Feed restriction of lactating cows triggers acute downregulation of mammary mTOR signaling and chronic reduction of mammary epithelial mass

While there is generally no consensus about how nutrients determine milk synthesis in the mammary gland, it is likely that the mechanistic target of rapamycin complex 1 (mTORC1) plays a role as a key integrator of nutritional and mitogenic signals that can influence a multitude of catabolic and anabolic pathways. The objectives of this study were to evaluate acute changes (<24 h) in translational signaling, in addition to chronic changes (14 d) in mammary gland structure and composition, in response to a severe feed restriction. Fourteen lactating Holstein dairy cows were assigned to either ad libitum feeding (n = 7), or a restricted feeding program (n = 7). Feed-restricted cows had feed removed after the evening milking on d 0. Mammary biopsies and blood samples were collected 16 h after feed removal, after which cows in the restricted group were fed 60% of their previously observed ad lib intake for the remainder of the study. On d 14, animals were sacrificed and mammary glands dissected. In response to feed removal, an acute increase in plasma nonesterified fatty acid concentration was observed, concurrent to a decrease in milk yield. In mammary tissue, we observed downregulation of the mTORC1-S6K1 signaling cascade, in addition to reductions in mRNA expression of markers of protein synthesis, endoplasmic reticulum biogenesis, and cell turnover (i.e., transcripts associated with apoptosis or cell proliferation). During the 14 d of restricted feeding, animals underwent homeorhetic adaptation to 40% lower nutrient intake, achieving a new setpoint of 14% reduced milk yield with 18% and 29% smaller mammary secretory tissue dry matter and crude protein masses, respectively. On d 14, no treat-ment differences were observed in markers of protein synthesis or mammary cell turnover evaluated using gene transcripts and immunohistochemical staining. These findings implicate mTORC1-S6K1 in the early phase of the adaptation of the mammary gland’s capacity for milk synthesis in response to changes in nutrient supply. Additionally,


INTRODUCTION
After over a century of research into milk yield responses of dairy cows to dietary nutrient supply, it can be surprising to realize that there is no universally accepted description of the mechanism by which dietary nutrients determine the rates of milk synthesis in the mammary glands (Akers, 2017, Schwab andBroderick, 2017).In well-fed cows, simple mass action effects of the precursors of milk protein, fat and lactose on ratecontrolling enzymatic steps in the biosynthetic pathways are too small to be of significance (Cant et al., 2003).The mechanistic target of rapamycin complex 1 (mTORC1) signaling pathway (Shimobayashi and Hall, 2014) shows great promise for explaining nutritional effects on the rate of synthesis of milk components in the mammary glands because this one kinase controls protein synthesis, lipogenesis, cell growth and proliferation in response to an integrated sensing of insulin and IGF-1 signals and intracellular amino acid and energy status, which are the very factors known to mediate dietary effects on milk yield (Cant et al., 2018, Pszczolkowski andApelo, 2020).Castro et al. (2016) deftly illustrated the potential reach of this signaling pathway over milk protein yield with a mathematical model of the hormonal and nutritional regulation of mTORC1 based on data from numerous incubations of mammary epithelial cells in vitro.In addition to mTORC1, there is an integrated stress response (ISR) pathway that controls global mRNA translation rate in response to amino acid and other nutrient deficiencies via phosphorylation of the key eukaryotic initiation factor 2 (eIF2; Proud, 2005).Glycogen synthase kinase-3 (GSK-3) is another second-messenger of insulin and IGF-1 signaling that accelerates mRNA translation by phosphorylating the eIF2B enzyme that activates eIF2 (Proud, 2005).The mTORC1, ISR and GSK-3 pathways mediate short-term (i.e., within a day) effects of amino acids, energy metabolites, insulin, and IGF-1 on mammary protein synthesis in vitro and in vivo (Appuhamy et al., 2012, Burgos et al., 2013, Arriola Apelo et al., 2014), but after several days of dietary intervention, activity state of these pathways in the mammary glands is not related to milk protein yields of cows (Doelman et al., 2015a, Doelman et al., 2015b, Nichols et al., 2016, Bajramaj et al., 2017).It is possible that early activation of mRNA translation by nutritional stimuli leads to larger epithelial cells, faster progression to mitosis, and ultimately greater epithelial mass with which to synthesize and secrete milk.That hypothesis inspired the current experiment in which a severe feed restriction of 40% (i.e., 60% of previous ad libitum intake) was chosen to alter milk yield nutritionally in cull cows, and mammary glands were biopsied within the first 24 h to assess translational signaling, and then collected at slaughter 2 weeks later for dissection and weighing of secretory tissue mass, counting of proliferating and apoptotic cells, and assessment of markers of mRNA translation, secretory differentiation, and cell turnover.

Animals and Experimental Design
All animal procedures were approved by the University of Guelph Animal Care Committee (AUP# 3316).Fourteen lactating Holstein-Friesian dairy cows (parity 3 ± 1.4, 224 ± 110.4 DIM; mean ± SD) that had been previously selected for culling from the research herd were housed in individual tie stalls at the Ontario Dairy Research Centre (Centre Wellington, ON) between July 2015 and April 2016.Animals used in this experiment were predominantly selected for voluntary culling due to reproductive issues.Animals selected for culling due to issues that may have affected mammary gland integrity (e.g., mastitis) were excluded from the study.Cows were blocked into pairs by enrolment date and average daily milk yield over the previous month of lactation and assigned to one of 2 feeding levels of a conventional TMR (Table 1): ad libitum (CON) or restricted to 60% of ad libitum intakes (RES) determined over the 3-d period immediately before the start of the experiment.Cows assigned to the RES treatment group had feed withdrawn from 1700h on d 0 until after mammary biopsies had been collected on d 1 (approx.Sixteen h total).Following collection of mammary biopsies, animals in the RES group were feed restricted for the remainder of the experiment.After the morning milking on d 14, animals were transported to the abattoir at the Department of Animal Biosciences, University of Guelph (Guelph, ON), where they were sacrificed by captive bolt stunning and exsanguination.

Production Variables
Animals were milked in place twice daily at approximately 0530h and 1700h, and the yield at each milking was recorded.Samples for milk composition were collected at both milkings one day before feed restriction (i.e., d −1), as well as at both milkings from d 9 to 13.Samples were analyzed for content of crude protein, fat, and lactose by infrared spectroscopy (Agriculture and Food Laboratory, University of Guelph).Milk compo- Net energy expenditure was calculated as the sum of maintenance energy requirements (0.08 × BW 0.75 ) and milk NE L output, and net energy balance was calculated as the difference between NE L intake and net energy expenditure.Fresh feed was delivered daily at approximately 1100h, and any refusals from the previous day were weighed to determine voluntary feed intake.Samples of TMR and refusals were collected daily, pooled by week (refusals additionally pooled by cow within week), and stored at −20°C until further analysis.Samples of TMR were analyzed for nutrient content by near-infrared spectroscopy at SGS Agrifood Laboratories (Guelph, ON; Table 1), while samples of both TMR and feed refusals were analyzed for DM content by oven drying at 100°C for 24 h (AOAC International, 2005, method 967.03) to determine voluntary DMI.

Mammary Biopsies
Mammary biopsies were collected on d 1 between 0900h and 1100h.Samples of tissue were collected from both rear quarters as described by Curtis et al. (2014).In brief, cows were sedated with 0.35 mL xylazine (Rompun, 20 mg/mL; Bayer, Shawnee Mission, KS) administered intravenously via the coccygeal vessels.The udder was then clipped and washed with soap and water, and the biopsy sites were sanitized with 4% chlorhexidine gluconate (E-Z Scrub 747; Becton, Dickinson and Co, Mississauga, ON), 70% ethanol and 2.5% chlorhexidine.Following sanitization, approximately 3 mL 2% lidocaine hydrochloride (100 mg/50 mL; Alveda Pharma, Toronto, ON) were administered subcutaneously at each biopsy site for local anesthesia.After confirming lack of sensation at the biopsy site, an incision of approximately 3 cm was made, and samples of secretory tissue (approx.500 mg) were retrieved using a drill and trocar (Farr et al., 1996).Samples were rinsed in phosphate-buffered saline (PBS) and snap frozen in liquid N 2 and stored at −80°C until analysis.Following collection of samples, pressure was applied to each biopsy site with sterilized cloths for approximately 15 min, or until bleeding ceased.Incisions were closed with surgical staples and sealed with Aluspray external bandage (Vetoquinol, Lavaltrie, QC).Following the procedure, 15 mL ketoprofen (Anafen, 100 mg/ mL; Merial Canada, Baie-d'Urfé, QC) were adminis-tered intramuscularly in the neck, and biopsy sites were checked at least twice daily until healed.

Blood Samples
On d 1 immediately before mammary biopsies and on d 13 at 1000h and 1200h, blood samples were collected by tail venipuncture into one EDTA-coated and one sodium heparin-coated blood collection tube (Becton, Dickinson and Co.).Samples were immediately stored on ice until centrifugation at 2,000 × g for 15 min.Plasma was aliquoted into polypropylene microcentrifuge tubes and stored at −20°C until analysis.Concentrations of plasma glucose, BHB and nonesterified fatty acids (NEFA) were determined by enzyme-linked spectrophotometry as described by Weekes et al. (2006).

Mammary Dissection
Immediately following sacrifice on d 14, udders were removed from the carcass and samples of secretory tissue were collected from the deep proximal region of each quarter.Tissue samples for quantification of DNA and gene expression (approx.Two g each) were rinsed in PBS, snap frozen in liquid N 2 and stored at −80°C until analysis.Samples for immunohistochemical staining (approx.50 g each) were washed in PBS and stored in 4% paraformaldehyde.Following sample collection, dermal tissue was separated from the udders and weighed, after which the udder was quartered and weighed.Tissue that was visibly non-secretory (e.g., adipose, ductal tissue) was separated, and both secretory and non-secretory tissues were weighed individually.Additional samples of secretory tissue (approx.200 g) were dried in a forced-air oven at 100°C until sample weight stabilized.Dried samples were ground using a blade grinder, then dried further at 110°C until sample weight stabilized, after which they were analyzed by wet chemistry at SGS Agrifood Laboratories using the following methods of AOAC International (2005): moisture (930.15);ash (942.05);crude protein (990.03);fat (954.02).

DNA and RNA Quantification
Samples of mammary secretory tissue (1 sample from each rear quarter per cow) collected via dissection on d 14 were ground under liquid N 2 using a mortar and pestle pre-chilled to −80°C and stored at −80°C until analysis.Duplicate samples from each animal (approx.50 -150 mg per sample) were weighed into pre-chilled microcentrifuge tubes.The weight of each sample was recorded, and samples were transferred to liquid N 2 until further processing.Tissue samples were subsequently homogenized in 1.5 mL RIPA lysis buffer (1% Triton X-100, 0.1% SDS, 50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.5% sodium deoxycholate) and centrifuged at 12,000 × g for 10 min at room temperature.Immediately following centrifugation, 2-µL aliquots of each supernatant were transferred to duplicate 0.5-mL PCR tubes, after which one tube was used to quantify RNA (Qubit RNA BR assay kit, catalog #Q10210; Invitrogen, Carlsbad, CA) and one to quantify DNA (Qubit 1X dsDNA HS assay kit, catalog #Q33230; Invitrogen).Both assays were performed using a Qubit 4 fluorometer (Invitrogen) following the manufacturer's instructions.Average tissue DNA & RNA content per animal was then multiplied by the total secretory tissue mass to estimate the total secretory tissue DNA & RNA mass per cow.

RNA Isolation and quantitative RT-PCR
Tissue samples collected on d 1 and d 14 were ground under liquid N 2 using a mortar and pestle chilled to −80°C.RNA was isolated from 50 to 100 mg of ground tissue using a phenol-chloroform extraction method (TRIzol Reagent; Invitrogen).The manufacturer's protocol was modified such that RNA pellets underwent an additional wash in 75% ethanol, and the wash fluid was aspirated from the pellets twice after each wash.RNA was resuspended in 50 µL of diethyl pyrocarbonatetreated water (Fisher Scientific, Hampton, NH) and stored at −80°C.RNA concentration and quality were determined using a NanoDrop 8000 spectrophotometer (Thermo Scientific, Waltham, MA).Absorbance was measured at 260 and 280 nm, and only samples that met or exceeded an A 260 /A 280 ratio of 1.8 were used for downstream analysis.RNA quality was confirmed by electrophoresis (2100 Bioanalyzer; Agilent Technologies, Santa Clara, CA), where samples were confirmed to have an RNA integrity number (RIN) greater than or equal to 6 before further analyses.
1-µg aliquots of RNA from 2 mammary tissue samples from each cow at each time point were treated with amplification grade DNase I (Invitrogen) following the manufacturer's protocol.cDNA was synthesized from 500 ng of DNase I-treated RNA with random hexamers using a High Capacity Reverse Transcription kit (Applied Biosystems, Waltham, MA) following the manufacturer's protocol.Quantitative RT-PCR was performed using PerfeCta SYBR Green FastMix (Quanta BioScience, Gaithersburg, MD) with a StepOnePlus real-time PCR system (Applied Biosystems).Primers for genes of interest were designed using NCBI primer-BLAST (Ye et al., 2012) to yield PCR products of 100 to 300 bp (Table A1).Fold change in gene expression relative to CON biopsy samples were calculated by the 2 -ΔΔCt method (Livak and Schmittgen, 2001) after normalization to reference gene expression.
The most stably expressed gene was determined by ANOVA of cycle threshold (C T ) value due to treatment, block, sampling time, and sample using the GLM procedure of SAS v9.4 (SAS Institute Inc., Cary, NC).For each gene, a coefficient of variation was calculated by dividing the total sum of squares of the model by 9 degrees of freedom and the mean C T value.The gene with the lowest coefficient of variation was found to be rps6kb1 so it was selected as the reference gene for normalization.

Gel Electrophoresis and Western Blotting
50 mg ground mammary tissue collected on d 1 were transferred to pre-chilled cryotubes containing 0.5 mL lysis buffer (1% Triton X-100, 0.1% SDS, 50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.5% sodium deoxycholate, 0.01% Halt protease inhibitor cocktail; Thermo Scientific).The mixture was homogenized, mixed under inversion for 1 h at 4°C, and subsequently centrifuged at 13,000 × g for 20 min at 4°C.The supernatant was recovered and stored at −20°C until analysis.Total protein content was determined with the Pierce BCA protein assay kit (Thermo Scientific) using bovine serum albumin (BSA; Millipore-Sigma, Darmstadt, Germany) as a standard.Aliquots of lysates were diluted with lysis buffer to achieve a protein concentration of 1.5 µg/µL.An equal volume of 2 × Laemmli sample buffer (100 mM Tris-HCl pH 6.8, 4% SDS, 20% glycerol, 0.2% bromophenol blue, 10% β-mercaptoethanol) was added, and samples were heated for 5 min at 90°C using a heating block.Fifteen µg protein were loaded per lane, resolved on 10% SDS-polyacrylamide gels and transferred to Immobilon-P PVDF transfer membranes (Millipore-Sigma).BLUeye pre-stained protein ladder (Froggabio, North York, ON) was run on each gel to evaluate protein molecular weight.Protein transfer was confirmed by staining membranes with Fast Green FCF dye, after which they were dried and stored at room temperature until immunoblotting.
Membranes were blocked for 1 h at 4°C in a solution of either 5% BSA in Tris-buffered saline-Tween (TBST) buffer or 5% skim milk powder in TBST based on recommendations provided with the primary antibody being used.After blocking, membranes were incubated overnight at 4°C with the primary antibody solution (Table A2), rinsed in TBST, and incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature.Membranes were then washed in TBST 6 times for 5 min each under agitation at room temperature, followed by application of Clarity Western ECL reagent (Bio-Rad, Hercules,
Coplin jars were filled with sodium citrate buffer (10 mM sodium citrate pH 6.0, 0.05% Tween 20) and heated in a water bath to 95°C.Slides were deparaffinized by washing 3 times in 100% xylene for 2 min each, followed by 3 washes in 100% isopropanol for 2 min each.Slides were then rehydrated in 70% isopropanol for 2 min, rinsed with double-distilled water, and washed twice in PBS for 5 min each at room temperature.Tissue was then permeabilized by incubating slides in preheated Coplin jars for 1 h.Following termination of the TU-NEL labeling reaction according to the manufacturer's protocol, slides were washed in PBS 3 times at room temperature for 5 min each.Slides were then blocked in 1% BSA PBS with Tween 20 for 30 min at room temperature.Excess blocking solution was discarded and slides were incubated with 50 µL Ki-67 primary antibody solution (Ki-67 rabbit mAb, product #9129, Cell Signaling Technology, Danvers, MA; 1:200 in 0.1% Triton X-100 PBS) in a humidity chamber for 2 h at room temperature.After washing slides 3 times in PBS for 5 min each, excess washing fluid was discarded and 50 µL of the secondary antibody solution (antirabbit IgG, product #5366, Cell Signaling Technology; 1:1000) was applied and slides were incubated for 1 h in a humidity chamber at room temperature.Slides were washed in PBS 3 times for 5 min each, after which 50 µL 4',6-diamidino-2-phenylindole (DAPI) was applied, a glass coverslip placed over the tissue section, and sealed with clear nail polish.Slides were stored at 4°C in dark conditions to retain fluorescence.
For each cow, mammary tissue sections from 2 quarters were selected at random, and the number of DAPIstained nuclei and cells positive for Ki-67 and TUNEL were counted in 10 randomly selected non-adjacent fields of view at 400 × magnification per tissue section.The technician performing counts was blinded to treatment.

Statistical Analyses
Before analysis, raw data for protein abundance and C T values were screened for potential outliers using a 2-step process.First, data were analyzed using the GLIMMIX procedure of SAS according to the linear models described below, with the addition of a compound symmetry covariance structure to model the correlation between duplicate measurements within each animal (protein abundance) or tissue (gene expression).The residuals from these analyses were then modeled using the ROBUSTREG procedure of SAS considering the same fixed effects as used in the linear mixed models, and automated detection of outliers was requested using the DIAGNOSTICS option.Forty-three C T values of 911 total (4.7%) and 10 of 243 (4.1%) values for protein abundance were detected as outliers and removed from subsequent analysis.Following removal of outliers, protein abundance values were averaged by animal before subsequent analysis and C T values were averaged by tissue within animal.Average C T values were then used to calculate relative fold change using the 2 -ΔΔCt method of Livak and Schmittgen (2001), with changes in gene expression normalized to expression of rps6bk1 and expressed relative to biopsy samples from the negative control group.Additionally, cell counts for DAPI, Ki-67 and TUNEL immunohistochemical data from individual fields of view were summed within animal before analysis.
Data pertaining to milk production and gene expression were analyzed using the GLIMMIX procedure of SAS according to the model, where y ijkl is the response of interest, μ is the intercept, Trt i is the fixed effect of treatment (CON or RES), Time j is the fixed effect of time, Trt i × Time j is the treatment-time interaction effect, block k is the random effect of block, cow l (block k ) is the random effect of cow within block, and ε ijkl is the residual error variance.For production outcomes measured over several days (e.g., milk yield, DMI), repeated measures on cows within blocks were modeled using a first-order autoregressive covariance structure.For outcomes measured at 2 time points (e.g., blood metabolites, gene expression), a compound symmetry covariance structure was used.
When analyzing data pertaining to gene expression, it was found that the statistical model was overspecified when the effect of block was included, and it was therefore removed from the model for those data.Data pertaining to mammary tissue composition, cell population dynamics and protein abundance were analyzed according to the model, where all terms are as previously described.A compound symmetry covariance structure was used to accommodate negative covariance between blocks (Stroup et al., 2018).For immunohistochemical data, the proportions of Ki-67-and TUNEL-stained cells relative to DAPIstained cells were modeled using a binomial distribution and the default logit link function, while the total number of stained cells were modeled using a Poisson distribution.For all analyses, the Kenward-Roger correction was used to adjust the denominator degrees of freedom, and a Newton-Raphson optimization with ridging was requested using the NLOPTIONS statement.For outcomes with a Trt i × Time j effect, tests of simple effects were requested using the SLICEDIFF option of the LSMEANS statement, with associated Pvalues adjusted using the Tukey-Kramer method where applicable.Tests were considered statistically significant where P < 0.05 and were considered trends where 0.05 ≤ P < 0.15.

Production
It was confirmed that during the restricted feeding period (i.e., d 1 to d 13), animals assigned to the RES treatment consumed 42% less feed per day relative to CON animals (23.1 vs. 39.8 ± 1.26 Mcal NE L /d, mean ± SED; P < 0.001).Daily milk yield did not differ (P = 1.00) between CON and RES animals during the 7-d period immediately before experimental period (Figure 1).When comparing yields at individual milking times (Figure 1), milk yield at the evening milking of d 0 did not differ between groups (P = 0.64); at the subsequent morning milking (d 1), the RES group milk yield tended to be 2.97 kg lower (P = 0.07), and by the evening milking on d 1, animals in the RES group produced 4.82 kg less (P = 0.01) compared with CON.Between d 0 and d 13, animals assigned to the RES group produced 6.82 kg/d less milk (P = 0.02) on average (Figure 1).Average intake and production between d 9 and d 13 are presented in Table 2. Yields of milk protein and lactose were decreased by 22.9% (P = 0.01) and 24.2% (P = 0.01), respectively, while milk fat yield tended to decrease by 14.6% (P = 0.12; Table 2).As a result, ECM yield decreased by 14.0% and milk NE L output decreased by 21.3% with a difference in net energy balance of 9.35 Mcal/d (P ≤ 0.04; Table 2).No differences were observed in milk component concentrations.

Blood Metabolites
Responses in plasma metabolite concentrations are presented in Table 3.While there was no response in BHB due to treatment (P = 0.47), concentrations at d 13 in RES animals tended to be increased by 39.3% (P = 0.11) relative to those observed on d 1 in the same treatment group; this temporal response was not observed in the control group (P = 0.78).Concentration of NEFA was 236 µM higher (129%; P < 0.01) in RES animals on d 1; by d 13, no difference in concentration was observed between treatment groups (Table 3).In contrast to BHB, concentration of NEFA at d 13 in RES animals was decreased by 53.0% (P < 0.01)  relative to the concentration at d 1 in the same group; this temporal response was not observed in the CON group.No differences were observed in plasma glucose concentrations (P ≥ 0.37; Table 3).

Mammary Tissue Composition
Data pertaining to mammary tissue composition are provided in Table 4. Whole-udder secretory tissue mass tended to be lower (P = 0.13) in RES animals, with trends to be higher in fat content (P = 0.08) and lower in CP content (P = 0.08; Table 4); this contributed to reduced secretory tissue crude protein mass (P = 0.05).No differences were observed in mammary cell population dynamics due to treatment (Table 5).

Protein Abundance
Responses in protein abundance as determined by Western blotting are presented in Figure 2. In response to acute feed removal, the abundance of cyclin D1 decreased by 31.2% (P = 0.04) relative to control, while total S6K and mTORC1 Ser2448 tended to decrease by 35.7% (P = 0.08) and 51.5% (P = 0.12), respectively.No differences were observed in abundance of mTORC1 or (P = 0.17), however the ratio of mTORC1 Ser2448 : mTORC1 decreased by 16.0% (P = 0.04) in RES animals relative to CON.

mRNA Expression
Gene expression normalized to rps6kb1 in mammary samples collected at 16 h from control animals is presented in Table 6.In biopsy tissue collected after 16 h feed removal, expression of ccnd1 and bax were decreased (P ≤ 0.02) due to feed restriction, with trends for decreased expression of eif2a1 and increased expression of rna45s5 (P ≤ 0.14).These differences were not present in tissue collected at dissection following 14 d of restricted feeding (Table 6).In tissue collected at d 14, ddit3 tended to decrease (Table 6) in response to feed restriction.

DISCUSSION
Our feed restriction experiment was designed to explore the hypothesis that an acute, nutritionally-induced change in protein translational signaling in cells of the mammary glands produces a chronic change in secretory cell number and/or activity.The hypothesis was developed to reconcile observations that mTORC1, ISR and GSK-3 signaling to the translational apparatus of mammary cells is upregulated by anabolic nutrients and hormones within minutes to hours, but after several days of a milk yield-stimulating nutritional treatment, these signaling pathways do not appear to remain upregulated in the mammary glands (as reviewed by Cant et al., 2018).It is possible that the early activation of translational signaling leads to a greater capacity for milk synthesis in the mammary glands through an  increase in secretory cell number and/or secretory capacity per cell.
For our experiment, feed restriction was chosen as the nutritional perturbation that would affect milk synthesis rate, and mammary tissue was collected after 16 h and 14 d of restriction.The feed restriction period started with 16 h of a complete fast to rapidly induce the restricted state that was maintained thereafter by feeding at 60% of ad lib intake, i.e., 40% feed restriction.The DM available for digestion during the initial fasting phase can be estimated assuming first-order kinetics of ruminal DM disappearance (sum of degradation and passage) according to a rate constant k, where disappearance = k × mass 0 × e -kt , and mass 0 is the rumen DM mass at time 0. For a typical k-value of 10%/h (Bruining et al., 1998), after 16 h of not eating, DM disappearance rate will be 20% of the ad lib value.
Over the entire 16-h feed withdrawal, DM disappearance will average mass 0 /t × (1 -e −kt) , equivalent to 50% of the ad lib value.This rough estimate shows that the preliminary 16-h feed withdrawal approximated a rapid 40% feed restriction.

Transition to a New Setpoint of Milk Production
Within the first 12 h of feed removal, milk yield began to decrease in RES animals and by 24 h, it was at the level at which it would remain for the duration of the 2-week study.Others have reported a similar time course of the decline in milk yield following feed restriction (McGuire et al., 1995, Chelikani et al., 2004, Toerien and Cant, 2007, Abdelatty et al., 2017).Likewise, the increase in plasma NEFA concentration within the first 24 h of feed removal is consistent with previous reports (McGuire et al., 1995, Chelikani et al., 2004, Toerien and Cant, 2007), indicating faster mobilization or slower accumulation of body fat reserves in response to feed removal (McNamara andHillers, 1986, Leduc et al., 2021).The lack of a change in plasma BHB concentration during the first 24 h of a fast was also observed by Toerien and Cant (2007) and represents a delay between the mobilization of body fat reserves and β-oxidation of the resulting NEFA in the liver.Bjerre-Harpoth et al. (2012) and Pires et al. (2019) reported a similar delay between NEFA and BHB buildups in plasma following the initiation of feed restriction in early lactation cows but the BHB concentration had already started to rise by 24 h.In mid-to late-lactation cows, restriction of NE intake to 48% of requirement only affected NEFA concentrations and did not cause an increase in plasma BHB even after 4 d (Bjerre-Harpoth et al. ( 2012), possibly because of the higher plasma glucose concentrations that facilitate BHB oxidation in cows at these later stages of lactation.Twelve of the 14 animals used in the present study would be classified as mid-to late-lactation based on the criteria used by  5. Counts and proportions of proliferating and apoptotic cells in mammary tissue from lactating dairy cows (n = 7 per treatment) after 14 d of ad lib (CON) or 40% restricted (RES) intake.Values are the total cell counts from 10 fields of view in each of two mammary quarters, summed within animal 1 LCL = lower 95% confidence limit of mean; UCL = upper 95% confidence limit of mean.By d 13 of 40% feed restriction relative to ad libitum intake, NEFA concentrations were no longer elevated in plasma (Table 3), suggesting that fatty acid sequestration in adipose tissue had returned to baseline, or that plasma NEFA were being removed by other tissues at a higher efficiency than at d 1.The drop in NE L balance from a positive to negative value in RES cows (Table 2) supports the latter explanation.In either case, RES cows appeared to have reached a new steady-state of nutrient flow at a lower milk yield with no changes in plasma glucose, NEFA or BHB concentrations compared with CON cows.Although NE L intake decreased by approximately 40% as designed, ECM yield decreased by only 14%.It is likely that the remainder of this imbalance was accounted for by increases in body reserve mobilization, as well as reductions in splanchnic energy expenditures due to prolonged reductions in feed intake (Johnson et al., 1990).The results show that a setpoint of lower milk production was established within 1 to 2 weeks of 40% feed restriction of ad lib intake when there was complete feed withdrawal during the first day.

Lower Milk Production Setpoint Associated with Smaller Mammary Mass
In support of our hypothesis of early signaling events, the phosphorylation state of mTOR suggested that mammary translational signaling was acutely inhibited within the first 16 h of feed removal.Phosphorylation of mTOR at Ser2448 is carried out by S6K1, which is activated by mTORC1 in response to insulin and other growth factors, amino acids, and cellular energy status (Magnuson et al., 2012), all of which may have been diminished to some extent in mammary tissue of feed-restricted cows (Leduc et al., 2021).In addition to mRNA translation, S6K1 also promotes ribosome biogenesis by activating transcription of ribosomal genes including rna45s5 (Chauvin et al., 2014).The decreased expression at 16 h of eif2s1 and rna45s5, which play key roles in mRNA translation and ribosome biogenesis, respectively, are consistent with early mTORC1-S6K1 downregulation.
The trend for decreased abundance of xbp1s in response to 16 h of feed removal implicates endoplasmic reticulum biogenesis in the mammary response (Hetz andPapa, 2018, Hetz et al., 2020).The xbp1s translation product is a transcription factor involved in differentiation of specialized secretory cells, including those of the mammary gland (Moore and Hollien, 2012, Davis et al., 2016, Yonekura et al., 2018).We previously found that stimulation of milk protein production in  cows with abomasal amino acid infusion was associated with elevated mammary expression of xbp1s (Nichols et al., 2017).Subsequently, Sharmin et al. (2021) reported that IGF-1 induced xbp1s expression in bovine mammary epithelial cells in an mTORC1-dependent manner.Thus, it is possible that xbp1s expression at 16 h decreased due to early downregulation of mTORC1, in accord with our other findings.
In addition to indications that protein synthesis and secretory epithelial cell differentiation were negatively impacted during the first day of feed deprivation, the decreased expression of ccnd1 mRNA and abundance of cyclin D1 protein in mammary tissue suggest an acute downregulation of cell proliferation as well.Cyclin D1 is a gatekeeper for initiation of proliferative DNA synthesis whose mRNA expression is increased by mitogenic signaling through numerous transcription factors (Pawlonka et al., 2021).Translation of ccnd1 is also considered to be mTORC1-dependent (Musgrove, 2006).The decline in cyclin D1 expression in our experiment indicates there was a smaller proportion of cells in the tissue sample entering into the DNA synthesis phase of the cell cycle, and it was likely connected to the lower mTORC1 activity during the first day of feed withdrawal.It is of note that expression of bax, encoding the pro-apoptotic BAX protein, was downregulated in mammary tissue collected via biopsy at 16 h.However, the abundance of BAX protein did not differ between treatments, in contrast to the findings for ccnd1 and cyclin D1 described above.If we assume that changes in gene expression precede changes in protein abundance, this could indicate that at the time of mammary biopsy, cells were transitioning to a state of reduced apoptosis.However, recent work has highlighted that interactions and (co)localization of BCL-2 family proteins within the cell may be the biggest determinant of cell fate, in contrast to the previously used rheostat model, where the balance between pro-and anti-apoptotic factors determined cell survival (Kale et al., 2018).As such, we believe the data do not provide enough information to make firm conclusions regarding acute apoptosis in the mammary gland due to feed removal.
After a period of slower protein synthesis, endoplasmic reticulum biogenesis, and cell proliferation in the mammary glands, one might expect mammary secretory tissue mass to decrease.Indeed, after 2 weeks on treatment, the mass of both secretory and non-secretory mammary tissue tended to decrease, and secretory CP mass decreased, but secretory DNA mass was not significantly affected.It should be noted, however, that while mammary tissue was sampled in such a way as to minimize RNA degradation, the samples collected may not accurately reflect the heterogeneous composition of the udder, and as such may have impacted the estima-tion of RNA and DNA content when extrapolating to the whole udder.During the first 11 wks of lactation on an all-forage diet of 1.28 Mcal/kg NE L that caused a 38% reduction in milk yield compared with a control diet of 1.61 Mcal/kg, cows in the experiment of Dessauge et al. ( 2011) exhibited a 38% reduction in udder mass.In cows given a concentrate allocation of 3 kg/d compared with 9 kg/d, Gibb et al. (1992) reported an 11% decrease in milk yield and a 17% decrease in udder mass.We observed a 14% drop in ECM yield and 18% drop in mammary secretory tissue DM mass.Similar equivalences in milk production per unit of mammary secretory mass have been described for the effects of stage of lactation post-peak (Knight and Peaker, 1984, Gibb et al., 1992, Capuco et al., 2001) or cow genetic merit (Knight, 2000).These data corroborate the notion that chronic impairment of nutrient supply on daily milk yield are enacted through changes in mammary epithelial mass.
Despite the reductions in udder mass and milk yield induced by 14 d of feed restriction to 60% of ad lib intake, there were no differences compared with CON on d 14 in mammary mRNA expression of cell turnover and protein synthesis markers, or in proportions of proliferating or apoptotic cells.In addition to apoptosis, mammary epithelial cells are sloughed into milk during lactation in a process known as exfoliation (Herve et al., 2016).In the present experiment, we did not quantify cell loss though this mechanism, though it has been previously estimated that approximately 1.6% of epithelial cells may be lost in this method over the course of lactation (Boutinaud et al., 2019).Herve et al. (2019) reported the same lack of differences in proliferating and apoptotic cell percentages after 4 weeks of 20% feed restriction post-peak lactation that decreased milk yield 9%.When feed intake was more severely restricted, by 40% from 2 weeks before parturition until 11 wks into lactation, milk yield dropped 38% and, 13 wks after treatment started, apoptotic cells in mammary tissue were twice as numerous as those in unrestricted cows (Dessauge et al., 2011).Proliferating cell concentration was not affected.Norgaard et al. (2005) observed the opposite: restricting energy intake for the first 8 wks of lactation to decrease milk yield 18% was associated with no difference in apoptotic cell concentration in mammary biopsies at wk 8 compared with the unrestricted control but there was a smaller proportion of proliferating cells.By wk 16 of the energy restriction, there were no treatment effects on apoptotic or proliferating cell counts despite a sustained depression in milk yield of 24% (Norgaard et al., 2005).After peak lactation, it is as if there is a short window for enacting temporary changes in cell turnover rates; the rates of secretory cell apoptosis and/or proliferation can be changed to allow the mammary cell population to reach a new state associated with a lower milk yield, after which the new state can be maintained at the old proliferation, differentiation, and death rates.Such a transient alteration in secretory cell dynamics is evident in cases where a switch in dietary energy intake changes daily milk yield of cows but does not affect the rate of decline in yield as their lactations progress (Chen et al., 2016, van Hoeij et al., 2017), which is almost entirely explained by the balance between secretory cell gain and loss (Dijkstra et al., 1997, Capuco et al., 2001).What is most puzzling about this homeorhetic adaptation scenario is how it is established that the time for modifying gain and loss rates is over and the new mammary gland size is appropriate to the new workload.

Secretory Activity vs. Cell Number
While it is possible to visually distinguish secretory mammary tissue from non-secretory tissue, such as ductal tissue, one of the limitations of this study is that we are unable to differentiate between terminally differentiated secretory cells from non-secretory cell types (e.g., mammary stem cells) within the secretory tissue.The 18% decrease in mammary tissue mass that we observed to be associated with a 14% decrease in ECM yield was related to a 29% lower CP mass.Because the 15% decrease in mean DNA mass was not statistically significant, we cannot conclude that epithelial cell number went down due to restriction.However, the daily ECM yield per g DNA, or per cell, was not calculated to be reduced to any extent, on average, between treatments.The ratio of CP: DNA in mammary tissue, which represents size or activity of cells, was also not significantly affected although it decreased 16% on average.Likewise, the RNA: DNA ratio representing protein synthetic capacity of epithelial cells decreased non-significantly.The lower CP mass at 2 wks, which is consistent with early downregulation of mTORC1-S6K1 signaling, could represent a change in both secretory cell number and activity per cell, the 2 of which cannot be teased apart with the current data.

CONCLUSIONS
Following 16 h of complete feed removal, approximating the nutrient supply of a rapid 40% reduction in feed intake, an acute increase in net fatty acid mobilization from adipose tissue was elicited in an effort to support the energetic demands of lactating Holstein dairy cattle.Accompanying this was an acute decrease in milk yield, and a reduction in mTORC1-S6K1 signaling and expression of markers of protein synthesis, endoplasmic reticulum biogenesis and cell proliferation in the mammary glands.Following 13 d at a 40% reduction in feed intake, a new homeostatic setpoint was established, and milk yield was sustained at a decreased level from a reduced mass of mammary tissue.Effects on cell proliferation and apoptosis were not detected at 14 d.The findings support the hypothesis that nutritionallymediated downregulation of mTORC1-S6K1 signaling within the first day(s) of feed restriction leads to a smaller mammary CP mass and reduced milk synthesis per day.
Seymour et al.: MTORC1 ACUTELY ATTENUATES MILK SYNTHETIC CAPACITY Seymour et al.: MTORC1 ACUTELY ATTENUATES MILK SYNTHETIC CAPACITYCA) and imaging on a ChemiDoc XRS+ system (Bio-Rad).Protein abundance was normalized to imaging exposure time and β-actin band intensity.

Figure 2 .
Figure 2. Protein abundance measured by Western blot in mammary biopsies collected from lactating Holstein cattle (n = 7 per treatment) following 16 h feed removal.Values presented are average abundance in RES animals normalized to both image exposure time and β-actin band intensity, expressed as a percentage of CON.

Table 2 .
Average intake and milk production in lactating dairy cows 1 ECM = energy corrected milk yield.
Seymour et al.: MTORC1 ACUTELY ATTENUATES MILK SYNTHETIC CAPACITY

Table A1 .
Seymour et al.: MTORC1 ACUTELY ATTENUATES MILK SYNTHETIC CAPACITY Primer sequences for quantitative real-time PCR in bovine mammary tissue