Impact of milking interval and time on milk spontaneous lipolysis and composition in dairy cows

Milk lipolysis is defined as the hydrolysis of triglycerides, the major component of milk fat, resulting in the release of short-chain fatty acids ( FA ) responsible for rancid flavor and partial glycerides that impair functional properties such as foaming and creaming abilities. Milk lipolysis is a complex phenomenon that depends on both animal parameters and breeding factors. Milk spontaneous lipolysis is known to be higher in milk from evening milkings than from morning milkings. This may be related to the longer length of overnight milking intervals or to the nycthemeral cycle. In this experiment, our objective was thus to study the impact of both milking intervals and time of day on milk spontaneous lipolysis in twice-daily-milking systems with one of 3 milking intervals: Short Day – Long Night ( SD-LN, 6.30 a.m. and 4.30 p.m.,); Long Day – Short Night ( LD-SN , 6:30 a.m. and 8:30 p.m.,); and Balanced Day and Night ( BDN , 6:30 a.m. and 6:30 p.m.,). To achieve this goal, 21 multiparous dairy cows in mid-lactation were used in a 3 × 3 Latin square design over 3 periods. The experiment lasted 5 weeks, corresponding to 3 experimental periods of 6 d alternating with 8 d of milking with conventional hours (morning-evening gap of 10 h). We confirmed that milk spontaneous lipolysis was influenced by milking interval, but not the milking time. Indeed, we observed more lipolysis in SD-LN evening milk (+0.20 mEq/100 g fat) and LD-SN morning milk (+0.22 mEq/100 g fat), both of which corresponded to a 10 h interval between successive milkings. High lipolysis milk came from cows that produced less milk with a higher milk fat content. No significant difference between milkings was observed for BDN. Milk protein, total P and citrate contents increased according to the duration of mammary gland storage of


INTRODUCTION
Milk lipolysis is defined as the hydrolysis of triglycerides, the primary component of milk fat, by the lipoprotein lipase ( LPL) enzyme (Deeth, 2006).This leads to the release of mono and di-glycerides and free fatty acids (FFA), including the short-chain fatty acids (FA) responsible for rancid flavor.In addition, the presence of partial glycerides impairs the functional properties of milk, such as foaming and creaming abilities (Deeth, 2006;Kamath et al., 2008).There are 3 types of lipolysis: spontaneous lipolysis, induced lipolysis and microbial lipolysis.In cows, spontaneous lipolysis is initiated by the cooling of raw milk to a temperature below 10°C after milking, in the absence of mechanical shock (Deeth, 2006).Three biochemical factors appear to largely determine the susceptibility of milk to spontaneous lipolysis: the amount of lipase activity; the integrity of the milk fat globule (MFG) membrane; and the balance of lipolysis-activating and -inhibiting factors (Cartier and Chilliard, 1990;Sundheim, 1988).A weak correlation has been found between lipase activity and the level of spontaneous lipolysis in bovine milks (Cartier and Chilliard, 1990).Similarly, the integrity or fragility of the MFG membrane would appear to be an obvious factor in predisposing milk to spontaneous lipolysis.Spontaneous lipolysis depends on both the animal as well as factors related to breeding conditions.Induced lipolysis occurs following mechanical or thermal shocks, which take place during milk transport and storage.In practice, it is impossible to dissociate spontaneous and induced lipolysis in milk tanks on dairy farms.On the other hand, microbial lipolysis arises due to the effect of microbial enzymes on milk fat globules, which becomes significant after 3 to 4 d of storage of raw milk of satisfactory health quality.In the present paper, we will focus on spontaneous lipolysis since milk samples have been analyzed just after milking.
It has been shown that spontaneous lipolysis is more pronounced in evening milk than in morning milk (Murphy et al., 1979;Ahrné and Björck, 1985;Vanbergue et al., 2017;Hurtaud et al., 2023).This contrast between morning and evening milk lipolysis could be due to unbalanced intervals between milking times (Wiking et al., 2019), which lead to differences in the quantity of milk produced in the morning and evening.Indeed, Bachman et al. (1988) observed no difference between morning and evening spontaneous lipolysis levels when milking intervals were equal.However, evening milk could also be directly affected by the circadian rhythm of the secretions of some specific hormones (Jellema, 1980).
The objective of this study was therefore to evaluate the impact of unbalanced versus balanced milking intervals on spontaneous milk lipolysis in a twice-daily-milking system with one of 3 milking intervals: a short-day/longnight milking interval (SD-LN); a long-day/short-night milking interval (LD-SN); and a balanced day/night milking interval (BDN).This experiment also enabled us to study the effect of milking time of day (morning or evening) within each milking system.Our hypothesis was that milk lipolysis would be affected by milking interval, with an increase for shorter intervals, as well as by animal activity during daylight hours.

Animals and experimental design
The experiment was conducted at IE PL, INRAE, Dairy nutrition and physiology (IE PL, 35650 Le Rheu, France; https: / / doi .org/ 10 .15454/yk9q -pf68; animal housing agreement number C-35-275-23) in accordance with French legislation on animal experimentation and with approval by the French National Committee for Consideration of Ethics in Animal Experimentation (Authorization: APAFiS #23235-2019120920557765 v2 delivered on 22 January 2020).For the experiment, we used 21 Holstein cows (6 in first and 15 in second lactation; 48 ± 8 DIM; 594 ± 35.4 kg of BW; average ± SD) producing 32.7 ± 5.0 kg of milk/d with 4.06 ± 0.37% fat, 3.01 ± 0.18% protein, with a spontaneous lipolysis estimated to 0.51 ± 0.24 mEq/100 g of fat for evening milk.All cows were kept indoors, with a mean area of 8.75 m 2 per cow.Cows were allocated to 3 groups of 7 animals each, according to the following criteria and in this order: milk yield, spontaneous lipolysis, lactation stage, parity (primiparous vs. multiparous), milk fat and protein content, SCC, and BW.
Starting in October 2020, the experiment was conducted using a Latin square design with milking interval as the main factor for 3 periods of 6 d alternating with 8 d of milking with conventional hours (morning-evening gap of 10 h) (Figure 1).Three treatments were applied: a short-day/long-night milking interval (milking at 6:30 a.m. and 4:30 p.m.; SD-LN); a balanced day/night milking interval (milking at 6:30 a.m. and 6:30 p.m.; BDN) and a long-day/short-night milking interval (milking at 6:30 a.m. and 8:30 p.m.; LD-SN).
All feed refusals were collected and weighed daily to determine individual DMI.To calculate DMI, refusals were assumed to have the same composition as the offered diet.

Sample collection and laboratory analyses.
Milk yield and traits Cows were milked in the milking parlor twice daily at different times according to their milking interval treatment.Milk yield was recorded individually at each milking.Milk fat, protein and lactose content, and SCC were determined from 6 consecutive milkings (Monday, Tuesday and Wednesday) corresponding to the last days of each experimental period and were also determined from 6 consecutive milkings during the intervals between experimental periods.These analyses were performed by FTIR mid-infrared spectrometry for fat, protein and lactose content and by flow cytometry for SCC at the MyLab dairy laboratory (Châteaugiron, France).Milk secretion rate was calculated by dividing milk produced by milking interval duration.
An additional individual milk sample was collected from milk cans from individual morning and evening milkings at the end of each experimental period (last 2 milkings), and then stored in containers of different volumes at 4°C or at −20°C according to the analyses.Spon-taneous lipolysis, FA profile, milk fat globule (MFG) diameter, and milk protein and mineral composition were determined from these milk samples.
Spontaneous lipolysis of milk Milk was immediately heated after milking in a water bath at 100°C during 3 min and after 24 h of storage at 4°C to stop the activity of LPL, and FFA were measured by the copper soap method as described in Hurtaud et al. (2023).Spontaneous lipolysis of milk was quantified using the FFA content of milk after 24 h of storage at 4°C, from which the initial FFA content was subtracted.

Milk fat globule and casein sizes
After adding bronopol (Merck, Darmstadt, Germany), vials of milk were kept at room temperature for a maximum of 16 h to assess MFG diameter distribution using laser-light scattering (Mastersizer 3000, Malvern, UK).Using the Malvern software, we calculated mean ), and MFG area S = 6/(ρ × d 3,2 ), with N i = the number of MFG in diameter class d i , and ρ = the density of the particle considered (0.92 for fat).Another sample of milk without bronopol was skimmed by 2 successive centrifugations (3157 × g, 4°C).The mean diameter d 4,3 of casein micelles was measured using the Mastersizer 3000.
Milk fatty acid composition and LPL activity.Milk FA methyl esters of the freeze-dried milk samples were then prepared and analyzed after injection into a gas chromatograph (Agilent 7890A GC System, Massy, France) equipped with a flame ionization detector and a CP-Sil 88 capillary column (100 m × 0.25 mm, 0.2 μm thickness; Agilent Technologies, Inc., Santa Clara, California, USA), as previously described (Fougere et al., 2018).Milk LPL (EC 3.1.1.34)activity was measured from morning milk stored at −20°C, as described by Bernard et al. (2005).
Milk N and mineral composition.Total nitrogen, NPN, non-CN, and CN were determined according to the Kjeldahl method described by Alais (1984).Urea was analyzed on milk ultrafiltrate (Vivaspin® Turbo 15 Centrifugal Concentrator Polyethersulfone, Sartorius, Göttingen, Germany) in 2 replicates with colorimetric enzymatic reactions assessed using a multi-parameter analyzer (KONE Instruments 200 Corporation, Espoo, Finland).Total and soluble Ca, total Na, total K, and total Mg were respectively analyzed via inductively coupled plasma optical emission spectroscopy (ICP-OES 5110 Agilent Technology, Les Ulis, France) of milk and milk ultrafiltrate (Ca only), as previously described by Hurtaud et al. (2023).Total and soluble P and Cl contents were determined using a KONE PRO multiparameter analyzer (ThermoFisher Scientific, Illkirch, France) according to the Allen method for P (Pien, 1969) and as described by Henry et al. (1974) for Cl.
Plasma metabolites and hormones.Blood was sampled from the tail using 5 mL heparinized and EDTA tubes (VT-050SHL, Venoject, Terumo Europe, Leuven, Belgium) after the morning milking during the last day of each experimental period.Blood was centrifuged at 2,264 × g for 15 min, and plasma was removed and stored at −20°C until analysis.Plasma glucose, urea, acetate, nonesterified FA (NEFA), triglyceride, lactose and BHB contents were analyzed in 2 replicates by colorimetric enzymatic reactions, as reported by Delamaire and Guinard-Flament (2006).Plasma insulin concentration was determined by RIA using the Wizard2 gamma counter 2470 (Perkin-Elmer) with commercial kits (Insulin RIA kit, PI-12K, Millipore, Billerica, Massachusetts, USA).

Statistical analyses
Unless otherwise noted, all statistical analyses were performed using the SAS software (SAS 9.2 Institute Inc., Cary, North Carolina, USA).Milking interval treatment (SD-LN, BDN and LD-SN), cow, and period effects were evaluated using univariate analyses of daily values for DMI, BW, milk traits and plasma parameters.For milk traits, daily values were obtained by calculating the average of morning and evening values weighted by milk yield.The effect of milking interval was also evaluated on milk from morning and evening milkings.We fit for each value a linear mixed model (SAS MIXED procedure) including the cow as a random effect and the experimental period (1, 2 or 3), milking time (morning or evening), interval of milking (SD-LN, LD-SN, or BDN) and interaction between milking time and interval as fixed effects: Multivariate statistical analyses were also used to identify and visualize trends across a subset of daily values measured in morning and evening milks [milk production and composition, milking duration and milk secretion rate, FFA, lipolysis, MFG, and casein diameter, Ca, and P, and some milk FA (C4:0, C6:0, C8:0, C10:0, C14:0, C16:0, C18:0, trans-10-C18:1, trans-11-C18:1, cis-9-C18:1, cis-9,cis-12-C18:2 (n-6), cis-9,cis-12, cis-15-C18:3 (n-3), cis-9,trans-11-CLA, cis -9 -C14: 1/ C14: 0, even FA < C16:0, FA > C16:0, odd FA < C17:0)], and in plasma.Multivariate analyses were performed using R (version 4.3.0).Missing values were first imputed with a multiple factor analysis using the missMDA package (version 1.18) (Josse and Husson, 2016).We then performed a coinertia analysis (Dray et al., 2003) using the ade4 package (version 1.7-22) (Thioulouze et al., 2018) to quantify the co-structure between morning and evening milk traits matched by cow within interval treatment group.An RV coefficient, corresponding to the multivariate generalization of the squared Pearson correlation coefficient, was used to quantify the co-variability in milk traits between morning and evening milkings.A permutation test (n = 9999 random permutations) was used to evaluate the significance of each RV coefficient.Finally, a partial redundancy analysis (RDA) (Legendre and Legendre, 2012) using the vegan R package (version 2.6-4) was performed to summarize the variation in the full set of daily values for milk traits explained by milking time, milking interval, and their interaction after adjusting for each individual cow.Analysis of variancelike permutation tests (n = 999 permutations) were used to assess significance for each term.A partial RDA was similarly performed to summarize the variation in plasma parameters explained by milking interval after adjusting for each individual cow.For all analyses, the threshold for statistical significance was set at P < 0.05, while that for a trend was set at 0.05 < P < 0.10.

Milking interval
The major result of this experiment is that the SD-LN and LD-SN milking intervals have an inversed impact on most of the measured parameters, with somewhat stronger differences observed in evening than morning milk (Figure 2).
The milking interval duration had no significant effect on DMI and BW.It also had no significant impact on milk yield, milk fat and protein content and yield, lactose yield and SCC.Only milk lactose content increased with BDN (+0.8 g/kg, P < 0.001) compared with SD-LN and LD-SN (Table 1).When comparing morning versus evening milkings, the unbalanced SD-LN and LD-SN had opposite results.Milk yield and milk protein content were significantly higher in long milking intervals, that is SD-LN morning compared with evening (+3.7 kg and +1.6 g/kg, respectively), and LD-SN evening compared with morning (+3.9kg, P < 0.001; +1.4 g/kg, P < 0.001, respectively).The milk secretion rate and milk fat content were both significantly lower in long milking intervals, that is SD-LN morning versus evening (−0.17 kg/h and −4.5 g/kg, respectively) and LD-SN evening compared with morning (−0.12kg/h, P < 0.001; −3.8 g/ kg, P < 0.001, respectively).Milk fat, protein and lactose yields were higher in the morning with SD-LN (+81 g, +142 g, and +187 g, respectively) and higher in the evening with LD-SN (+99 g, +146 g, and +192 g, P < 0.001 respectively).Milk SCC tended to be higher in the morn-ing with LD-SN (+6.5) and in the evening with SD-LN (+7.3, P = 0.053) (Table 1).
With BDN, milk production during morning or evening milking was intermediate to that of SD-LN and LD-SN, but significantly different from both (−1.8 kg relative to the highest value with SD-LN or LD-SN).Milk fat content with BDN was not significantly different from SD-LN in the morning and from LD-SN in the evening.Milk protein content with BDN was significantly lower than SD-LN in the morning (−0.6 g/kg, P < 0.001) and not significantly different from LD-SN in the evening.With BDN, milk lactose content was not different from SD-LN in the morning and was significantly higher than LD-SN in the evening (+1.4 g/kg, P < 0.001).Milk fat, protein and lactose yields for BDN were between those of SD-LN and LD-SN in the morning and evening.With BDN, SCC tended to be lower than with short milking intervals, i.e., LD-SN morning milk and SD-LN evening milk (Table 1).Milking intervals had no effect on NPN, casein content and casein on protein ratio (supplementary Table 1, https: / / doi .org/ 10 .57745/HUBGDM) Lipolysis was lower for long milking intervals, whether SD-LN morning milk (−0.19 mEq/100 g fat) or LD-SN evening milk (−0.27 mEq/100 g fat, P < 0.001).Overall, the activity of LPL was higher with SD-LN compared with LD-SN (+62.9 and +165.4 nmol/min/mL for morning and evening values, respectively, P < 0.001).However, LPL activity was lower in morning milk for both SD-LN and BDN (−97.6 and 15.3 nmol/min/mL, respectively), and there was no significant difference between morning and evening milks for LD-SN.There was no effect of milking intervals on MFG diameter (d 4,3 , d 3,2 ) and area (Table 2).
With BDN, milk lipolysis was not significantly different from SD-LN morning milk or from LD-SN evening milk.For this treatment, LPL activity was not different from SD-LN in morning milk and intermediate to SD-LN and LD-SN in evening milk (Table 2).

Milking time
A coinertia analysis of daily milk traits confirmed the strong shared global co-structure between morning and evening milk regardless of milking interval (RV = 0.826, P < 0.001 for SD-LN; RV = 0.826, P < 0.0001 for BDN; RV = 0.872, P < 0.0001 for LD-SN).Interestingly, morning milk from SD-LN more strongly resembled evening milk from LD-SN (RV = 0.673, P < 0.0001) than did morning LD-SN milk and evening SD-LN milk (0.576, P < 0.001).
With BDN, the milk secretion rate was higher for morning milk than for evening milk (respectively +0.11, P = 0.003).Milk protein and lactose contents were higher in the evening with BDN (respectively +1.0 g/kg, P < 0.001 for protein and +2.4 g/kg for lactose, P < 0.001), whereas milk fat yield was lower (−43 g, P = 0.015) (Table 1).
Lipolysis was higher in the morning only with SD-LN (+0.30 mEq/100 g fat, P < 0.001).There was no effect of milking time on MFG.However, we did identify an effect of milking time on LPL activity that increased in evening milks (+18 ηmol/min per mL, P < 0.001).

DISCUSSION
We showed here that the composition of milk is dependent upon the milking interval, especially with regard to lipolysis values.Overall, SD-LN morning milk strongly resembled LD-SN evening milk.Daily milk production was not affected by milking intervals, probably because the duration between successive milkings never surpassed 16 h.Indeed, according to Stelwagen et al. (2008), the rate of milk secretion remained relatively constant until 12 h and tended to be lower at 18 h, after which it began to fall.However, when considering morning and evening milking separately, as expected milk production was greater for longer milking intervals (in the morning for SD-LN, and in the evening for LD-SN).Conversely, the hourly milk production was higher for the shorter 10 h-intervals.Indeed, the rate of secretion by alveolar epithelial cells proceeds at a uniform rate until the alveoli expand with milk and exert pressure on the milk-producing cells.This fluid stasis inhibits and may eventually stop the synthetic processes.In addition, serotonin, which plays a role in the control of alveolar volume homeostasis in the mammary gland, initially causes a transient decrease in tight junction permeability, followed by a precipitous increase in permeability   (Marshall et al., 2014).When cows are milked at 10-and 14 h-intervals, higher milk yields were obtained after 14 h due to total fluid accumulation; however, the rate of secretion or yield per h is greater during the 10 h-interval (Nickerson, 1995).For BDN, milk morning yield was greater due to a higher rate of milk secretion during the night for a similar milking interval.This result could be associated with day-to-night differences in cow activity with more metabolites available for milk synthesis (Ferlay et al., 2010).
With regards to the main objective of the present study, we have shown that lipolysis is primarily affected by milking interval, and that the higher lipolysis typical of evening milks is essentially related to a shorter milking interval rather than the time of day.Globally, multivariate analyses of daily milk measurements highlighted a strong clustering according to milking time and milking interval length (Figure 2).Interestingly, a somewhat more marked distinction among milking interval lengths was observed for evening milk than for morning milk.These results suggest that when considering the daily milk measurements globally, the 2 unbalanced milking intervals systems have opposite but not perfectly symmetrical effects on morning and evening milk parameters.This observation is in line with the coinertia analysis results, which showed somewhat stronger co-variability between morning SD-LN and evening LD-SN milks than between morning LD-SN and evening SD-LN milks.
Taken together, these results confirmed that the shorter the interval, the higher the lipolysis (Wiking et al., 2006), regardless of the time of day.One hypothesis is that the reconstitution of the membrane surrounding the triglycerides is penalized by a shorter milking interval (Connolly, 1978).However, MFG diameter was not affected by milking interval.It is possible that only the composition of the membrane was modified, which could explain its increased fragility (Wiking et al., 2006).Another hypothesis could be that LPL cofactors may be involved in milking interval effect on lipolysis.Indeed, Ahrné and Björck (1985) found higher milk LPL activity in the milk  fraction of evening milk compared with morning milk with a 9 h interval.In the present study, there was no correlation between LPL activity and lipolysis as previously observed in Cartier and Chilliard (1990); here LPL activity only increased with SD-LN especially in evening milk.Measuring the amount of LPL present in the cream, as suggested by Cartier and Chilliard (1990), might have been a better indicator of lipolysis.There was no significant difference in lipolysis between morning and evening milk with BDN, similar to the absence of difference in lipolysis between morning and evening milk with equal intervals found by Bachman et al. (1988).Unsurprisingly, the milk FA profile was not altered by milking intervals as the rearing conditions, in particular the diet, were not varied (Kliem and Shingfield, 2016).However, regardless of milking interval, the percentage of ∑ < C16:0 percentage (de novo FA) was lower in the evening milk mainly due to the decrease in the percentage of ∑ C16.The ∑ > C18:0 percentage (preformed FA) was higher in the evening particularly due to an increase in the percentage of cis-9-C18:1, as well as PUFA (P < 0.001).These differences between morning and evening milks are consistent with those found by Ferlay et al. (2010) and Vanbergue et al. (2018), and could be associated with day-to-night differences in cow activity, and the subsequent impact on nutrient availability for milk fat synthesis.
The increase in the Na + /K + ratio with LD-SN evening milk alone seems to indicate an opening of tight junctions associated with a loss of integrity of the udder.Surprisingly this effect was not observed with SD-LN morning milk.This difference could be linked to the drop in activity with an increase in rest time during this period (Leliveld et al., 2022).
The increase in citrate concentration in long milking intervals, that is SD-LN morning and LD-SN evening milk, could be due to a longer synthesis time.Indeed, according to Linzell et al. (1976), citrate is synthesized in the mammary gland, citrate carbon being derived from both glucose and acetate.Its increase in milk cannot be linked to the opening of tight junctions because in normal lactation, the mammary epithelium is almost totally impermeable to citrate (Linzell et al., 1976).The effect of milking time on milk minerals is difficult to explain and may be linked to bone accretion phenomena.Milk Cl was lower in evening milk (P < 0.001).The Cl concentration decreased due to passage from milk into blood indicated by the increase of Na + /K + ratio indicating tight junction opening (Sorensen et al., 2001).

CONCLUSIONS
Taken together, our results highlight the strong global similarity in lipolysis, fat composition, and structure between milks obtained after similar milking intervals, whether short (10 h; LD-SN morning and SD-LN evening milks), long (14 h; LD-SN evening and SD-LN morning milks), or intermediate (12 h; BDN morning and evening milks).We have thus shown that differences between milks from morning and evening milking are associated primarily with different milking intervals rather than with the day-night transition.However, this original study leaves some unanswered questions about the mechanisms underlying lipolysis (activators, inhibitors, milk fat globule membrane composition etc.) that will need to be unraveled to better control milk lipolysis through adapted rearing conditions.It would be worthwhile in future work to test a strategy for modifying milking intervals over the course of a single day or over successive days to manage lipolysis.
Hurtaud et al.: Milking interval and milk lipolysis

Table 2 .
Milk free fatty acids, lipolysis and milk fat globule structure of cows submitted to different milking times and intervals

Table 3 .
Milk fatty acids of cows submitted to different milking times and intervals

Table 4 .
Hurtaud et al.:Milking interval and milk lipolysis Milk nitrogen composition of cows submitted to different milking times and intervals Means in the same row with no common superscript differ.

Table 5 .
Milk minerals of cows submitted to different milking times and intervals