The effect of N-arachidonoylethanolamide administration on energy and fat metabolism of early lactating dairy cows

The aim of the study was to investigate the effect of N-arachidonoylethanolamide (AEA), an endocannabinoid with orexigenic characteristics, on plasma endocannabinoid concentrations, feed intake, energy balance, lipomobilisation, and hepatic lipid metabolism of early-lactating dairy cows. The experiment involved 10 pairs of Holstein half-sibling cows (end of 2nd–3rd pregnancy). Half-sibs of each pair were randomly assigned to either AEA (n = 10) or control (CON) group (n = 10). From day 1 to 30 postpartum, the AEA group received 5 intraperitoneal injections per week of 3 µg/kg body weight AEA and the CON group 0.9% NaCl. In week 1–3 postpartum, AEA administration had no effect on dry matter intake, body weight, or lipomobilisation, but increased plasma triglyceride concentration on d 21 p.p. and mRNA abundances of genes related to hepatic triglyceride synthesis. In week 4 postpartum, the AEA group showed reduced feed intake and whole-body carbohydrate oxidation, but increased whole-body fat oxidation and hepatic lipid accumulation, likely as a result of a counter-regulatory leptin increase. In conclusion, the present study shows a tissue-specific AEA insensitivity and may point to a leptin-controlled regulation of the ECS in early-lactation.


Body anatomy and body condition.
The heart girth (Supplementary Fig. 2a) and body condition score (BCS) (Supplementary Fig. 2b) did not differ between groups, in neither the a.p. nor p.p. period, except for the heart girth in week 4 p.p., which tended to be higher in AEA cows.However, AEA cows tended to have higher amounts of estimated mesenteric adipose tissue in week 4 a.p., (P = 0.083; Fig. 2a) and more mesenteric adipose tissue in week 3 and 2 a.p. (P < 0.050; Fig. 2a).In addition, AEA cows tended to have more omental adipose tissue in week 3 a.p. and week 2 p.p. (P < 0.100) and had more omental adipose tissue in week 3 p.p. (P = 0.040; Fig. 2b).By contrast, the amounts of retroperitoneal adipose tissue, subcutaneous adipose tissue, and total abdominal adipose tissue (Fig. 2c-e) were not different between groups, but subcutaneous adipose tissue tended to be affected by a group by time interaction before parturition (P = 0.078).The thickness of the fat layer over the 12th rib and the back fat thickness were not affected by group before and after parturition (P > 0.1; Fig. 2f,g).However, mBW declined during the first 4 weeks of the p.p. period more in AEA than CON cows (group x time interaction: P = 0.021; Supplementary Fig. 2c).
Plasma EC and NAE concentrations.Next, we examined the effect of postparturient i.p.AEA administration on EC concentrations in the circulation.Before start of AEA administration, i.e. on day (d) 10 a.p., plasma AEA concentrations did not differ between groups (Fig. 3a).After start of the AEA injection series, plasma AEA concentrations were 1.72-to 1.81-fold higher in AEA than CON cows (P < 0.05).However, the plasma concentrations of 2-AG, LEA, OEA, PEA and DHEA were not affected by AEA treatment at any time p.p. (Fig. 3b-g), except for EPEA, which tended to be higher in the AEA group on d 14 p.p. (P = 0.094, Fig. 3e).In addition, plasma AEA, 2-AG, LEA, OEA, EPEA and PEA concentrations increased, while the DHEA concentration decreased in both groups from the a.p. to the p.p. period (Fig. 3g).

Plasma metabolite concentrations.
Plasma glucose and urea concentrations, NEFA, high-density lipoprotein cholesterol and TG were comparable between groups (P > 0.1; Fig. 4a-e), although AEA cows had higher TG concentrations on d 21 p.p. (P = 0.040, Fig. 4e).Similarly, plasma β-hydroxybutyrate and leptin concentrations, as well as the phospholipid transfer protein (PLTP) activity did not differ between groups (P > 0.1; Fig. 4f-h), except on d 28 after calving, on which AEA cows had a higher leptin concentration (P = 0.026, Fig. 4g) and tended to have a lower PLTP activity compared to CON cows (P = 0.087, Fig. 4h).www.nature.com/scientificreports/Ex vivo adipose tissue lipolysis.We observed no group differences in the amount of glycerol released of from cultivated subcutaneous adipose tissue after stimulation with norepinephrine, neither a.p. nor p.p. (Supplementary Fig. 3).stained liver sections (Fig. 5c,d).The percentage of the lipid droplet area of the total area in the liver tended to be higher in the AEA compared to the CON group (P = 0.054, Fig. 5a), however, the number of lipid droplets per mm 2 liver tissue on d 30 p.p. did not differ between the groups (P = 0.401, Fig. 5b).In regard to hepatic fat metabolism, diacylglycerol O-acyltransferase 1 (DGAT1) and diacylglycerol O-acyltransferase 2 (DGAT2) expression were lower in AEA compared to CON cows (P < 0.05, Table 1), whereas the transcript levels of ATP binding cassette subfamily a member (ABCA1), apolipoprotein B (APOB100), www.nature.com/scientificreports/microsomal triglyceride transfer protein (MTTP), sterol regulatory element binding transcription factor 1 (SREBF1) and peroxisome proliferator activated receptor alpha (PPARA ) were comparable between the groups on day 16 p.p. Before parturition, the hepatic mRNA expression of genes involved in the ECS (CNR1, GPR55, FAAH, NAPE-PLD) were not different between groups (P > 0.1, Table 1).In addition, genes related to fat metabolism (ABCA1, APOB100, MTTP, SREBF1, PPARA, DGAT1, DGAT2) did also not differ between groups.On day 16 of the p.p. period, the GPR55 expression level tended to be lower in the AEA compared to the CON group (P = 0.091), however, AEA administration had no effect on CNR1, FAAH and NAPEPLD mRNA expression levels (P > 0.1, Table 1).
The mRNA expression analyses in liver obtained after slaughter revealed a higher FAAH abundance in the AEA compared to the CON group (P = 0.028, Table 1).In addition, MTTP mRNA abundance tended to be lower in AEA cows (P = 0.061), whereas AEA administration did not affect the abundance of genes involved in the ECS.
Mammary gland ECS and fat metabolism.The relative mRNA expression levels of genes related to ECS and fat metabolism were not different between the AEA and CON group in the mammary gland p.p., however, acetyl-CoA carboxylase alpha (ACC1) mRNA expression was lower in the AEA than the control group (P = 0.031, Table 2).
Short-term feed intake and energy metabolism in the fourth week postpartum.On d 27 p.p., cows were transferred to a respiration chamber to measure the short-term responses of feed intake and energy metabolism after AEA administration.Cumulative DMI normalized to mBW was not different in the first two h after feeding, but tended to be higher in the CON group 3 h after feeding (P = 0.076) and was higher than in the AEA group 4-7 h after feeding start (P < 0.05; Fig. 6a).Moreover, FOX normalized to mBW decreased more, and COX normalized to mBW increased more in CON cows to differ from AEA cows after feeding (P = 0.019, Fig. 6b,c).However, HP/mBW did not differ between groups at any time post feeding (P > 0.1, Fig. 6d).

Discussion
In the present study, repeated administration of AEA was applied to increase plasma AEA concentrations in half-sib early-lactating cows.Activation of the ECS was expected to increase dry matter intake and decrease lipolysis, as already described in studies with rodents 17,[30][31][32][33][34] .

Effect of AEA administration on the plasma concentration of endocannabinoids and endocannabinoid-like compounds.
In our study, we were able to successfully increase the plasma AEA concentration in early-lactation by repeated daily i.p. injections of 3 µg AEA/kg BW in the first four weeks of lactation.The treatment was specific to AEA, because it did not affect the plasma concentration of other endocannabinoids or endocannabinoid-like compounds, except for EPEA, which tended to be elevated on d 14 p.p. To date, no study has investigated, if AEA administration influences plasma EPEA concentration.In rodents, diet composition can affect EPEA levels in plasma and adipose tissue 35 , but all animals in our study were fed the same diet and consumed comparable amounts of feed.Thus, it remains elusive whether AEA administration or an unknown influencing factor caused this temporary increase in EPEA.The latter is a CNR1 and CNR2 agonist 3 and possess anti-inflammatory properties 36 ; however, further studies are needed to evaluate the role of EPEA on earlylactating cows.
Effect of AEA on energy balance and hepatic gene expression in the first three weeks after parturition.Because we observed major differences in energy metabolism between weeks 1-3 p.p. and weeks 4-5 p.p., and that the transition period is defined 3 weeks before until 3 weeks after calving 37 , the discussion distinguishes between these 2 p.p. periods.Despite elevated plasma AEA concentrations, we did not observe an increase in feed intake during the first three weeks after calving in the AEA group.This result is in contrast to a study in rodents, which used a comparable low AEA dosage, but non-lactating animals 31 .Furthermore, AEA administration at similar dosage increased feed intake in mid-and late-lactation cows, although this effect was limited to 1 h after i.p. injection and to 10 h after intracerebroventricular injection 14,15,28 .The authors of these studies explained this short-term effect with the short half-life of AEA, because it is rapidly inactivated by reuptake and degrading enzymes 38 .It is conceivable that the AEA treatment in the present study also caused a short-term increase in feed intake but this effect could not be detected on the daily basis.Furthermore, it has to be taken into account that the endocrine status differs substantially between early, late and non-lactating animals.Another important factor is that the sensitivity towards various hormones regulating feed intake, i.e., leptin, and perhaps also AEA, is diminished in early-lactation.Finally, we cannot exclude that AEA administrations on only 5 days a week affected the results.
Administration of AEA had also no effect on milk yield, ECM yield, and milk constituents, which together with the unaltered feed intake resulted in a comparable energy balance between the groups in the first 3 weeks p.p.This finding is underlined by comparable plasma NEFA and β-hydroxybutyrate concentrations between the groups.A previous study reported also no changes in plasma NEFA and β-hydroxybutyrate concentrations after i.p.AEA administration to late-lactating cows 15 .However, a reduction in milk yield after intracerebroventricular injection was observed in mid-and late-lactating cows 16 .
Previous studies in rodents, humans, and cows reported the involvement of the ECS in the regulation of lipolysis 17,27,39 .In early-lactating cows, fat mobilisation occurs due to the negative energy balance, resulting in a loss of adipose tissue mass and increase of plasma NEFA concentration.To our knowledge, the present study is the first to investigate the influence of AEA administration on fat depots in vivo.In our study, we did not observe a clear effect of AEA treatment on various fat depots and fat layer thicknesses, although the amount of omental adipose tissue tended to be higher in the AEA group in week 2 p.p. and was higher in week 3 p.p.One explanation could be that omental fat depots have a different sensitivity towards AEA than the other tissue types investigated, and this assumption is supported by the finding that various adipose tissue types reveal different responsiveness to lipolytic signals 40,41 .However, mesenteric fat depots were greater already in the a.p. period and could have affected cow's metabolism p.p.A further limitation of the study is that we had no information about milk yield in the previous lactation which might have influenced body fat accretion a.p.
The plasma NEFA concentration is a marker for fat mobilisation 42 .The plasma NEFA concentration were not different between the first 3 weeks p.p.Moreover, we did not observe any effect on the norepinephrinestimulated in vitro lipolysis in week 2 p.p.This finding is consistent with a previous study showing no changes in isoproterenol-induced lipolysis after CNR1 receptor agonist administration 27 .Overall, our and earlier results indicate that AEA has no or only a limited effect on lipomobilisation in early-lactation.This finding supports the hypothesis proposed by Myers et al. 27 , who suggested a resistance to ECS activation in adipose tissue of earlylactating cows 27 .The authors reported no effect of CNR1 receptor activation on the lipolysis rate in adipose tissue explants from cows obtained 1-3 weeks p.p. 27 .
Lipid mobilisation and the resulting increased influx of NEFA into the liver can exceed the hepatic capacity for oxidizing fatty acids, thus leading to lipid accumulation in liver of cows during early-lactation 43 .In rodents, the activation of the ECS increases fatty acid synthesis and lipid deposition in the liver 20 .In the present study, we showed that i.p.AEA administration decreased hepatic GPR55 mRNA expression.Downregulation of G-protein Table 1.Relative mRNA abundances in the liver of cows treated intraperitoneally with N-arachidonoylethanolamide (AEA, n = 9) or NaCl (CON, n = 10) postpartum (p.p.).Tissue biopsies were obtained on d 25 antepartum (a.p.), d + 16 p.p. and on d 30 p.p. after slaughter and are presented as means ± SD.DGAT1/2 diacylglycerol O-acyltransferase 1/2; ABCA1 ATP binding cassette subfamily a 1; APOB100 apolipoprotein B; MTTP microsomal triglyceride transfer protein; SREBF1 sterol regulatory element binding transcription factor 1; PPARA peroxisome proliferator activated receptor alpha, CNR1 cannabinoid receptor 1; GPR55 G protein-coupled receptor 55; FAAH fatty acid amide hydrolase; NAPEPLD N-acyl phosphatidylethanolamine phospholipase D. www.nature.com/scientificreports/coupled receptors is induced, among others, by prolonged exposure to an agonist 44,45 .Therefore, the diminished GPR55 expression in the present study is likely due to chronic exposure of AEA, a GPR55 agonist 46 .To date, the physiological role of GPR55 in the liver of ruminants is unresolved.In humans and rodents, GPR55 is involved in the regulation of hepatic lipid metabolism 47,48 and insulin signalling 49 .Thus, the lower GPR55 mRNA abundance suggests that administered AEA alters hepatic lipid metabolism through GPR55 in cattle.Indeed, AEA treated cows showed reduced DGAT1 and DGAT2 mRNA abundance.Inhibition of DGAT1 has been shown to result in less lipid droplet accumulation and TG concentration in primary calf hepatocytes after incubation with fatty acids.Thus, repeated AEA treatment may limit TG accumulation in liver and thus the progression of fatty liver in week 3 p.p.In addition, we observed higher plasma TG concentration on d 21 p.p. in the AEA group.In mice, activation of the ECS resulted in higher plasma TG and cholesterol levels by impairing apolipoprotein E-mediated clearance 50 .In contrast to the finding in mice, we did not observe increased plasma cholesterol levels.In general, the increase in plasma TG can be induced either by increased synthesis of the liver or by decreased utilization in the mammary gland.Due to the comparable milk yield and milk fat content between the cow groups in week 3 p.p., a change in TG utilization for milk fat synthesis is rather unlikely.The synthesis and secretion of TG in cows occurs primarily in the liver through the re-esterification of NEFA and subsequent TG export as VLDL.The increase in plasma TG concentrations in week 3 p.p. might be due to an increased VLDL synthesis and secretion.Yet, we observed no difference in the mRNA expression of APOB100 and MTTP, both genes involved in VLDL assembly 51 , but we cannot exclude that other hepatic genes related to VLDL export account for the different plasma TG concentrations.
Effect of AEA on energy balance, hepatic and mammary gene expressions in week 4 and 5 postpartum.Unexpectedly, AEA treated cows stopped increasing feed intake in week 4 p.p, while feed intake, as expected, further increased in the CON group.The reduced feed intake in the AEA group was also reflected by the lower milk protein percentage as previous studies showed a decrease in milk protein content during restricted feeding 52 .Moreover, the insufficient feed intake of the AEA group was also accompanied by a higher proportion of the lipid droplet area of the liver, indicating a higher lipid accumulation in the AEA group.
In rodents and cows, AEA administrations either increased or did not affect short-term feed intake 14,16,31,53 , but there are no previous studies reporting a decrease in short-term feed intake in mammals.Here we report that long-term administration of AEA decreases feed intake of cows after 4 weeks of treatment.One possible explanation for the reduction in feed intake could be that the long treatment duration induced a contra-regulatory mechanism in which plasma leptin concentrations increased in AEA cows in week 4 p.p.In rodents, humans and ruminants, leptin decreases feed intake and controls energy balance [11][12][13][54][55][56] . Becaus plasma leptin concentrations are usually determined by the amount of stored triglycerides 57 , a negative energy balance and fat mobilisation results in a decrease in plasma leptin concentration 58 .However, in our study we found that the AEA group had a similar BCS throughout the experimental period and a similar amount of adipose tissue as the CON group, except for minor differences in omental adipose tissue in week 2 and 3 p.p.These facts cannot explain the abrupt increase in leptin concentration in week 4 p.p., and indicates that plasma leptin concentrations do not correspond to adipose tissue mobilisation.
There is evidence that leptin can modulate the AEA level in non-ruminants 10,59 .Hence, it is conceivable that the increase in plasma leptin concentration in week 4 p.p. could be a counter-regulatory response to the chronic AEA administrations.However, no study has proved this hypothesis and thus further studies are needed to elucidate the interaction between the AEA tone and leptin release.
In the liver, the FAAH mRNA abundance was higher in the AEA group than in the control group on d 30 p.p. FAAH encodes the enzyme responsible for the degradation of AEA 60 and thus, upregulation of FAHH implies increased degradation of AEA.Unfortunately, we were not able to measure the AEA concentration on d 30 p.p. to support this assumption.Previous studies reported increased FAAH mRNA abundance after AEA administration 61 .However, another reason for the change in FAAH mRNA abundance could be the higher leptin  concentration in the AEA group.In rodents, it has been shown that i.p. administration of leptin increased FAAH activity and thus AEA hydrolysis in the hypothalamus, however, FAAH gene expression was unchanged and the authors proposed a post-translational mechanism increasing FAAH activity 59 .From these findings, we conclude that upregulation of the FAAH mRNA abundance was either triggered directly by repeated administration of AEA or indirectly via leptin.The analysis of mRNA of genes related to lipid metabolism revealed a tendency to a lower MTTP mRNA abundance in the AEA group on d 30 p.p.The MTTP is involved in apolipoprotein assembly and the export of TG from the liver 62 .Bremmer et al. reported a trend for a negative correlation between MTTP mRNA expression and liver TG concentration on d 35 p.p. 64 .Similarly, cows with higher liver fat concentrations had a reduced MTTP mRNA expression level compared to controls 65 .In the present study, MTTP downregulation was also accompanied by higher lipid accumulation, as reflected by the trend to greater lipid droplet area in the liver of AEA cows; but whether AEA or the liver fat content regulates MTTP expression is not known.Furthermore, the biological significance of an altered MTTP expression remains elusive, because a change in MTTP mRNA abundance is not necessarily accompanied by a change in MTTP activity 63 .The higher lipid accumulation on day 30 was probably also the result of the lower feed intake and the numerically higher plasma NEFA concentration of the AEA group.Another possible explanation could be that AEA administration directly promoted hepatic fat accumulation, as CB1 activation stimulated the expression of lipogenic genes in the liver of mice 20 .However, in our study the expression of genes involved in lipid metabolism was not altered on day 30 p.p., except for MTTP.
As mentioned above, activation of the ECS regulates lipid metabolism in the liver and adipose tissue of rodents and ruminants, and involves among others upregulation of ACC1 in lipogenic tissues 20 .However, little is known about the influence of endocannabinoids on the metabolism of the mammary gland.In the present study, AEA cows have a lower mammary gland ACC1 mRNA abundance, which is involved in de novo milk fatty acid synthesis 66 .Surprisingly, downregulation of ACC1 was not accompanied by a reduction in milk fat content.Perhaps the de novo milk fat synthesis from acetate and butyrate was inhibited and milk fat concentration was maintained due to increased uptake of long-chain fatty acids from the circulation.However, the effect of AEA on milk fatty acid composition needs to be evaluated in future studies.Downregulation of ACC1 could also be due to higher leptin concentrations in AEA cows on d 28 p.p. Leptin inhibits ACC1 by activating AMP-activated protein kinase 67,68 , however, whether this pathway is also activated in the mammary gland of AEA cows requires further investigations.Furthermore, the decreased PLTP activity may be a result of greater leptin concentrations in AEA cows on d 28 p.p.The PLTP mediates the transfer of phospholipids to high-density lipoprotein cholesterol 69 , however, high-density lipoprotein cholesterol plasma concentration was unchanged in our study.Nonetheless, our result corresponds to the finding in heterozygous PLTP +/-mice, which had reduced PLTP activity abut no change in high-density lipoprotein cholesterol level compared to the wildtype 70 .
Effect of AEA on whole-body energy metabolisms in week 4 postpartum.Relative to the CON group, AEA cows showed a higher postprandial FOX and lower postprandial COX.In addition, AEA treated cows tended to or had higher cumulative feed intake beginning 3 h after feeding.Because feed intake is negatively and FOX positively correlated with COX 71 , the observed differences in whole-body energy metabolism could be related to the different cumulative feed intake between groups.Our results are in contrast to a previous study in late-lactating cows, in which AEA administration increased short-term feed intake, metabolic heat production, and COX, while decreasing FOX 15 .These contrasting findings suggests that the effect of AEA administration clearly depends on the energy balance differing between stages of lactation.The AEA treatment in late-lactating cows supports anabolism 15 , whereas, as shown herein, it induces whole-body catabolism and increases lipolysis in early-lactating cows.However, the observed differences in cumulative feed intake and metabolism between AEA and CON early-lactating cows may not be attributed to the AEA administration itself, but be overridden by the higher plasma leptin concentrations.In contrast to AEA, leptin exerts catabolic effects, reduces feed intake 11 , and increases fat oxidation 68 .

Comparative analysis of the AEA effect in different stages of lactation. Myers et al. proposed
that the sensitivity to endocannabinoids varies in the adipose tissue due to the physiological status 27 .However, a lactation stage-dependent sensitivity to ECS activation could also be present in other tissues.In mid-and latelactating cows, AEA administration increased short-term feed intake (1-10 h) but had no effect on total daily feed intake 15,28 , probably due to the short half-life of AEA.If the short AEA half-life affected the results of the present study remains questionable.
In early-lactation, AEA administration also showed no positive effect on total daily feed intake, but, this may be due to the short half-life of AEA or the dominating role of leptin.Nevertheless, endocannabinoid concentrations were found to directly correlate with an increase 28 or decrease 72 in feed intake during early-lactation, suggesting their involvement in the regulation of feed intake when not disturbed by leptin.When cows in late lactation are treated with AEA, they respond with a reduction in plasma NEFA concentration and thus lower lipomobilisation 15 .In contrast, AEA administration did not reduce lipomobilisation in early-lactation.Consistent with in vitro studies, CNR1 activation did not alter lipolysis rate in adipose tissue explanted from periparturient cows, whereas the lipolysis rate was reduced in adipose tissue collected from non-lactating and non-gestating cows 27 .
In the liver, we observed downregulation of the mRNA of genes related to TG synthesis, which may lead to less lipid accumulation in this organ.In contrast, in late-lactating dairy cows, AEA administration did not affect the mRNA abundance of genes involved in fat metabolism 15 .These results suggest a tissue-specific sensitivity to ECS activation depending on the physiological status of the cows, as proposed by Myers et al. 27 .However, further research is needed to elucidate the underlying mechanisms.In conclusion, the present study shows that repeated AEA administration in the first three weeks p.p. did not affect feed intake, energy balance, milk yield or milk composition.Furthermore, repeated AEA administration did not alter lipomobilisation.However, a three-week AEA treatment affected TG synthesis in the liver, underscoring a tissue-specific AEA insensitivity in early-lactation.Chronic elevation of the AEA level after 4 weeks of administration may resulted in a counter regulatory leptin increase, which coincided with a reduction in feed intake and consequently a higher hepatic lipid accumulation, increased whole-body fat oxidation and lower whole-body carbohydrate oxidation.Further investigations are needed to understand the interaction between leptin and AEA in early-lactation.

Material and methods
Animals and housing.The experimental protocol was approved by the Federal Office of Agriculture, Food Security and Fishery Mecklenburg-Western Pomerania, Rostock, Germany (LALLF, permission no.7221.3-1-015/19) and conducted in accordance with the ARRIVE guidelines (https:// arriv eguid elines.org/), the European Directive 2010/63/EU, the German Animal Welfare Act and the German Regulation on the Protection of Animals in Connection with Slaughter or Killing and on the Implementation of Council Regulation (EC) No 1099/2009.All persons involved in this study were blinded, except the persons who did the i.p. injections, the ultrasound measurements, the biopsies and the blood sampling.
For this study, 20 German Holstein cows at the end of their 1st (n = 12) or 2nd (n = 8) lactation were purchased from a local farm (Agrarprodukte Dedelow GmbH, Dedelow, Germany; the farm has consented to the use as experimental animals) in 10 blocks of 2 cows.Cows of each pair were half sibs, were in the same lactation number, but differed in age (± 5.5 months) and estimated calving day (± 6 d).Half sib pairs were chosen to reduce the genetic variance known to influence fat metabolism, milk yield and feed intake [73][74][75][76] .Pairs were transferred on d 56 (± 18 d) before expected calving to the free stall barn of the Experimental Facility for Cattle (Research Institute for Farm Animal Biology, Dummerstorf, Germany).Animals were habituated the respirations chambers on 3 different days between day -53 and day -23 before expected calving date.The duration of stay in the chambers was successively increased from initially 2 h to 8 h per day.Animals were considered habituated to the respirations chambers when they consumed feed and water, laid down, and ruminated.

Feeding and milking.
Cows were dried-off on d 51 (± 24 d) before the estimated calving date (except one cow on d 137 before expected calving).During the dry period, cows received a far-off diet until d 25 (± 8 d) before expected calving, following a close-up diet from d 24 (± 8 d) until parturition.After parturition, cows received a lactation diet (Table 3).All diets were offered as total mixed ration.Cows had ad libitum access to water and feed, except between 0500 and 0745 h for reasons of maintenance and to synchronize the start of feed intake after the morning feeding.Feed samples were taken weekly and the dry matter content was determined by drying samples for 24 h at 60 °C and subsequently for 4 h at 103 °C.Nutrient composition was analysed by the Landwirtschaftliche Untersuchungs-und Forschungsanstalt (LUFA GmbH, Rostock, Germany) using near infrared spectroscopy according to VDLUFA (2004) (Table 3).The individual daily feed intake as measured by the Roughage Intake Control system (RIC, Insentec B. V., Marknesse, The Netherlands) was used to calculate a weekly mean.
After calving, cows were milked twice daily at 0500 h and 1630 h and the milk yield was recorded.Milk composition was analysed by infrared spectroscopy (MilkoScan; Foss GmbH, Hillerød, Denmark) at the State Inspection Association for Performance and Quality Testing Mecklenburg-Western Pomerania e.V. (LKV Güstrow, Germany).For this purpose, milk samples from the evening and morning milking were pooled once a week.The ECM yield was calculated according to the GfE (2001) 77 : The energy balance (EB) was calculated according to GfE (2001) 77 as follows: For antepartum period: For postpartum period: NE L intake, NE L maintenance, NE L pregnancy, and NE L milk were calculated as follows: Body weight, BCS and ultrasound measurements.Body weight (BW) was recorded once a week in the a.p. period and twice daily in the p.p. period after milking using a walk-through scale.From the weekly mean, the mBW was calculated.Once a week, the BCS was determined according to Edmonson et al. 78 and the heart girth was measured according to Branton and Salisbury 79 .The back fat thickness was measured according to Staufenbiel 80 and the subcutaneous fat layer over the 12th rib according to Raschka et al. 81 , each once a ECM kg/d = milk yield kg/d × 1.05 + 0.38 × milk fat % + 0.21 × milk protein % /3.28 .week via ultrasound.Additional ultrasound measurements were taken to calculate the amount of subcutaneous adipose tissue, retroperitoneal adipose tissue , omental adipose tissue and mesenteric adipose tissue according to Raschka et al. 81 .The total abdominal adipose tissue was calculated as the sum of retroperitoneal, omental and mesenteric adipose tissue.The devices and ultrasound probes used for the respective ultrasound measurements are shown in Supplementary Table 1.
Treatments.Each cow of the half-sib pair was randomly assigned to either a treatment group (AEA, n = 10) or control group (CON, n = 10).Treatment started on d 1 after parturition if the cow calved before 1400 h or on d 2 if the cow calved after 1400 h.The AEA group received 3 µg/kg BW/d of N-arachidonoylethanolamine (AEA; Tocris, Bristol, UK) diluted in 50 mL 0.9% NaCl via i.p. injections.The i.p. injections were administered at the right paralumbar fossa as described previously by van Ackern et al. 14 .We chose i.p. injection because it allows rapid AEA absorption into the systemic circulation and the activation of the splanchnic ECS and vagal afferents of the gut-brain-axis 15 .
Subcutaneous fat biopsy and liver biopsy.On d 25 (± 10 d) before expected calving and on d 16 (± 2 d) p.p. a liver biopsy and a subcutaneous adipose tissue biopsy were taken.The liver was scanned using ultrasound [L52x Rectal Transducer (10-5 MHz), Fujifilm SonoSite Inc., Bothell, WA; SonoSite MicroMaxx; Fujifilm SonoSite Inc., Bothell, WA; USA] to locate the insertion site and avoid injuries to large hepatic blood vessels or the intestine.A 15 × 15 cm area located approximately 10 cm below the processus transversi was shaved, washed, disinfected, and 10 mL of 2% procainhydrochlorid (Procamidor, WDT, Garbsen, Germany) were injected subcutaneously and into the muscles of the 12th, 11th, or 10th intercostal space.After stab incision in the respective intercostal space with a scalpel, liver tissue was obtained using the Pro-Mag Ultra 2.2 Liver device (Plano, USA) with a 13 gauge needle (1st and 2nd block) or a tailor-made two-part trocar needle with an outer diameter of 6 mm (3rd to 10th block).Liver tissue samples were immediately snap frozen in liquid nitrogen and stored at − 80 °C until analysis.
For the adipose tissue biopsy, cows received epidural anaesthesia with 5 mL of 2% procainhydrochlorid (Procamidor, WDT, Garbsen, Germany).Furthermore, 10 mL procainhydrochlorid were subcutaneously injected above the sacrotuberal ligament.After shaving, cleaning and disinfection, the skin of the ischiorectal fossa was cut and adipose tissue was taken through a 2.5 cm long incision.It was not possible to obtain subcutaneous adipose tissue from one CON cow p.p.
Immediately after collection, half of the harvested adipose tissue was frozen in liquid nitrogen and stored at − 80 °C for further analysis.The other half of the obtained adipose tissue biopsy was used to determine in vitro lipolysis according to Kokkonen et al. 83 .Briefly, biopsy tissue was immersed in 37 °C Krebs-Ringer solution (Thermo Fisher Scientific Inc., Waltham, USA) supplemented with 15 mmol/L NaHCO 3 and 2.5 mmol/L CaCl 2 , and transported within 10 min to the laboratory.The tissue was cut into 5 mm pieces.Approximately 250 mg tissue were incubated in 3 mL of the medium mentioned above but saturated with O 2 at 37 °C.After 15 min of incubation, 1 mL of the incubation medium was removed and frozen at − 20 °C.Ten µl of a solution containing 10 mM noradrenaline-hydrochloride and 2 M glucose were added to the remaining medium before the medium was gassed with O 2 and incubated for 120 min at 37 °C.Thereafter, 1 mL medium was taken and frozen at − 20 °C and the fat tissue subjected to freeze-drying to determined tissue dry matter.The 1-mL media samples were thawed and the dissolved protein precipitated by adding 0.9 mL acetonitrile to 0.3 mL medium.After mixing and centrifugation at 13,000 g and 4 °C for 20 min, the supernatant was evaporated to dryness and re-dissolved in ultra-pure water by sonication with a concentration factor of 2 and centrifuged again for 10 min.The glycerol concentrations were measured using HPLC with a refractive index detector (1200/1260 infinity Series, Agilent Technologies).Chromatographic separation of 50 µL solution was carried out on a 300 × 7.8 mm Rezex ROA-Organic Acid H + (8%) column (Phenomenex, Aschaffenburg, Germany) protected with a 4 × 3 mm Carbo-H + guard cartridge (Phenomenex, Aschaffenburg, Germany) at 75 °C using sulphuric acid (5 mM) as eluent with a flow rate of 0.4 mL/min.Calculations were done by the use of external standards from 0.05 to 0. Cows received the i.p. administrations as described above and were fed 45 min post administration.The 24-h gas concentration measurement started after over-night gas equilibration on d 28 (± 2 d) at 0700 h.CO 2 , O 2 , CH 4 concentrations and feed intake were measured in 6-min intervals as described by Derno et al. 84 .The airflow through the chamber was set to approximately 30 m 3 /h and was measured by a differential pressure type V cone flow meter (McCrometer, Hemet, CA).The mean CO 2 recovery rate of each chamber was 99.9%.
Total CO 2 production is composed of CO 2 from fermentative (CO 2 ferm) and metabolic (CO 2 metab) processes.CO 2 ferm was estimated according to Chwalibog et al. as CO 2 ferm (L) = 1.7 × CH 4 (L) 85 .Metabolic CO 2 was calculated by subtracting CO 2 ferm from total CO 2 .
Due to technical issues, only the data obtained before feeding and up to 7 h after feeding were considered in the following calculations: The heat production (HP) was calculated according to Brouwer 86 : Net carbohydrate oxidation (COX) and net fat oxidation (FOX) were calculated according to Frayn 87 : Urinary N excretion (N u ) was estimated to 100 g/d, considering real N u values ranging from 75 to 150 g/d 88 .Thereby, an error of less than 5% of HP, COX and FOX was accepted.
Data for HP, FOX and COX were normalized to mBW.Changes in FOX, COX and HP, as well as cumulative DMI were calculated in relation to the start of feeding (0745 h) and evaluated in hourly intervals.

Slaughter and quantitative real-time-PCR (RT-qPCR).
On d 30 (± 1 d) p.p., cows were sacrificed 1-1.5 h after AEA or CON administration by captive bolt stunning and subsequent exsanguination at the institutional abattoir.Samples were taken from the lobus quadratus of the liver and from the right rear quarter of the mammary gland, snap frozen in liquid nitrogen, and stored at − 80 °C for further analysis.These tissue samples as well as the liver samples obtained from the biopsies were ground in a mortar under liquid nitrogen.Approximately 18 mg powdered tissue were extracted, respectively, for total RNA using the innuPREP RNA Mini Kit 2.0 (AJ Innuscreen GmbH, Berlin, Germany).The RNA samples were DNA digested (innuPREP DNase I Digest Kit, AJ Innuscreen GmbH) and the RNA quality assessed with an Agilent 2100 Bioanalyzer (Agilent Technologies Inc., Santa Clara, CA, USA) yielding RNA integrity numbers (RIN) between 6.4 and 9.1 (mean: 8.3, SD: 0.6).For first-strand cDNA synthesis, 1 µg RNA was reverse transcribed (SensiFast™ cDNA Synthesis Kit; Bioline, London, UK) using a Thermocycler (peqstar 96 × HPL, VWR International, Pennsylvania, USA).

Liver fat.
Frozen liver tissue samples taken after slaughter were cut into 6-μm sections with a cryostat microtome (CM3050 S, Leica, Bensheim, Germany) and stained with Oil Red O (Chroma Gesellschaft, Münster, Germany).Nine randomly selected images (total area 3.3 mm 2 ) per cow were taken, except for one animal for which 8 images were taken (total area 2.9 mm 2 ), using an Olympus BX43 microscope (Olympus, Hamburg, Germany) equipped with a UC30 colour camera (OSIS, Münster, Germany) and cellSens image analysis software (Evident, Hamburg, Germany).The lipid area was identified using a colour threshold operation and the number of lipid droplets and the lipid droplet area were determined using the "Count and Measure" function of the cellSens software.In addition, the number of lipid droplets/mm 2 and the percentage of lipid droplet area of the total measured area were calculated.

Statistical analyses.
The sample size required was calculated with CADEMO for Windows (ANOV version 4.03, BioMath GmbH, Rostock) using ANOVA-F-Tests, assuming a biological relevance of at least 1.4 residual standard deviations, and a probability for the type I error of α = 0.5 and type II error of β = 0.2.
All statistical evaluations were performed using R Statistical Software (v4.2.0; R Core Team 2021, R Foundation for Statistical Computing, Vienna, Austria).Outliers were detected by using cooks distance (olsrr package, v0.5.3;Hebbali 2020) and visual inspection of boxplots.Two outliers in the dataset of AEA plasma concentration on d + 14 were detected and excluded from the subsequent statistical analysis.Furthermore, one outlier was HP (kJ) = 16.18 * O 2 (L) + 5.02 * CO 2 total (L) − 2.17  detected and excluded in the dataset of mRNA analysis of hepatic genes (CNR1, GPR55, DGAT1/2 in liver tissue obtained on d -25 and on d 30).Due to technical problems, two animals were excluded from the indirect calorimetry analysis.In addition, one animal was excluded from the fat depot analyses, because it had haematomas at the measurement sites.Another animal was excluded from all analyses in week 4 and 5 p.p. due to intestinal obstruction.
Data were analysed with a linear mixed model (LMM, lmer function, lme4 package, v1.1-29; Bates, Maechler, Walker 2015 89 ).For the analysis of gene expressions, the plasma concentration of endocannabinoids, N-acylethanolamines, plasma metabolites, leptin., PLTP activity a.p. and the analysis of the number of lipid droplets and lipid droplet area in liver tissue the model contained "group" (level: AEA and CON) as fixed effect and "sire" as random effect.The same model was used for dataset of lipolysis rate a.p. and p.p., respectively.To evaluate the plasma concentration of endocannabinoids, N-acylethanolamines, plasma metabolites, leptin, PLTP activity p.p. and data obtained from indirect calorimetry, the model contained the fixed effects "group" (level: AEA and CON), "time" and the interaction (group x time), with "time" as repeated variable.Additionally, the model contained "sire" and "cow ID" as random effects.The equal model was used to evaluate the data of the BW, DMI/ mBW, energy balance, milk yield, milk constituents and data obtained from ultrasound measurements, whereby data were considered separately for a.p. and p.p. period.The fixed effect "lactation number" was tested but did not improve the models.All models were tested for homoscedasticity and normal distribution of the residuals (check_normality and check_heteroscedasticity function, performance package, v0.9.0; Lüdecke et al. 2021 90 ).Heteroscedasticity has only a marginal impact on model estimates 91 , allowing the models to be used despite the violation of the constant error variance.If the assumption of normality was violated, data were transformed with the Johnson transformation.In case the transformation was not successful, a generalized linear mixed model with a gamma distribution was used.The Wilcoxon signed rank test was used to analyse EPEA during a.p. period, because no normal distribution of the residuals could be achieved after transformation and gamma distribution did not fit.
Pairwise differences between levels of fixed effects were tested by using the Tukey Kramer test.For the fixed effect of interest, estimated marginal means and their standard errors (SEs) were estimated.Effects and differences were considered significant at P < 0.05 and as a trend at P < 0.1.For interpretation purposes, the means of the observed data and their standard deviations (SDs) are presented in the figures and tables.

Figure 5 .
Figure 5. Percentage of the lipid droplet area of the total area in the liver tissue (a), and number of lipid droplets per mm 2 in liver tissue (b) of cows treated intraperitoneally with N-arachidonoylethanolamide (AEA, n = 9) or NaCl (CON, n = 10) postpartum.Data are presented as means ± SD; # P < 0.1, * P < 0.05.Oil red O staining of liver sections from the AEA (c), and the CON group (d).The black framed areas from the top images are shown enlarged at the bottom.Scale bar represents 50 µm.
Vol.:(0123456789) Scientific Reports | (2023) 13:14665 | https://doi.org/10.1038/s41598-023-41938-0www.nature.com/scientificreports/ results were corrected for recovery.The lipolysis rate was calculated by subtracting the glycerol concentration at the time t = 15 min from the concentration at t = 120 min.On d 27 (± 2 d) p.p., each cow was transferred into one of four open-circuit respiration chambers and individually kept in tie-stalls at 15 °C until d 29 (± 2 d).The BW was recorded before and after entering the respiration chamber.Milking was performed at 0630 h and 1700 h, and the milk yield was recorded.