
]. The significant phenotypic effect of the GDF9
c.1111A - allele observed in this study could therefore indicate that this allele originate from Finnish landrace. The development in allele frequency in the AI rams population since 1990 also resemble the pattern observed for another imported allele, namely the Texel MSTN
c.*1232A - allele [22
]. In both cases a new breed has been imported to improve key traits, litter size and meatiness respectively, and after some latency the organized breeding system catches the causal allele and rapidly increase its frequency, first in the AI ram population and subsequently in the whole ewe population.