The maximum or minimum of the lactation curve for the different milk production traits has been stated to be between 28 and 56 days after calving in Holstein Friesian cattle [25, 26], which would be during the last three 10-day intervals for 31–60 days in milk (DIM) in our analysis.
We could identify changes in allele effects over the first 60 lactation days and compared to 305-day records. One reason for the change in allele effects could be that they only follow the change in EBVs due to the strong correlation (>0.99) in the response variable between consecutive intervals. However, the change in allele effects seems to be independent from the change in EBVs which is particularly obvious for milk yield, where the EBVs decreased over the first 60 lactation days but most of the allele effects increased. Furthermore, the variation in EBVs was fairly constant and only marginally increased for all traits over the first 60 lactation days, therefore presented comparisons of absolute allele effects over time is reasonable.
Three groups of markers were identified that showed different directions in allele effect changes. Group 1 consisted of five markers in the DGAT1 region between 0.05 and 2.6 Mb on chromosome 14 that showed increasing effect sizes for all traits during the first 60 days of lactation. The rest of 29 markers on chromosome 14 were regarded as group 3 with increasing effect sizes, apart from fat yield. DGAT1 is a major gene for a QTL on chromosome 14 with opposing effects on milk yield and milk fat [9, 10, 27, 28]. Even though it was shown that the enzyme activity of the DGAT1-protein can depend on the mutation variant [29, 30], it was not yet demonstrated that functional changes or quantity of expression is altered during the course of a lactation [29, 31, 32]. Therefore, other loci in that QTL region may explain the time-dependent changes of effect sizes in early lactation as well as the opposing development of allele effects for fat yield in group 1 and 3, as found in this study. For example, the CYP11B gene, which is located 1 Mb upstream of the DGAT1 gene and on the edge of a haplotype block of three significant markers, showed an opposite effect on fat yield compared to DGAT1. This opposing effect could decrease the overall effect of the DGAT1 region on fat yield, which could explain the herein reported decreasing effect size of markers associated with fat yield in group 3.
Overlapping with a haplotype close to CYP11B was also a region containing eight genes encoding proteins for the lymphocyte complex. Markers in that region were significant for the first 60 lactation days as well as for the 305-day records and strongly affecting most milk production traits. This region could therefore play a role in the immune response of the mammary gland and prevents inflammation during lactation.
Group 2 included markers on chromosomes 6, 18 and 27. The significant markers showed decreasing effects for all traits, which were significant for protein content for markers on chromosomes 6 and 18. The markers on chromosome 6 in the Casein-gene cluster were only significant during the first 60 lactation days with largest effects at the beginning of the lactation. This co-occurrence with changes in gene expression patterns of the casein genes during early lactation . The same pattern of changes of genetic effects was found for a marker on chromosome 18 that is located in the SICLEC12 gene and linked to SICLEC genes 10 and 14 which are all involved in immunoglobulin production. The co-occurrence of changes in allele effects over time of markers in the Casein-gene cluster and the SICLEC genes can be explained by their biological functions. In the first days after calving, when the colostrum is produced, milk is rich in proteins, especially in immunoglobulins, which have bioactive functions and help to activate the calf’s immune system [34, 35]. Furthermore, other genes of the SICLEC family have been reported to affect productive live and fertility in Holstein cattle and are linked to leptin signaling [36, 37]. Leptin plays a major role in regulating foot intake and thus the energy status of a cow which could affect both fertility and productivity [38, 39].
The significant region on chromosome 27, with decreasing effects on fat yield and content during the first 60 days of lactation, was supported by only one marker in the initial model, but five more markers when accounting for the DGAT1 effect. The 1-acylglycerol-3-phosphate O-acyltransferase 6 gene (AGPAT6) is located 79.38 Kb upstream of the initially only significant marker on the edge of a haplotype block. AGPAT6 has a similar function as DGAT1. It is an enzyme in the phospholipid and triglyceride biosynthesis, and thus, contributes to the production of milk fat . The expression of AGPAT6 in the mammary gland was shown to increase drastically over the first 60 lactation days and decreases afterwards . The change in the expression profile is not consistent with the herein reported trend of decreasing allele effects over the first 60 days of lactations but the fact that markers in the AGPAT6 region were only significant at the beginning of lactation showed that the impact on milk fat must be diminishing in late lactation.
Additionally, the GO terms chemokines and peptidases occurred significantly often for our marker set and especially in the region of markers on chromosome 6, 18 and 27 which were only significant in early lactation. Chemokines attract lymphocytes and macrophages and are thus related to immune response, either to help the offspring or prevent infections of the udder itself in a time of high activity and metabolic stress . Something similar can be said about peptidases, especially the kallikrein-related peptidases that were predominant in the region on chromosome 18; connection to immune response and also associations with breast cancer have been reported [42–44]. Chemokines and peptidases however, show no direct link to the traits that were investigated in this study.
Not included in the three described effect groups were markers on chromosomes 5. Only one marker was initially significant for the last 10-day interval, and thus, no change in effect size could be determined. However, five more markers for the 10-day intervals and six markers for the 305-day records became significant when accounting for the DGAT1 effect. The markers had a tendency towards increasing allele effects for fat yield and content over the first 60 lactation days. This, and the fact that initially most of the markers on chromosome 5 were significant for the 305-day records concurs with a decrease in fat content when energy availability is low during early lactation .
Only a few studies have focused on time-dependent genetic associations in livestock to date and, as this study shows, the investigation of association at certain lactation stages seem to be a promising approach to detect loci with only small overall effects [13, 46–48]. Thus, significant associations can sometimes only be found when the phenotype was recorded at the right time. Because the impact of loci changes over time, an investigation at certain developmental or lactation stages might be a contribution to detect parts of the otherwise missed genetic variance and may be better in detecting quantitative trait genes with overall small effect.
Finally, we want to propose two candidate genes for markers detected for the non-return rate in heifers (NRH). The first gene is the kelch-like 8 (KLHL8), located on chromosome 6 within 68.91 Kb of the nearest significant marker. The KLHL8 gene was reported to be preferentially expressed in the female gonads of fish (zebrafish and carbiomedaka) where it may play a role in oogenesis . Even though fish are rather distantly relate to cattle, the gene function might be evolutionary conserved. The second gene is nidogen 1 (NID1) on chromosome 28 and within 303.97 Kb of the nearest significant marker. NID1 is increasingly expressed in the focimatrix (follicular basal laminas) when follicles enlarge before ovulation . Thus, both genes, KLHL8 and NID1, could affect the non-return rate by regulating the ovulation process.
An antagonistic relation between production and fertility traits reflects the competition between these traits for the same body resources [16, 17]. We confirm previous studies reporting an unfavorable genetic correlation between yield and fertility traits [17, 51–53]. The EBVs for content traits were mostly positively correlated with the EBVs for fertility which might be due to a spurious effect resulting from both content and fertility traits, being negatively correlated with milk yield. Additionally, we report an increase of the negative genetic correlation over the first 60 lactation days for the EBVs of milk and protein yield and smaller negative and larger positive correlations for the EBVs of content traits. The stronger negative correlation between EBVs for yield and fertility traits could result from the also increasing milk production in early lactation, depleting important body resources needed for most fertility traits.
Additionally to the time dependency of genetic correlation between production and fertility traits, we reported differences for genetic correlation within genotypes of the most significant markers from the GWAS. With only few exceptions, alleles that significantly increased the EBV for milk yield showed a weaker negative correlation with most of the fertility traits if this allele was present in the genotype. Though for different markers compared to this study, Pimentel et al. (2011) reported SNPs with favorable effects on production and fertility traits and came to the same conclusion that a selection for higher performance based on the genotypes of certain loci does not further influence the fertility negatively .
The content traits showed mainly stronger positive correlations with most of the fertility traits if the yield decreasing allele was present in the genotype. Again, a spurious effect due to the negative correlation of content traits as well as fertility traits with milk yield cannot be ruled out.