Edited by: Wendy Mercedes Rauw, Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria, Spain
Reviewed by: John B. Cole, United States Department of Agriculture, USA; Wen Huang, North Carolina State University, USA
*Correspondence: Eva M. Strucken, School of Environmental and Rural Science, University of New England, Room 4, W47 Homestead Building, Armidale, NSW, 2351, Australia
This article was submitted to Livestock Genomics, a section of the journal Frontiers in Genetics
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Lactation is a dynamic process, which evolved to meet dietary demands of growing offspring. At the same time, the mother's metabolism changes to meet the high requirements of nutrient supply to the offspring. Through strong artificial selection, the strain of milk production on dairy cows is often associated with impaired health and fertility. This led to the incorporation of functional traits into breeding aims to counteract this negative association. Potentially, distributing the total quantity of milk per lactation cycle more equally over time could reduce the peak of physiological strain and improve health and fertility. During lactation many factors affect the production of milk: food intake; digestion, absorption, and transportation of nutrients; blood glucose levels; activity of cells in the mammary gland, liver, and adipose tissue; synthesis of proteins and fat in the secretory cells; and the metabolic and regulatory pathways that provide fatty acids, amino acids, and carbohydrates. Whilst the endocrine regulation and physiology of the dynamic process of milk production seems to be understood, the genetics that underlie these dynamics are still to be uncovered. Modeling of longitudinal traits and estimating the change in additive genetic variation over time has shown that the genetic contribution to the expression of a trait depends on the considered time-point. Such time-dependent studies could contribute to the discovery of missing heritability. Only very few studies have estimated exact gene and marker effects at different time-points during lactation. The most prominent gene affecting milk yield and milk fat,
Lactation is an orchestrated process aimed at providing nutrition and immune protection to the offspring; however, the mother must also retain sufficient resources to ensure her own survival. Thus, the quantity and composition of milk produced is strongly dependent on the developmental stage of the offspring and the maintenance requirements of the mother. As such, milk production is a classic exemplar of a time-dependent dynamic process.
The domestication of animals inevitably led to selective breeding for increased productivity. The uninterrupted increasing global demand for dairy products necessitated a concurrent increase in milk production. Thus, in order to meet market requirements, the dairy sector implemented selective breeding programs which have led to a doubling in the amount of milk produced per cow over the last 50 years, such that total milk production is increasing despite a decline in dairy cattle populations (Food and Agriculture Organization of the United Nations, FAO, 2012
These breeding indices have enabled dairy farmers to breed for milk production and functional traits without requiring knowledge on how these practices impact upon the dynamics of milk production or change the expression of underlying genes. However, the continuous development of genetic and genomic tools, as well as computational capacities, will allow breeders of the future to base their decisions not only on phenotypically observable traits or indirect genetic marker information but also on the direct causative genetic variants.
As with many other complex traits important in livestock production, milk production is influenced by many genetic loci that act directly, interact with each other and/or interact with the environment (Lemay et al.,
The milk production of a cow follows a dynamic curve (Figure
From a nutritionist's point of view it might be necessary to reduce, rather than to increase, peak milk yield in order to decrease the energy deficiency experienced during early and peak lactation, and thereby improve health and fertility traits. However, this is in direct conflict with the desire to increase overall milk production. Therefore, an alternative method of increasing overall milk production might be via increasing production persistency. A better production persistency raises the overall gain per lactation due to an increased persistency affecting the longest part of the lactation (i.e., late lactation; Dekkers et al.,
Production persistency is most often defined as a lesser decrease in milk production after the peak, i.e., a flatter shape of the lactation curve compared to another animal or the heard average. Such calculations can be based on the difference of peak yield to a 305d measurement, on test-day deviations, or on parameters of lactation curve models (Gengler,
One problem with persistent production is the requirement to dry-off a cow between lactations. However, if the production system does not require yearly calving, the duration of the lactations can be chosen according to daily yield. Subsequently, with increased lactation duration, the time point of insemination has to be postponed. Assuming that the peak production remains around 6 weeks into the lactation cycle, a later time point for insemination has the added benefit that a new pregnancy begins after the energy deficit caused by the high peak production. Thus, fertility issues potentially arising from an energy deficit will be reduced. Regardless of lactation duration, the general recommendation for days dry is still 45–70 days (Kuhn et al.,
Current methods in animal breeding apply an index of traits weighted according to their economic importance as well as heritability in the breeding goal. Further, phenotypic and genetic correlations between traits within the index are included, on the one hand to increase accuracy on lowly heritable traits, and on the other hand to account for potential negative correlations (Dekkers,
The development of the mammary gland is the primary factor affecting milk production. A well-developed mammary gland with many fully differentiated secretory cells, good blood supply, and strong connective tissue will be highly productive over a long time.
The mammary gland forms a rudimentary duct tree during fetal development in response to maternal hormones (Watson and Khaled,
Hormones and growth factors are important in determining how many secretory cells develop, and thus how much milk can be produced in the mammary gland (Watson and Khaled,
A second crucial point for milk production concerns the quantity and quality of the secreted milk. Milk is an emulsion of fat and water containing dissolved carbohydrates, proteins, vitamins, and minerals that all have to be produced in or transported to the mammary gland. During lactation, quantitative milk yield is primarily regulated by lactose within the alveoli. Alveolar lactose influences the osmotic pressure between blood and alveoli and thereby the amount of water drawn into the alveoli (Figure
Proteins and fat are important for qualitative milk yield in terms of organoleptic properties of the milk and downstream industries such as cheese and butter production (Bailey et al.,
Lactation is coupled with changes in the activity of genes in the mammary gland but also in other organs. In the liver, fat and glucose synthesis is highly increased from pregnancy to early lactation to provide fatty acids and blood glucose for milk production (Figure
In conclusion, to understand the genetics behind a lactation cycle, a number of gene pathways need to be considered. These include genes regulating food intake and blood glucose levels; the digestion, absorption, and transportation of nutrients; the activity of the secretory cells in the mammary gland, liver, and adipose tissue; the synthesis of proteins and fat in the secretory cells; and the pathways which provide triglycerides, fatty and amino acids, transporter proteins, and transcription factors.
The establishment of the
Only around a dozen candidate genes have been consistently identified between studies and described more extensively with regards to their association with the main milk production traits (Table
3 | 80,071,689–80,147,000 | Milk yield Milk fat | Banos et al., |
|
4 | 93,249,874–93,266,624 | Milk yield Milk fat | Banos et al., |
|
5 | 66,532,879–66,604,699 | Milk yield Milk fat | Mullen et al., |
|
6 | 37,959,536–38,030,585 | Milk protein Milk fat | Cohen-Zinder et al., |
|
6 | 38,120,578–38,127,541 | Milk yield Milk protein Milk fat | Leonard et al., |
|
6 | 44,854,113–44,960,533 | Milk yield Milk fat | Khatib et al., |
|
6 | 87,141,556–87,392,750 | Milk protein | Velmala et al., |
|
14 | 1,795,425–1,804,838 | Milk yield Milk fat | Winter et al., |
|
15 | 59,164,519–59,200,908 | Milk fat | Zielke et al., |
|
18 | 22,118,201–22,541,540 | Milk fat Milk protein | Zielke et al., |
|
20 | 31,890,736–32,064,200 | Milk yield Milk protein Milk fat | Viitala et al., |
|
20 | 39,073,246–39,137,480 | Milk yield Milk protein Milk fat | Bole-Feysot et al., |
|
23 | 35,105,135–35,113,750 | Milk yield Milk protein Milk fat | Bole-Feysot et al., |
|
27 | 36,212,352–36,228,987 | Milk yield Milk fat | Bionaz and Loor, |
Only little is known in regards to time-dependent genetic effects causing a dynamic curve in dairy cattle but the next section summarizes the efforts and results made in this field.
Whether a single marker for a candidate gene is used or thousands of indirect markers for a GWAS, finding associations between markers and a trait that displays dynamic expression over time can be difficult. The simplest solution may be to estimate associations over various time-points, i.e., treat each measurement as a separate phenotype. Automated milking systems could provide an accurate measurement of milk production for each day of lactation. However, this approach would mean that several hundred phenotypes would have to be analyzed. Further, whilst such measurements would provide daily milk yield, persistency cannot be estimated from a single time point. Ergo, daily measurements should not be treated as separate phenotypes. Therefore, appropriate phenotypic and genetic correlations have to be incorporated or repeated measurement analyses performed. Whilst daily measurements provide a highly accurate description of lactation performance, it might be computationally too time-consuming to be practically applied. Further, milking systems have still not penetrated the entire dairy sector and analyses solely relying on daily measurements would require additional methods to include animals with missing records. Most countries with national evaluation networks record milk production once a month. Assuming the lactation period of a cow lasts for 340 days, one record a month sums up to approximately 11 test-days. Because crucial changes such as peak yield occur roughly 6 weeks into the lactation cycle, one analysis every month could still give a fairly thorough picture of the lactation performance. However, analyzing 50k or 800k markers (the marker number of the most commonly used SNP-chip in dairy cattle at present) for thousands of animals would still take time.
Instead of using the measurements of the actual test-days, fewer parameters can be sufficient to describe an entire lactation. The profile of milk production, and its components, over the course of a single lactation has been described by various mathematical and biological functions (Pollott,
Such approaches are known as functional modeling in human genetics where it is mostly applied to map dynamic loci affecting disease traits using growth curves such as cubic splines (Hou et al.,
The few results of time-dependent association studies in livestock reflect reported dynamic expressions of genes involved in milk production (Bionaz and Loor,
In general, the highest variation in associated loci were reported for early and late lactation, suggesting that those time periods provide the best opportunity for alteration through breeding schemes. This would also serve the idea of decreasing peak production through a slower increase in early lactation and increase the persistency of production in late lactation. Furthermore, by analyzing marker associations over time, we are more likely to find genetic markers with small effects over the whole lactation but strong effects at a specific time-point as they are not masked by major candidate genes such as
Finally, differences in genetic effects were not only found for different lactation stages but also between lactations, especially between the first and later lactations (Strucken et al.,
In animal breeding, the ability of an animal to improve a trait in the next generation can be summarized using an estimated breeding value (EBV). The current standard for breeding value estimation is to include the animal's own performance records as well as the records of relatives, assuming that related individuals share a certain amount of genes with each other. In milk production, obviously a bull does not produce milk, and therefore its EBV is entirely dependent on milk production records of female relatives.
Because milk production is routinely recorded once a month in most countries, EBVs are based on these monthly test-day data. To account for the fluctuation of milk yield throughout a lactation, test-day models have been developed through the incorporation of appropriate lactation curve models (Misztal et al.,
One possible solution could be to include the shape of the production curve into the selection index and set a standard curve based on how much milk a cow can produce without inducing an energy deficiency under natural feeding conditions, and how much milk a cow should produce to be still profitable for the farmer. Based on such a standard curve, breeding values could be weighted according to their deviation from the standard curve resulting in a single EBV per animal.
A similar approach could be applied for genomically estimated breeding values (gEBVs) where the information of the genome-wide markers along with the production records of all relatives are included. gEBVs seem to be the way forward as they use the genetic constitution of the animal itself, and thus, what is actually inherited from generation to generation (Goddard and Hayes,
Most of the reviewed studies on time-dependent genetic effects, applied a GWAS approach where the effects of each marker were estimated independently from all other markers. However, it is assumed that a quantitative trait such as milk production is shaped through the activity of many genes that might affect and even depend on each other. Thus, marker effects should not be estimated independently of all other markers in a study. A method which includes all markers at the same time has been termed Snp-BLUP, which is an extension of the original BLUP (best linear unbiased prediction) equation used to estimate the EBV of an animal (Goddard,
Information obtained through dynamic GWAS or Snp-BLUP would make it possible to weight each marker according to its effects on the dynamic expression of the trait at different time-points, and thus provide a gEBV that includes the shape of the production curve. With the reducing cost of sequencing, causal genomic variants may be discovered and ultimately used in animal breeding to perform the most accurate selection possible.
Milk production is a dynamic process and factors influencing this process occur as early as the fetal development. Whilst many physiological aspects of a dynamic milk production have been discovered, research on time-dependent genetic effects is still a wide open field. The animal breeding industry considers dynamic milk production by incorporating appropriate lactation curve models into their breeding value estimates to improve accuracy. Further, through an index of traits, breeders attempt to tackle the detrimental effects of a high milk production on other functional traits. However, if we assume that some of the negative issues arising from a high milk production can be overcome by altering the shape of the production curve, the impact of such an index on the actual dynamics of the lactation cycle are poorly understood. Since genetic and genomic tools are constantly developing with whole genome sequencing already being applied, our understanding of genes, their interactions and pathways will improve and direct causative mutations might be the target of future animal breeding programs. Understanding the time-dependent effects of genes and their variants is therefore an important field to study. Finally, whilst the dynamic of milk production is an obvious example, other time-dependent traits such as growth and weight gain, marbling, or onset of puberty could also benefit from a deeper understanding of the underlying dynamic of gene effects.
EMS conceived of the topic and wrote the manuscript. YCSML and GAB advised and critically revised the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank Dr. Adam Wilkins for his constructive comments on the manuscript, his teaching and time.
1