Livestock species are raised primarily for their economic benefit to humans. Most dairy cows are culled before they reach the end of their potential lifespan due to poor milk production or fertility and/or an increased prevalence of diseases such as those causing mastitis or lameness [3, 8, 9]. Various cow longevity indexes have been defined, some of which also take account of lifetime milk production, which is in turn affected by both the milk yield capacity and the number of lactations achieved . The main focus of previous studies into longevity in cows has been to increase the average survival time in the milking herd, so improving the profitability of the dairy industry. For example, one genome wide association study (GWAS) which investigated longevity found genes such as NPFFR2, previously identified as a candidate gene for mastitis resistance and two zinc finger proteins (ZNF613, ZNF717), which have been associated with calving difficulties . Information about the age-related morbidities and causes of death that afflict cattle due to natural ageing is, however, limited. In contrast, there is a growing body of previous work into the underlying causes of cellular ageing which has been based on studies of human populations and model organisms.
This study is the first, to our knowledge, to assess changes in the global gene expression in leukocytes of dairy cows associated with increasing lactation number. For this we compared first lactation PP cows versus older multiparous MP > 3 cows using WCGNA analysis in order to identify potential genes and related pathways involved in the ageing process. The blood samples were all collected in early lactation, a time when lactating cows are placed under metabolic stress due to the nutrient requirements of milk production [11, 38] but none are pregnant. These aspects improved comparability between samples but meant that we did not include any data from younger growing heifers of < 2 years. The general aim of most dairy farmers is to calve animals for the first time at 2 years of age and then to achieve an annual calving interval, although in practice these targets are rarely met . Although we did not record the exact ages of the cows in this study, the expectation would be for the PP cows to have been between 2 and 2.5 years old, the MP2–3 between 3.5–6 years old and the MP > 3 (which were in lactations 4–7) would have been between 6 and 10 years old. In most countries worldwide, dairy cow longevity has declined over the past 50 years and has been negatively related to the rise in milk yields achieved over the same period . Despite the many differences in physiology, lifestyle and lifespan between species, many of the genes and pathways identified in our study as being associated with increased lactation number in dairy cows were nevertheless the same as those highlighted in previous studies of ageing in other species [18,19,20, 23].
All the samples analysed were whole blood leukocytes. In dairy cows the average leucocyte population is between 5 and 12 cells per ml of blood, predominantly consisting of lymphocytes and neutrophils . The blood does, however, also contain natural killer cells, platelets, peripheral blood mononuclear cells (PBMC), eosinophils, and basophils. These different cell types all vary in both their basal gene expression and their transcriptional amplification responses to particular stimuli . Furthermore, many aspects of the immune system alter during ageing, eventually leading to immunosenescence . During this process different immune cell subsets are affected in different ways . There is also an increase in baseline systemic inflammation with age, termed “inflammaging” [42, 43]. This may arise as a result of multiple mechanisms, including the accumulation of misfolded proteins, impaired clearance of senescent cells and obesity. Most individual leukocytes have a short lifespan in the circulation, measured in days or weeks, before they are destroyed by the lymphatic system, although there is a small pool of long-lived T and B lymphocytes which can survive for years, providing immunological memory. In humans, the relative abundance of naive T-cells decreases with chronological age while the population of memory T-cells increases [18, 44, 45]. Another possible source of variation in our analysis was the potential differences between herds, but this was accounted for by removing the batch effect in the WGCNA. Two of the six herds also fed three different lactational diets. Although these did influence cow metabolism this was not yet apparent at Day 14 when the samples used in the present study were collected. A comparison of the transcriptome in PBMC collected from the same UK herd failed to detect any DEG attributable to diet at this time .
The leukocyte transcriptome thus provides a reflection of the basic functions required for cell survival together with the various responses of the different cell types to the changing environment within the body, which alters with key factors such as disease exposure and nutrition. Cells may proliferate, undergo apoptosis or migrate to or from tissues into blood in response to different signals . The changes reported here which were associated with lactation number will, to some extent, reflect altered abundance of different cell types as well as their changing expression patterns. Despite this caveat, many previous studies have now reported that transcriptional signatures of whole blood can reliably differentiate individuals with a variety of infections, for example Johne’s disease (Mycobacterium avium subsp. Paratuberculosis) in cattle . Our analysis also benefitted from the use of WGCNA, which identifies networks of co-expressed genes whose expression is highly correlated. The use of whole blood also avoided the pitfall of potential artefacts which can be induced during cell separation procedures and provided greater consistency in a multi-site study.
Previous work based on studies of human populations and model organisms has identified a number of transcriptional signatures for cellular ageing which occur repeatedly across different tissues and organisms and which segregate into six main groups [24, 49]. In brief these are: i) downregulation of genes encoding mitochondrial proteins; ii) downregulation of the protein synthesis machinery (including ribosome biogenesis); iii) dysregulation of immune system genes, immune senescence; iv) reduction in growth factor signalling; v) constitutive responses to stress and accumulated DNA damage, and vi) dysregulation of processes regulating gene expression and mRNA processing (transcription and translation). We have obtained evidence that all of these were associated with lactation number within our population of cows.
Mitochondria and oxidative stress
Mitochondria regulate a multitude of different metabolic and signalling pathways and also play an important role in programmed cell death . Oxidative metabolism causes endogenous production of free radical molecules and oxidative damage accumulates in multiple tissues and species with age . For example, accumulated mutations in somatic mitochondrial DNA (mtDNA) and respiratory chain dysfunction were associated with ageing in mice . In our study, genes encoding proteins involved in fatty acid beta-oxidation were more highly expressed in leukocytes in PP cows. There were also changes in expression of genes encoding mitochondrial ribosomal proteins, with higher expression of MRPS9, MRPS27 and MRPS31 in leucocytes of PP cows whereas MRPL17 and MRPL38 showed enriched expression in MP > 3 cows.
Protein homeostasis is essential to maintain protein structure and function, but the control of this process declines during ageing . In our study, GO enrichment analysis of all the genes in the violet module, which were more highly expressed in the first lactation cows, revealed their involvement in protein autophosphorylation and catalytic activity (enzyme activity). Autophosphorylation is a type of post-translational modification of proteins. In eukaryotes, this process occurs by the addition of a phosphate group to serine, threonine or tyrosine residues within protein kinases, normally to regulate the catalytic activity. Genes involved in regulation of nitrogen compound metabolic process were identified in the yellow module as being down-regulated in the older MP > 3 cows. Maintenance of the proteome is essential to enable cells to respond appropriately to their environment. This requires correct synthesis and assembly of proteins and is controlled by molecular chaperones and clearance mechanisms which help to prevent protein misfolding and the associated accumulation of toxic aggregates. The efficiency of this process declines with age and has previously been associated with both metabolic and immunological diseases [49, 52].
As we were studying a leukocyte population, changes in gene expression relating to immune function were expected. As animals age they are exposed to an increasing variety of disease causing microorganisms, while a progressive loss of function of the immune system increases their vulnerability to infection . Notable age-related changes within the immune cell population include reduced cytokine signalling, diminished production of nitric oxide and peroxide, decreased phagocytic ability and reduced ability of dendritic cells to migrate and process antigens . Our results are in accord with the study by Peters et al. , who investigated leucocytes from ageing human populations, in finding pathways which were either up- or down-regulated with increasing age. The GO term immune system process was enriched in both PP and MP > 3 cows, with genes involved in adaptive immunity up-regulated in the PP cows (FYN, ITK, LCK, etc) and genes related to innate immunity up-regulated in MP > 3 cows (CTSS, CTSH, IRAK2, TLR2, etc). The black module was down-regulated in the PP cows and contained genes involved in the regulation of cytokine secretion (IL10, CD14, FGR, IL17RC). Expression of genes in the turquoise module increased in older cows, containing genes involved in neutrophil degranulation and innate immune system. The darkred and midnightblue modules were both more highly expressed in the PP cows and contained the terms disease and immune system.
There was also a significant overlap between the DEG with immune functions identified in our population of cows and candidate age-related genes in the human transcriptome . Those up-regulated in PP cows were mainly associated with T-cell development and function (CCR7, CD27, IL7R, CAMK4, CD28) while those up-regulated in MP > 3 cows included genes encoding proteins involved in immune defence. Of these LYZ encodes lysozyme, an antimicrobial peptide, CTSZ encodes cathepsin Z a lysosomal cysteine protease with multiple roles in host immune defense mechanisms, SREBF1 encodes a transcription factor involved in TLR4 signalling, while GRN encodes granulin, involved in TLR9 signalling, ANXA5 is involved in T-cell activation and ADARB1 encodes a deaminase enzyme with A-to-I RNA editing activity, which is important for the maintenance of cellular health but may also play a role in response to viral infection. The results from our study therefore suggest that the higher lactation number cows are more actively engaged in combatting disease pathogens through activation of the innate immune system and also support a higher level of inflammation with ageing. Parturition is itself an inflammatory process and, in addition to higher rates of infectious diseases including mastitis and metritis, older cows are also more vulnerable to metabolic disorders including milk fever, ketosis and displaced abomasum after calving. The prevalence of all of these metabolic diseases increased significantly in cows which were ≥ 5 lactations . Our previous study showed that metabolic disorders led to prolonged uterine inflammation by up-regulating the genes and pathways related to immune and inflammatory processes .
Growth factor signalling
The relationship between ageing and metabolic regulation is bidirectional, as ageing impairs the activity of key metabolic signalling pathways and the ensuing metabolic dysregulation results in accelerated ageing . Cell signalling pathways that sense the availability of nutrients and the energy status of the cells communicate with hormonal and growth factor signalling pathways to co-ordinately regulate whole body metabolic homeostasis. Ageing results in a gradual deterioration of various cellular functions including metabolic regulation . The turquoise module was the second highest correlated module with lactation number, containing genes more highly expressed in MP > 3 cows. The insulin-like growth factor binding protein pathway was indeed enriched with genes like INSR, IGF1R, IGF1, LDLR, HTRA1, IGFBP7 etc. (Cluster 8; Supplementary Table 4). The latter pathway is interlinked with metabolic pathways to ensure coordinate regulation and fine-tuning of cellular metabolic responses in line with cellular energy status, nutrient availability and hormonal/growth factor signalling input . We have shown previously that circulating IGF1 concentrations are significantly lower in PP compared with MP cows, falling to a lower nadir in the first week after calving .
Stress and DNA damage
Accumulation of genetic damage represents one of the major contributions to ageing of cells and organisms. Cellular DNA is constantly exposed to exogenous and endogenous DNA-damaging agents like reactive oxygen species, nitric oxide metabolites, and alkylating agents  leading to accumulation of mutations in the genome aggravated by loss of capacity in the DNA repair systems . DNA damage is tightly linked to various ageing stresses, such as oxidative stress, telomere shortening, inflammation, irradiation, exposure to chemicals, and mitotic stress . This is supported by a recent study which took repeated measurement of the relative leukocyte telomere length in a dairy herd. Higher rates of telomere attrition in individual cows was predictive of a shorter productive lifespan, suggesting a link between telomere loss and health .
In our dataset, NELL2 was the gene most highly correlated with lactation number in the turquoise module and was also overall the most differentially expressed gene between the PP and MP > 3 cows, with greater leucocyte expression in the older animals. The encoded protein Neural EGFL Like 2 is highly conserved in mammals and is a glycoprotein containing several von Willebrand factor C and epidermal growth factor (EGF)-like domains. This has a variety of possible roles but amongst these a cell survival-promoting effect mediated by an intracellular mitogen-activated protein kinase (MAPK) pathway has been relatively well studied in neural tissues [59, 60]. NELL2 is also important in protecting cells from death caused by endoplasmic reticulum (ER) stress resulting in the accumulation of unfolded proteins which trigger the unfolded protein response. Within this context, overexpression of NELL2 decreased expression of ER stress-induced C/EBP homologous protein (CHOP) and the pro-apoptotic caspases 3 and 7 while increasing expression of ER chaperones and anti-apoptotic Bcl-xL .
Another relevant gene to consider is MTOR, which encodes Mechanistic Target of Rapomycin, a protein belonging to a family of phosphatidylinositol kinase-related kinases which mediate cellular responses to stresses such as DNA damage and nutrient deprivation. This gene is a central regulator of metabolic homeostasis and is associated with lifespan in many species [62, 63]. MTOR is a component of two distinct complexes of which mTORC1 controls protein synthesis, cell growth and proliferation. Genes which are part of the MTOR pathway include the transcription factors FOXO1 and FOXP1. A GWAS for longevity, based on a population of Holstein cows, previously identified a region on Bta16 containing MTOR . In our study FOXP1 expression was positively related to increasing lactation number, whereas FOXO1 expression was negatively related. Also, of potential relevance here is the opposing roles of these two FOX genes in the regulation of glucose homeostasis, through competition in binding to the insulin response element in gene promoters. Up-regulation of FOXP1 in mice inhibited the hepatic expression of key gluconeogenic genes, including PGC-1α, PEPCK and G6PC . LAMTOR1, − 2 and − 3 all also featured in the list of DEG which were negatively related to lactation number in our study. These genes encode late endosomal/lysosomal adaptor, MAPK and MTOR activator-1, − 2 and − 3 respectively, all subunits of the Ragulator complex. This functions as a lysosome anchor, which recruits Rag GTPase and its associated mTORC1 complex to the lysosomal surface prior to MTOR activation .
Regulation of gene expression
The control of gene expression becomes more dysregulated with cellular ageing. A large number of genes and pathways identified in this study are involved in regulation of gene expression. The violet module was the most highly correlated with lactation number and contained a group of genes involved in transcriptional regulation including ZNF462, SOX13, and SOX4, all of which featured individually in the top 20 genes whose expression was most highly correlated with age (Table 1). Genes in the orange module were more highly expressed in PP cows and this module was enriched with genes involved in gene expression (CHTOP, CPSF6, THOC1, UPF3B, etc.) and metabolism of RNA (AMDHD1, SRSF2, SRSF4, SRSF5, etc.). In the yellow module IGF2BP3 was down-regulated in the older cows and was the most highly correlated gene. IGF2BP family members were initially identified as post-transcriptional regulators of IGF2. They are RNA-binding proteins which direct nuclear RNA export and translation/degradation rates, so playing a major role as regulators of the RNA life cycle. IGF2BP3 has recently risen to prominence as a potential oncogene . Other genes in this module were also grouped under regulation of gene expression (TGFB3, NEUROG2, ZNF554, etc.). The yellow module also contained the gene WRN, which exhibited lower transcript abundance in MP > 3 cows. A mutation in this gene in humans causes Werner Syndrome, an autosomal recessive disorder characterized by the premature development of ageing features. The encoded protein is a member of the RecQ family of proteins and is involved in DNA replication and repair, and telomere maintenance, so playing a crucial role in genome stability . Expression of WRN was similarly reduced in leucocytes of older humans . ADARB1 was another candidate gene from previous studies which was up-regulated in the older MP > 3 cows. It encodes a deaminase enzyme with A-to-I RNA editing activity, was previously identified in a study of men aged 90–119 years, and is also associated with longevity in C. elegans .
One key difference between PP and MP > 3 is that milk production potential in dairy cows increases with age . The liver coordinates the extensive metabolic changes required for milk production and these are reflected in circulating metabolite concentrations [11, 38]. Milk synthesis has a high requirement for glucose. In ruminants this demand is met almost exclusively through hepatic gluconeogenesis and cows are at risk of glucose insufficiency during early lactation, the time period we investigated . NEFA are released from lipid stores as an alternative energy source and are either used by the udder to provide milk triglycerides, fully oxidized in the liver to provide energy, or partially oxidised resulting in the production of ketone bodies, in particular BHB. Circulating BHB concentrations are thus an index of fatty acid oxidation and concentrations are significantly higher in older cows . BHB, NEFA and glucose concentrations can all influence leukocytes directly. Immune cells require an adequate supply of nutrients including glucose to mount an effective immune defense . Neutrophils from cows with more elevated NEFA and BHB concentrations after calving had reduced expression of genes important for granulocyte recruitment, IFN signaling and apoptosis . This suggested that neutrophil survival time was longer in the circulation when exposed to pro-inflammatory conditions. Another study of the circulating leucocyte transcriptome in early lactation found that expression of genes in KEGG pathways relating to DNA replication, cell cycle, homologous recombination, base excision repair, and valine, leucine, and isoleucine biosynthesis were all inhibited as plasma BHB increased, whereas genes involved with vitamin metabolism, the endocrine system, signalling molecules and the immune system were activated .
In this study genes in both black and turquoise modules were negatively significantly correlated to cow parity and David functional annotation cluster analysis found enrichment of the terms fatty acid metabolism and glycolysis. Interestingly, we found that the majority of 27 genes in Cluster 4, with higher expression in MP > 3 cows, were involved in the Glycolysis/Gluconeogenesis pathway (Fig. 3). Although the underlying metabolic background is very different, up-regulation of a small cluster of genes relating to “Fatty acid metabolism, peroxisome activity” was also associated with ageing in the human blood transcriptome .
Other key genes associated with lactation number
Within the violet module LAMA4 was the most highly correlated gene overall, with greater expression in the PP cows. This encodes a subunit of laminin, part of the family of extracellular matrix glycoproteins which form the major non-collagenous constituent of basement membranes. Laminin is thought to mediate the attachment, migration and organization of cells into tissues by interacting with other extracellular matrix components, suggesting that these activities may be reduced in older cows. There is a second MTOR complex mTORC2, which acts as a regulator of the actin cytoskeleton. This is a network of actin and actin binding proteins which are important for a range of essential cellular processes including organelle transport, cell migration, phagocytosis, and cell cycle progression. This is in agreement with the blood transcriptomic study in humans  in which expression of a cluster of genes relating to the actin cytoskeleton and focal adhesion also increased with ageing (ACTA2, ACTN4, CSRP1, ILK, LPP, TAGLIN, TLN1, VCL and WDR1). In contrast, ARHGAP15 was more highly expressed in younger cows (this study) and in humans . This gene encodes Rho GTPase activating protein 15 which is involved in T- and B-cell signalling and promotes an increase in actin stress fibres and cell contraction .