Dose response to post-ruminal urea in lactating dairy cattle

Inclusion of urea in dairy cattle diets is often limited by negative effects of high levels of feed urea on dry matter intake (DMI) and efficiency of rumen N utilization. We hypothesized that supplying urea post-ruminally would mitigate these limitations and allow greater inclusion of urea in dairy cattle diets. Four rumen-fistulated Holstein-Friesian dairy cows (7 ± 2.1 lactations, 110 ± 30.8 d in milk; mean ± standard deviation) were randomly assigned to a 4 × 4 Latin square design to examine DMI, milk production and composition, digestibility, rumen fermentation, N balance, and plasma constituents in response to 4 levels of urea continuously infused to the abomasum (0, 163, 325, and 488 g/d). Urea doses were targeted to linearly increase the crude protein (CP) content of total DMI (diet plus infusion) by 0, 2, 4, and 6% and equated to 0, 0.7, 1.4, and 2.1% of expected DMI, respectively. Each 28-d infusion period consisted of a 7-d dose step-up period, 14 d of adaptation, and a 7-d measurement period. The diet was fed ad libitum as a total mixed ration [10.9% CP, 42.5% corn silage, 3.5% grass hay, 3.5% wheat straw, and 50.5% concentrate (dry matter basis)] and was formulated to meet 100, 82, and 53% of net energy, metabolizable protein, and rumen-degradable protein requirements, respectively. Linear, quadratic, and cubic effects of urea dose were assessed using polynomial regression assuming the fixed effect of treatment and random effects of period and cow. Dry matter intake and energy-corrected milk yield responded quadratically to urea dose, and milk urea content increased linearly with increasing urea dose. Apparent total-tract digestibility of CP increased linearly with increasing urea dose and ruminal NH 3 -N concentration responded quadratically to urea dose. Mean total VFA concentration was not affected by urea dose. The proportion of N intake excreted in feces decreased linearly and that excreted in urine increased linearly in response to increasing urea dose. The proportion of N intake excreted in milk increased linearly with increasing urea dose. Urinary urea excretion increased linearly with increasing urea dose. Microbial N flow responded cubically to urea dose, but the efficiency of microbial protein synthesis was not affected. Plasma urea concentration increased linearly with increasing urea dose. Regression analysis estimates that when supplemented on top of a low-CP diet, 179 g/d of post-ruminal urea would maximize DMI at 23.4 kg/d, corresponding to a dietary urea inclusion level of 0.8% of DMI, which is in line with the current recommendations for urea inclusion in dairy cattle diets. Overall, these results indicate that post-ruminal delivery of urea does not mitigate DMI depression as urea dose increases.


INTRODUCTION
Dietary urea is an effective rumen-available N source as it is rapidly hydrolyzed upon consumption and increases ruminal ammonia concentration (Veen and Bakker, 1977;Teller and Godeau, 1986;Boucher et al., 2007).However, increased ruminal ammonia concentrations can result in decreased efficiency of N use by ruminal microbes (Satter and Slyter, 1974;Bach et al., 2005;Boucher et al., 2007).Ammonia that is not utilized for microbial protein synthesis is absorbed from the gastrointestinal tract (GIT) into portal blood, where it is detoxified to urea in the liver.Hepatic urea production contributes significantly to the endogenous N economy of ruminants (Lapierre and Lobley, 2001).In dairy cows, recycling of endogenous urea-N to the GIT averages 61% of digestible N intake (Nichols et al., 2022), which can make a net contribution to MP supply if the recycled N is used for microbial protein synthesis.The magnitude of urea recycling and its contribution to microbial protein synthesis in the rumen depends on luminal pH and concentration of ammonia and VFA (Abdoun et al., 2007;Abdoun et al., 2010;Lu et al., 2014) which play a significant role in regulation of urea flux through facilitative urea transporters.Aquaglyceroporins may also play a role in urea movement into Dose response to post-ruminal urea in lactating dairy cattle K. Nichols,1* R. Rauch, 1 L. Lippens, 2 D. J. Seymour, 1 and J. Martín-Tereso 1 the gastrointestinal tract (Walpole et al., 2015;Scott et al., 2020;Zhong et al., 2020).The proportion of urea entering circulation that is recycled to the GIT decreases and urinary N excretion increases as dietary CP intake increases (Reynolds and Kristensen, 2008).
Hypophagic effects and altered feed intake patterns (e.g., less time spent eating, or slower eating during the meal period; Coppock et al., 1976;Kertz et al., 1982) have been reported when dietary urea inclusion approaches or exceeds 1% of dietary DM (Wilson et al., 1975;Polan et al., 1976;Brito and Broderick, 2007).In contrast, other studies report that feed intake was not affected by the addition of urea to the diet of dairy cows at levels above 1% of DMI (Clark et al., 1973;Kertz and Everett, 1975).Animal response to urea intake at or above this boundary may vary based on the approach to urea delivery (i.e., adaptation time to the diet, feeding a TMR versus concentrate supplements; Plummer et al., 1971;Coppock et al., 1976) and the composition of the rest of the diet (i.e., fermentability and non-protein N (NPN) content of forages; Huber et al., 1968;Poos et al., 1979;Kertz, 2010).
Given the relationship between ruminal ammonia concentration and urea recycling, some studies have aimed to reduce the canonical ammonia peak that occurs after a meal, improve the efficiency of N capture by rumen microbes, and maintain DMI at higher dietary urea inclusion rates by supplementing protein or NPN post-ruminally.Egan and Moir (1965) demonstrated that a pulse dose of urea infused into the duodenum increased feed intake and digestion of cotton thread in the rumen of sheep, and Wickersham et al. (2009) observed that increased urea-N recycling supported OM digestibility during RUP supplementation in steers.Urea derivatives that slow or resist ruminal urea degradation (e.g., biuret, isobutylidene diurea) have been shown to mitigate peaks of ruminal ammonia concentration upon consumption when compared with traditional urea (Komatsu and Sakaki, 1971;Veen and Bakker, 1977;Smith, 1986).Carvalho et al. (2020) demonstrated in nonlactating Holstein-Friesian heifers that continuous infusion of urea (1.7% of observed DMI) into the abomasum resulted in a more stable ruminal pH, a lower ruminal ammonia-N concentration throughout the day, and a 10% increase in apparent total-tract digestibility (ATTD) of NDF compared with an iso-nitrogenous pulse dose of urea into the rumen.Oliveira et al. (2020) demonstrated that continuous infusion of urea (1.4 ± 0.07% of observed DMI) into the abomasum of nonlactating Nellore heifers resulted in lower ruminal pH and ammonia-N concentration compared with continuous or pulse-dose ruminal urea delivery.Further, more microbial N was produced per kg of digestible OM and more of this microbial N originated from recycled N with post-ruminal urea compared with the ruminal urea pulse dose.Dry matter intake was not affected by location of urea supply in the study of Carvalho et al. (2020) nor Oliveira et al. (2020).
To our knowledge, post-ruminal supplementation of urea had not been evaluated in lactating dairy cattle.Therefore, we aimed to determine DMI, milk production, rumen fermentation, digestibility, and N metabolism of lactating dairy cattle in response to increasing doses of urea supplied post-ruminally.Our primary hypothesis was that post-ruminal delivery of urea would mitigate the negative effects of an increasing dose of urea on DMI.Our secondary hypothesis was that postruminally infused urea would be recycled to the rumen and support microbial protein synthesis as urea dose increased.

Experimental Design
This study was conducted at the dairy research facility of Trouw Nutrition Agresearch (Burford, Ontario, Canada) from February to May 2020.Animal procedures were approved by the Animal Care and Use Committee at Trouw Nutrition Agresearch and complied with the guidelines set forth by the Canadian Council on Animal Care (2009).Four rumen-fistulated, Holstein dairy cows (7 ± 2.1 lactations; mean ± SD) producing an average (mean ± SD) of 35.6 ± 1.80 kg milk/d at 110 ± 30.8 DIM and 698 ± 64.8 kg BW before the first experimental period were randomly assigned to treatment sequences within a 4 × 4 Latin square design.Experimental periods (total 28 d in length) consisted of a 7-d dose step-up period where the infused urea dose (g/d) was increased incrementally, followed by a 14-d period of adaptation to the final urea dose and a 7-d measurement period.Each period immediately succeeded the previous period.

Diet, Feeding, and Treatment Infusions
Cows were housed in tie stalls with individual and free access to fresh drinking water throughout the entire experiment.Cows were fed a low CP (10.9%,DM basis) basal diet as a TMR consisting of 42.5% corn silage, 3.5% grass hay, 3.5% wheat straw, and 50.5% concentrate on a DM basis (Table 1).The TMR was formulated to meet 100, 82, and 53% of NE L , MP, and RDP requirements (CVB, 2018), respectively, for cows producing 40 kg/d of milk containing 4.4% fat and 3.3% protein.The fresh TMR was mixed using a self-propelled mixer wagon equipped with an electronic weighing scale and allocated once daily between 0900 and 0930 h.Cows were acclimated to the basal TMR for 14 d before the first infusion period and were fed ad libitum for the entire experiment.Fresh feed allowance was adjusted daily to maintain at least 3 kg of feed refusal from individual cows to ensure ad libitum intake was observed.Treatments consisted of 4 linearly increasing doses of abomasally infused urea (0, 163, 325, and 488 g/d; Pestell Minerals and Ingredients, New Hamburg, ON, Canada) targeted to incrementally increase N intake such that the CP content of the diet plus infusion increased on each treatment by 0, 2, 4, and 6% (on a DM basis) over the basal diet.The urea doses corresponded to 0, 0.7, 1.4, and 2.1% of the mean DMI of the 4 cows observed during the final 5 d of the diet acclimation period (23.7 kg DM/d).The amount (g/d) of infused urea was kept constant within each treatment throughout the entire experiment, regardless of any observed changes in DMI.Due to potential risk of ammonia toxicity with urea infusion, cows were monitored daily for signs of abnormal behavior and treatments were not blinded to researchers or animal caretakers.The abomasal infusion device has been described by Nichols et al. (2019).Infusion lines were placed in the abomasum via the rumen cannula 7 d before the first experimental period.They were checked twice daily for correct placement and patency throughout the entire experiment to prevent infusion of urea into the rumen.
The daily dose of urea for each treatment was dissolved in 10 L of tap water.Treatment batches were mixed no more than 2 d in advance of delivery to the animal and stored at room temperature.The 10-L infusate batches were replenished at the same time each day and infused via a Watson-Marlow 205U/CA multi-channel peristaltic pump (Wilmington, MA) at a rate of 6.95 mL/min to facilitate continuous infusion.

Measurements and Sample Collection
Cows were milked daily at 0630 and 1730 h and milk production was measured and recorded electronically (DeLaval DelPro MU480; Peterborough, ON, Canada).Milk samples were obtained during the morning and afternoon milkings from each cow throughout the entire experiment via automatic samplers on the milking units that collected a fixed volume of milk per kg produced.Milk samples were preserved with bronopol, stored at 4°C, and analyzed within 4 d.Daily fresh feed intake was calculated as the difference in weight between allocated and refused feed after a 24-h period.Samples of corn silage, grass hay, wheat straw, and TMR were collected weekly, pooled per experimental period, and stored at −20°C until analysis.Samples of the concentrate were collected at the feed mill (Trouw Nutrition, St. Mary's, ON, Canada) at the time of manufacture and stored at −20°C until analysis.Feed refusals were collected and weighed daily to determine daily feed intake.
A 72-h total collection of urine and feces from each cow was conducted during each period (balance period; 1000 h on d 22 until 1000 h on d 25) to facilitate the estimation of N balance.Before the start of each balance period, cows were fitted with a urine funnel device which allowed urine to be collected directly into a 20 L carboy secured on the ground approximately 2 m behind the cow.The urine funnels were constructed by securing a 5 L rubber re-breathing bag (Upper Grand Veterinary Services, Guelph, ON, Canada), which served as the urine reservoir, to a specially designed rubber base.The base was cut from flexible rubber sheets (1.5 mm thickness) and was shaped to cover the external genitalia with flaps that wrapped over the pins alongside the tail head of the cow.A hole was cut in the rubber base at the location of the vulva to allow urine to flow into the re-breathing bag.The funnel was secured to the cow using adhesive glue (Bison Kit; Bison International B.V., Goes, the Netherlands).No glue was placed on the perineum, but a sponge was used to fill the space and eliminate contamination of feces into the urine funnel.Cows were continually monitored during the total urine and fecal collection period to ensure that the urine funnel remained functional and free of feces.The urine funnel was attached via the opening of the re-breathing bag to a braided PVC hose (internal diameter 1.91 cm; approximately 3 m in length) which was connected to the carboy.A vacuum was created across the carboy using a braided PVC hose (internal diameter 0.95 cm; approximately 3 m in length) attached to the carboy and a vacuum pump (Kozyvacu TA450; Kozyard LLC, Mercer Island, WA).At the moment of urination, the vacuum pump was manually turned on to direct all urine into the carboy resulting in no residual urine in the hose between the cow and the carboy.Before collection began, 250 mL of 20% (wt/ wt) H 2 SO 4 was added to the carboy.The same volume of acid was added to the carboy every 4-h during the balance period.This volume of acid was determined to be sufficient to acidify urine to pH ≤2 based on estimations of urine output and measured pH from earlier observations by our research group (J.B. Daniel, Trouw Nutrition Research and Development, Amersfoort, the Netherlands, personal communication).Urine was emptied from the carboy every 24 h into a tightly sealed composite container.At the end of the 72-h balance period, total urine was weighed, manually mixed, measured using an electronic pH meter to confirm that urine was maintained at pH ≤2, and subsampled.Urine samples were stored at −20°C until analysis.
To facilitate total fecal collection, stall dividers constructed from high density polyethylene boards were se-cured along the stall stanchions before the start of each balance period to prevent cross-contamination of feces between cows.Bedding material was removed from the stalls and the manure gutter was covered with rubber matting to allow clean and total collection of feces from the ground.Feces was collected during or immediately after defecation and was pooled in an individual container per cow.Collected feces was weighed, manually mixed, and quantitatively sampled every 24 h (5 g/ kg feces produced over 24 h).Daily fecal samples were added to a composite sample that was stored at −20°C until analysis.An additional milk sample (5 g/kg milk) was collected and pooled by cow at each milking during the balance period and stored at −20°C until analysis.Cows were weighed 2 d before the start of the experiment and at the end of each balance period.
Blood and rumen fluid samples were collected from each cow on d 26 of each period.Blood samples were collected from the coccygeal vessels by venipuncture into 10 mL sodium heparin and potassium EDTA Vacutainers (Becton Dickinson, Rutherford, NJ) at 0930, 1000, 1100, 1300, 1500, 1700, 1900, and 2100 h.Collection tubes were immediately placed in ice and centrifuged at 3,000 × g at room temperature for 15 min.Plasma from each time point was transferred into polypropylene tubes and stored at −20°C until analysis.Ruminal contents were collected from different sites (reticulum, caudal, ventral, and dorsal sacs; approximately 100 mL pooled sample in total) every 30 min beginning at 0930 h and ending at 2100 h.Samples were immediately strained through 4 layers of cheesecloth.Thereafter, 10-mL subsamples of rumen fluid filtrate were preserved with 2 mL of 25% (wt/vol) metaphosphoric acid pending analysis of NH 3 , and 1-mL subsamples were preserved with 1 mL of 5% (vol/vol) H 3 PO 4 pending analysis of VFA.Samples were stored at −20°C until analysis.
Rumen pH loggers (LRCpH T4 logger, Dascor, Escondido, CA) were inserted into each cow through the rumen fistula on d 21 and removed on d 28 of each period.Loggers were calibrated in buffer solutions of pH 4 and 7 before insertion and after removal and were programmed to record mV values every 30 s. Recordings were converted from mV to pH using the linear relationship established from the calibrations before insertion and after removal, where a linear drift in mV recording over time was assumed.

Analytical Procedures
Milk samples from each morning and afternoon milking were analyzed for fat, CP, lactose, and urea N contents by mid-infrared spectroscopy (Foss MilkoScan FT3; Agriculture and Food Laboratory, University of Guelph, ON).Total N content in the pooled milk samples collected during the balance period was analyzed using combustion according to the AOAC International (2000) method 990.03 by SGS Crop Science Canada (Guelph, ON, Canada).Samples of corn silage, grass hay, wheat straw, concentrate, and feces were analyzed for DM, ash, N, crude fat, starch (except grass hay), sugars (except corn silage and feces), amylase-treated NDF (aNDF), ADF (except feces), and ADL (except feces) by SGS Crop Science Canada in air-dried material.Forage DM was determined according to methods adapted from Goering and Van Soest (1970), and concentrate DM was determined using AOAC International (2000) method 930.15.Ash was determined after combustion at 600°C for 2 h (AOAC method 942.05,AOAC International, 2000).Nitrogen content in dry samples was determined using combustion according to the AOAC International (2000) method 990.03.Concentration of NH 3 -N in fresh corn silage and fecal samples was measured by ion electrode in the presence of 10 M NaOH.Crude fat content was determined using ANKOM Technology Method 2 (01-30-09; AOCS Official Procedure Am 5-04) with sample extraction using petroleum ether.Starch was determined enzymatically (AOAC method 996.11, AOAC International, 2000).Sugars were analyzed according to Dubois et al. (1956) using a sonicator for sample extraction.The aNDF content was analyzed according to Van Soest et al. (1991) after pre-treatment with α-amylase and sodium sulphite.The ADF and ADL contents were determined by AOAC International (2000) methods 973.18 and 973.18D, respectively.Reported values for nutrient content of the diet were calculated from diet composition and analyzed values obtained for the roughages and concentrate.Reported DVE (intestinal digestible protein), OEB (rumen degradable protein balance), and NE L (VEM (feed unit lactation) system; Van Es, 1978) content for corn silage, grass hay, wheat straw, and concentrate were calculated based on near-infrared spectroscopy analysis of corn silage, grass hay, and wheat straw (Masterlab, St Hyacinthe, QC), and the composition and table values of the ingredients for the concentrate (CVB, 2018).For the TMR, DVE, OEB, and NE L contents were calculated based on the diet composition.Urine samples were analyzed at Wageningen University and Research (Wageningen, the Netherlands).Total N content was analyzed using combustion according to the Dumas principle (ISO, 2008; CN828, LECO Instruments, Mönchengladbach, Germany) and urea concentration was analyzed using the urea liquicolor test (HUMAN, Wiesbaden, Germany) based on measuring light absorbance at 578 nm after a modified Berthelot reaction.Analysis of creatinine, uric acid, and allantoin was performed according to Shingfield and Offer (1999) using reversed-phase HPLC system equipped with a diode-array detector, autosampler, and heated column compartment (Thermo Ultimate 3000, Thermo Fischer Scientific, Waltham, MA).Separation was achieved using a Kinetex EVO C18 reversed-phase column (250 × 4.6 mm, i.d.Five μm; Phenomenex, Torrance, CA) with the use of a precolumn.The specific gravity of a vortexed urine sample was measured at each sampling time point by weighing 1.000 mL of acidified urine with a calibrated pipet.
Rumen fluid samples were prepared and analyzed for VFA (A&L Canada Laboratories; London, ON, Canada) using an ion chromatograph (Dionex ICS-5000; Thermo Fisher Scientific, Waltham, MA) operated at 20°C equipped with an IonPac ICE-AS1 column and ACRS-ICE 500 suppressor (Thermo Fisher Scientific).Heptafluorobutyric acid (0.4 mM) was used as the eluent at a flow rate of 0.8 mL/min, and tetrabutylammonium hydroxide (5 mM) was used as the regenerant.Rumen fluid was analyzed for NH 3 -N concentration by colorimetric assay using salicylate and a nitroprusside as a catalyst (Hach Method 8155; Hach Company, Loveland, CO).Plasma collected from each sampling time point was analyzed in the Department of Animal Biosciences at the University of Guelph (ON, Canada).Plasma AA and 3 methyl-histidine concentrations were determined for each sample time point using ultra-performance liquid chromatography in conjunction with Empower Chromatography Data Software (Waters Corporation, Milford, MA) according to the protocol described by Boogers et al. (2008).Plasma from each sampling time point was analyzed for glucose (kit no.GAGO20; Sigma Chemical Co., Oakville, ON, Canada), BHB (Cant et al., 1993), acetate (kit no.K-ACETRM; Megazyme Neogen, Lansing, MI), nonesterified fatty acids (NEFA; kit no.999-34691; Wako Chemicals GmbH, Neuss, Germany), triacylglycerides (TAG; kit no.TR22421; Thermo Scientific, Nepean, ON, Canada), and urea (kit no.B549-150; Teco Diagnostics, Anaheim, CA).Plasma was pooled over the sampling time points for analysis of insulin (kit no.10-1201-01; Mercodia, Winston Salem, NC) and IGF-1 (kit no.CSB-E08893b; Cusabio, Wuhan, China) by immunoassay.

Calculations and Statistical Analysis
Energy-corrected milk yield was calculated as milk yield (kg) × [(38.30× fat content (g/kg) + 24.20 × protein content (g/kg) + 16.54 × lactose content (g/ kg) + 20.7)/3140) according to Sjaunja et al. (1990).Apparent total-tract digestibility was calculated considering the nutrient inflow from the diet and the infusions.Infused urea was assumed to be 100% digestible and contributed DM, ash, OM, and CP.The DM of the infused urea was assumed to be 100%, and it was assumed that urea contributed 0.3 g ash/kg and 2919 g CP/kg.
Rumen pH recordings at 30 s intervals were used to calculate the mean pH per min over 6 d.These data were recalculated to the min/d that pH was below a certain value, ranging from pH 4.5 to 7.5, with steps of 0.1.For each cow and period, a logistic curve was regressed on the data points using the NLIN procedure of SAS (version 9.4; SAS Institute Inc., Cary, NC) according to the method of AlZahal et al. ( 2007) and the model: 1 440 The parameter B 0 is the slope of the logistic curve at the inflection point and reflects the diurnal variation in ruminal pH, where a higher slope value means rumen pH is more stable throughout the day.The parameter B 1 is the inflection point of the curve which represents the median rumen pH.
One cow was removed from the experiment during period 1 due to a mastitis infection, resulting in missing data for the 325-g/d urea dose from period 1.A reserve cow was maintained on the study diet during period 1 and was used for the remaining experimental periods.Unless otherwise described, all data were analyzed using the GLIMMIX procedure of SAS (version 9.4) where urea dose was considered a fixed effect and period and cow were considered random effects.The Kenward-Roger correction was used to adjust the denominator degrees of freedom.Daily DMI, milk yield, and milk composition was averaged over the 7-d measurement period.Variances in ruminal VFA and NH 3 -N concentrations were evaluated using a model assuming fixed effects of urea dose, time (sampling time points on d 27), urea dose × time, and the random effects of period and cow.Within-animal variation due to equally spaced repeated measures in time was modeled using a first-order autoregressive covariance structure.The same model was applied to variances in ruminal pH where time represented mean hourly pH values over 24 h.The same model was applied to variances in plasma metabolite concentration but the spatial power covariance structure was used to account for the unequal spacing of blood samples collected on d 26.Multiple comparisons between urea dose least squares means for each time point were requested using the SLICE option of the LSMEANS statement.Potential linear, quadratic, and cubic responses to urea dose were evaluated by polynomial regression using the GLIMMIX procedure, where the linear, quadratic, and cubic effects of urea dose were included as covariates and period and cow were random effects.Regression coefficients were estimated using Type I sums of squares.For all analyses, statistical significance was declared where P ≤ 0.05, and tendencies where 0.05 < P ≤ 0.10.We observed no carryover effects between periods, assessed by testing for an effect of the previous treatment in the ANOVA.

DMI, Milk Yield and Composition, and Digestibility
Dry matter intake of the basal diet, milk yield, ECM, and yields of milk fat, protein, and lactose, responded quadratically to urea dose (P ≤ 0.05; Table 2).Urea dose had a cubic effect on milk protein content (P = 0.02).Milk urea N content and feed efficiency increased linearly with increasing urea dose (P ≤ 0.03).Apparent total-tract digestibility of CP increased linearly with increasing urea dose (P < 0.01; Table 3).

Ruminal pH, NH 3 -N, and VFA
Ruminal pH was affected by time (P < 0.01; Supplemental Material S1), where pH decreased for all treatments after feeding, and there was no interaction with urea dose (P = 0.14).The measured mean daily ruminal pH and the inflection point parameter (B 1 ) of the logistic curve increased linearly (P < 0.01; Table 4) and the slope parameter (B 0 ) tended to respond quadratically (P = 0.07) to urea dose.Ruminal NH 3 -N concentration was affected by time, where the concentration decreased over the 12 h period following feeding for all treatments (P < 0.01; Figure 1), and there was no interaction with urea dose (P = 0.51).Mean ruminal NH 3 -N concentration responded quadratically to urea dose (P = 0.02; Table 4).Total VFA concentration was affected by time, where the concentration increased over the 12 h period following feeding for all treatments (P < 0.01; Supplemental Material S2), and there was no interaction with urea dose (P = 0.31).Mean total VFA concentration was not affected by urea dose (P ≥ 0.30; Table 4).The molar proportion of acetate increased linearly, and the molar proportion of propionate decreased linearly with increasing urea dose (P = 0.04).The proportion of acetate to propionate increased linearly with increasing urea dose (P = 0.03).Urea dose tended to have a cubic effect on the molar proportion of isovalerate (P = 0.08).

N Balance and Nitrogenous Metabolite Excretion
Intake of N from the basal diet and total N intake (diet + infusion) tended to respond quadratically to urea dose (P ≤ 0.08; Table 5).The contribution of infused urea-N to total N intake increased linearly with increasing urea dose (P < 0.01).Excretion of N in feces responded quadratically (P = 0.02), excretion of N in urine and incremental N excretion increased linearly (P < 0.01) with increasing urea dose, and excretion of N in milk tended to respond quadratically to urea dose (P = 0.09).Urea dose tended to have a cubic effect on N retention (P = 0.09).The proportion of N intake excreted in feces decreased linearly (P < 0.01) and the proportion of N intake excreted in urine increased linearly (P < 0.01) in response to increasing urea dose.The proportion of N intake excreted in milk decreased linearly with increasing urea dose (P = 0.01).
Urinary urea excretion increased linearly with increasing urea dose (P < 0.01).Excretion of allantoin responded quadratically (P = 0.04) to urea dose.Excretion of uric acid and total purine derivatives (allantoin + uric acid) responded cubically (P ≤ 0.05) to urea dose.Excretion of total purine derivatives was not affected by urea dose when expressed relative to kg of apparently digested OM (P ≥ 0.84).Urea dose had a cubic effect on microbial N flow (P = 0.05).The efficiency of microbial protein synthesis was not affected by urea dose (P ≥ 0.37).

Plasma Constituents
No interactions between urea dose and time were detected for any plasma constituents (P > 0.10).Arterial plasma concentrations of BHB, TAG, and acetate were affected by time (P ≤ 0.04), and concentration of NEFA tended to be affected by time (P = 0.07), where Data are least squares means calculated from feed and feces sampled during the 72-h balance period and were calculated considering the total nutrient inflow from the diet + infusions.
their concentrations increased over the day after feeding (data not shown).Mean plasma TAG concentration tended to respond quadratically to urea dose (P = 0.07; Table 6), and urea concentration increased linearly with increasing urea dose (P = 0.01).Arterial plasma concentration of total EAA and the individual concentrations of several AA were affected (Ile, Leu, Met, Phe, Thr, Trp, Val, Asn, Gly; P ≤ 0.05) or tended to be affected (Arg, His, Lys, Gln, Ser; P ≤ 0.10) by time over the day of sampling, where their concentration decreased until 3.5 h after feeding and then increased to similar concentrations as observed at feeding (data not shown).Mean plasma concentration of total EAA decreased linearly with increasing urea dose (P = 0.04).Individually, concentrations of Phe, Thr, Gly, and Tyr decreased linearly with increasing urea dose (P ≤ 0.03).The concentration of Asn tended to decrease linearly (P = 0.07) and the concentration of Glu+Cit tended to increase linearly (P = 0.06) with increasing urea dose.Plasma concentration of His and Met responded quadratically (P = 0.05) and concentrations of Trp and Orn tended to respond quadratically (P = 0.07) to urea dose.The concentration of Ile and Ala responded cubically to urea dose (P ≤ 0.03).

DISCUSSION
Feeding urea at inclusion levels >1% of dietary DM or >20% of total CP intake in dairy cattle diets often does not benefit nutrient intake, milk production, or N efficiency because it results in ruminal ammonia concentrations that exceed microbial requirements (Satter  and Slyter, 1974;Satter and Roffler, 1975;Polan et al., 1976), or DMI is depressed (Wilson et al., 1975;Polan et al., 1976;Brito and Broderick, 2007).We hypothesized that supplementing urea post-ruminally would 1) mitigate the negative effects of urea on DMI, and 2) stimulate recycling of urea to the rumen to support microbial protein synthesis.The linear increase in ATTD of CP in response to increasing urea dose disappeared when ATTD was calculated considering intake of the basal diet only (data not shown), indicating that the highly digestible nature of urea that is seen in the rumen (Veen and Bakker, 1977;Teller and Godeau, 1986) was maintained when urea was delivered postruminally.If the CP digestibility of 57% observed with 0 g urea/d represents CP digestibility of the basal diet, calculated ATTD of the infused urea-N was 106, 97, and 104% for 163, 325, and 488 g/d, respectively.These estimates support our assumption that the infused urea was 100% digestible in the small intestine and suggest that it did not contribute appreciably to microbial protein synthesis in the hindgut.Characterization of bac-terial urease activity in the post-ruminal digestive tract is scarce, but it has been detected in the digesta and mucosa of the post-ruminal GIT (abomasum through colon) of sheep (Michnová et al., 1979;Whitelaw et al., 1991;Marini et al., 2004), and in the digesta of several monogastric species (Kornberg and Davies, 1955;Patra and Aschenbach, 2018).If 100% of infused urea was hydrolyzed to ammonia in the small intestine, 5.4, 10.8, and 16.3 mol/d of ammonia from infused urea could have been absorbed from the post-ruminal GIT.This would be reasonable based on estimates of net portal ammonia absorption in lactating dairy cattle (Bach et al., 2000;Lapierre and Lobley, 2001;Røjen and Kristensen, 2012) and the fact that urease activity in the GIT appears not to limit urea hydrolysis even under physiologically high urea loads (Norton et al., 1982;Whitelaw et al., 1991;Marini et al., 2004).Urea not hydrolyzed by post-ruminal urease could have been absorbed through facilitative urea transporters or aquaglyceroporins which have been detected in the 2 N retention = N intake (diet + infusion) -N feces -N urine -N milk. 3 Incremental N excretion = Total N excretion in urine + feces corrected for N excretion observed on the negative control, i.e., 0 g urea/d.

4
Sum of allantoin excretion and uric acid excretion.

Stimulation of DMI, microbial N flow, and milk production
Dry matter intake responded quadratically to urea dose and increased 2.3 kg/d over the negative control (0 g urea infused/d) when 163 g of urea was infused per d.In line with this increase in DMI, infusion of 163 g urea/d increased the yield of ECM and milk protein by 3.5 kg/d and 137 g/d, respectively, and resulted in a numerically similar feed efficiency compared with the negative control.Quadratic regression based on this dose response (Supplemental Material S3) estimates that a maximum DMI of 23.4 (95% CI: 19.1, 27.7) kg/d would be observed at a post-ruminal urea dose of 179 g/d, which would correspond to a dietary urea inclu-sion rate of 0.8% of DMI.Indeed, urea inclusion up to 0.8% of DMI is within the range of what is considered a practical level for dairy diets (NRC, 2001), where generally no negative effects on DMI would be expected (Poos et al., 1979;Kertz, 2010).
Infusing 163 g urea/d increased the CP content of the total diet (basal diet + infusion) from 10.9 to 12.9% (DM basis).Increasing CP intake through post-ruminal protein or NPN supplementation has been shown to increase DMI in cattle when the basal diet consists of low-quality forage (e.g., low digestibility, < 10% CP; Bandyk et al., 2001;Wickersham et al., 2009), or when MP balance is negative (Martineau et al., 2016).Rumen-available N, either from RDP or recycled urea, can increase DMI due to the physical effect on rumen capacity resulting from improved microbial function and fiber digestion, particularly at low dietary CP contents (Köster et al., 1996;Allen, 2000;Wickersham et

2009
).An increased energy requirement for milk protein synthesis upon increased MP supply arising from microbial protein or RUP sources may also stimulate feed intake (Martineau et al., 2016).When the lowest dose of post-ruminal urea was infused in the current experiment, we expect that DMI was stimulated through improved fiber digestibility, possibly resulting from relatively greater urea recycling to the rumen compared with the negative control.Indeed, digestibility of OM, and aNDF increased numerically by 1.7 and 2.3 units, respectively, in response to infusion of 163 g urea/d compared with 0 g/d.Further, total ruminal VFA concentration and microbial N flow was numerically highest at 163 g/d of post-ruminal urea.Taken together, these observations are in line with previous studies with nonlactating ruminants that have observed improved digestion, increased feed intake, or greater microbial protein synthesis when supplementing low CP diets (<10% CP) with post-ruminal protein (Bandyk et al., 2001;Wickersham et al., 2009) or urea (Egan and Moir, 1965;Oliveira et al., 2020).

DMI depression
In contrast to our hypothesis, increasing urea inclusion above current practical recommendations via post-ruminal delivery did not mitigate DMI depression.Post-ruminal urea doses of 325 and 488 g/d decreased DMI by 2.3 and 4.9 kg/d, respectively, compared with the 163-g/d urea dose.Urea doses of 325 and 488 g/d were equivalent to 1.7 and 2.8% of observed DMI, respectively.These correspond to urea inclusions levels that would be expected to decrease DMI in lactating dairy cattle if fed into the rumen (Wilson et al., 1975;Polan et al., 1976;Brito and Broderick, 2007).Previous work has attributed reduced intake of urea-containing diets to poor palatability of urea (i.e., taste; Huber and Cook, 1972) or consequences of increased rumen ammonia concentrations (e.g., ammonia trapping in epithelial cells, reduced rumen motility; Conrad et al., 1977;Davidovich et al., 1977).However, the hypophagic response observed in the current study where urea was infused post-ruminally suggests that reduced DMI cannot be due simply to poor urea palatability.Indeed, Wilson et al. (1975) observed that urea supplementation at levels above 2% of dietary DM decreased feed intake of Holstein cows whether it was mixed with the diet, pulse-dosed into the rumen, or continuously infused into the rumen, further suggesting that urea palatability is not the only factor responsible for DMI depression when urea-containing diets are fed.Taken together, these results suggest that signals from the post-ruminal digestive tract, post-absorptive metabo-lism, or both, can play a role in hypophagia in response to urea supplementation.
Assuming the infused urea was readily absorbed as ammonia, the amount of ammonia absorbed into portal circulation would have increased with increasing urea dose.Depending on the intensity of hepatic ammonia load (i.e., magnitude of portal concentration and duration of elevated concentrations), ammonia detoxification via urea synthesis can require α-amino N from AA, which leaves the AA carbon available for gluconeogenesis or oxidative catabolism (Milano and Lobley, 2001;Brosnan and Brosnan, 2009).Based on in vitro observations with isolated caprine or ovine hepatocytes, increasing the concentration of ammonia in the incubation buffer decreased the use of propionate for gluconeogenesis and increased propionate oxidation (Aiello and Armentano, 1987;Demigné et al., 1991).In vivo, Oba and Allen (2003) reported a greater decrease in DMI when ammonium propionate was infused compared with sodium propionate, supporting an exacerbating effect of ammonia on hypophagic effects of propionate oxidation in the liver.Increased oxidation of AA-carbon and propionate as a consequence of high hepatic ammonia load is consistent with hypophagia related to elevated hepatic ATP concentrations translating a satiety signal via neural pathways to the center of the brain influencing feeding behavior (Allen et al., 2009).Equivalent N supply from abomasally infused casein or AA mixtures, which do not impose an immediate ammonia detoxification load on hepatocytes upon first-pass liver uptake, do not typically reduce DMI (Dhiman et al., 1993;Relling and Reynolds, 2008;Nichols et al., 2016).This further implicates hepatic ammonia metabolism in the feed intake response to abomasal urea infusion.Therefore, we speculate that the decrease in DMI at the highest levels of post-ruminal urea infusion was related to increased post-absorptive ammonia load in the liver resulting in oxidation of AAcarbon and propionate, and the associated signals of satiety in the brain.
Compared with the negative control, DMI was unchanged or decreased 13% with the 325 and 488 g/d urea doses, respectively, but ECM increased 8% with infusion of 325 g urea/d and decreased 7% with infusion of 488 g urea/d.The higher feed efficiency with 325 and 488 g urea/d compared with 0 and 163 g urea/d reflects this relative maintenance of milk production despite the reduced DMI as the infused urea dose increased.This relative maintenance of milk production despite the reduced DMI at the 2 highest levels of urea infusion agrees with the observed OM digestibility and total VFA concentrations at these urea doses.Digestibility of OM was numerically similar between 0 and 325 g urea infused/d, and was numerically highest with 488 g/d.Further, concentration of total VFA with 325 and 488 g/d remained numerically higher than the negative control.The maintained or numerically increased OM digestibility with 325 and 488 g urea/d, particularly at the highest dose, is likely a function of the lower DMI encouraging longer rumen retention time and increased fermentation (Colucci et al., 1982;Robinson et al., 1985), and was also observed by Brito and Broderick (2007) when urea inclusion at 1.9% of DMI resulted in a 2.5-kg/d decrease in DMI compared with diets supplemented with plant protein sources.The absence of significant response in OM or aNDF digestibility in response to urea infusion at any dose in this experiment could be due to a deficiency of branched-chain VFA to support fiber-digesting bacteria under the condition of relatively low dietary RDP (Kang-Meznarich and Broderick, 1980;Milton et al., 1997).

Ruminal ammonia-N
We intentionally formulated the basal diet to be rich in fermentable carbohydrates, low in CP content, and deficient in RDP relative to estimated requirements (CVB, 2018) based on the hypothesis that postruminally infused urea would be recycled to the rumen under these conditions of ruminal N deficit.Ruminal ammonia-N concentrations exhibited post-prandial peaks after 1 h for the negative control (5.36 mg/dL; Figure 3), and after 1.5 h for the urea infusion treatments (range of 8.27 to 9.17 mg/dL).Increased transepithelial urea-N transport from the blood through the rumen wall is facilitated by increased concentrations of short-chain fatty acids and CO 2 and depletion of ruminal ammonia that can be expected upon the intake of rumen-fermentable energy (Rémond et al., 1996;Abdoun et al., 2010;Lu et al., 2014).That peaks in rumen ammonia-N concentration were higher in response to urea infusion treatments compared with the negative control suggests that post-ruminal urea infusion may have allowed greater transfer of endogenous urea into the rumen than what was achieved by the basal diet alone, which supports our hypothesis.This speculation is supported by Wickersham et al. (2009) who supplemented graded doses of casein to steers via continuous abomasal infusion on top of a 4.7%-CP hay and reported that with increasing doses of post-ruminal casein, rumen ammonia concentrations and urea recycling increased linearly, microbial N synthesized from recycled urea increased quadratically, and duodenal microbial N flow and N retention relative to N intake increased linearly.Mean ruminal ammonia-N concentration in response to the basal diet (i.e., 2.16 mg/ dL with 0 g urea infused/d) was only slightly higher than the recommended minimum required for micro-bial growth (1.96 mg/dL; Satter and Slyter, 1974).At the 163 g/d urea dose, the increase in mean ruminal ammonia-N concentration to approximately double this minimum threshold for optimum microbial function agrees with the numerical increases in ATTD of DM, OM, and aNDF, DMI, microbial N flow, and ECM over the negative control.
Quadratic regression based on this dose response estimates that a maximum ruminal ammonia-N concentration of 5.41 (95% CI: 1.17, 11.9) mg/dL would be achieved at a urea dose of 395 g/d (Supplemental Material S3).This concentration is similar to that identified in vitro by Satter and Slyter (1974) where ruminal ammonia began to accumulate and no additional microbial growth was observed (5.04 mg/dL).Similarly, Weston and Hogan (1967) infused urea into the abomasum of sheep and observed a limit to the increase in ruminal ammonia-N concentration with increasing urea-N infusion rate.In the current study, ruminal ammonia-N concentration followed a quadratic response reaching a maximum, but plasma urea concentration increased linearly as urea infusion dose increased.Sunny et al. (2007) determined that blood urea concentration was correlated with return of urea to the GIT in sheep, but that the capture and use of recycled urea-N in the GIT is more rate limiting to N use efficiency than the rate of urea-N recycling.There is likely a discrepancy between ammonia flux across the rumen epithelium and the measured luminal ammonia concentration.Microbial colonization of the rumen epithelium leads to accumulation of ammonia from blood urea along the epithelium (Egan, 1980;Egan et al., 1986).This build-up along the rumen wall can result in reabsorption of ammonia and may inhibit urea diffusion or transport from blood into the rumen (Egan et al., 1986;Kristensen et al., 2010).This epithelial ammonia pool from hydrolyzed urea equilibrates poorly with the luminal ammonia pool (Egan, 1980;Egan et al., 1986), which limits the interpretation of ruminal fluid ammonia concentrations as representative of true ammonia absorption across the rumen wall.Indeed, Kristensen et al. (2010) determined that a large fraction of ammonia released into the ruminal vein arises from the epithelial ammonia pool, not from rumen fluid, and that the venous-arterial difference of ammonia across the rumen was poorly correlated with luminal ammonia concentration.As plasma urea concentration increased alongside post-ruminal infusion dose in the current experiment, it is possible that ammonia accumulation along the ruminal epithelium inhibited the transfer of blood urea into the rumen, resulting in ammonia concentrations measured in rumen luminal fluid that likely differed from that at the ruminal epithelium.Further, the reduced DMI at higher urea infusion doses could have resulted in a ruminal environment that was less favorable to urea entry via the rumen wall (i.e., reduced fermentable OM intake, increased ruminal pH; Lu et al., 2014).In contrast to urea influx from blood, urea derived from saliva is hydrolyzed in the ruminal digesta and would contribute to measured ammonia concentration in ruminal fluid.However, the diet fed in this study was relatively fermentable and high in concentrate relative to forage, so the contribution of urea from saliva to ruminal ammonia concentration was likely lower compared with epithelial transfer (Huntington, 1989).
It should be noted that suggestions of effects on urea recycling to the rumen are speculative, as urea recycling itself was not measured in the current study and rumen ammonia concentrations alone do not describe the flux of ammonia-N or urea-N across the rumen wall.Thus, we cannot clearly determine if changes in ruminal ammonia-N concentration were truly in response to changes in urea transport or if greater amounts of recycled urea were used to support microbial protein synthesis in lieu of hydrolysis of ruminal degradable protein.Further, changes in ruminal ammonia-N concentration could have also been in response to changes in fluid dilution rates which were not measured in the current study.

N Balance and Plasma AA
Total N intake tended to respond quadratically, where it increased with urea infusion at all doses relative to the negative control.The linear decrease and increase in N excretion in feces and urine, respectively, as a proportion of N intake is in line with the increased N excretion via urine typically observed when digestible N supply increases (Sunny et al., 2007;Wickersham et al., 2009;Chibisa and Mutsvangwa, 2013).The shift of N excretion from feces toward urine also agrees with the linear increase in ATTD of CP and the complete digestion and absorption of urea along the GIT.The contribution of urea-N to urinary N excretion increased from 33% on the negative control (23 g/d) to 79% at the highest urea infusion dose (196 g/d), which reflects the increased contribution of urea-N to total N intake.Increasing N intake from post-ruminal urea had little effect on the non-urea component of urinary N, which averaged (±SD) 48 ± 3.6 g/d across treatments (data not shown).Despite evidence that post-ruminal urea infusion increased urea recycling to the rumen, the efficiency of utilization of infused urea-N decreased as infusion dose increased.Marginal use efficiency of urea-N for milk N decreased from 24% on 163 g urea/d to 9% on 325 g urea/d and to 8% in response to 488 g urea/d, and milk N efficiency decreased linearly with increasing urea infusion dose.
Post-ruminal urea infusion increased N intake but could not contribute directly to absorbable AA supply upon digestion in the small intestine.The higher DMI and microbial N flow with infusion of 163 g urea/d may have resulted in a slight increased AA absorption.However, the arterial plasma concentration of NEAA was not affected by urea dose, and total EAA concentration decreased linearly with increasing urea dose.That arterial plasma concentrations of total AA and NEAA were not affected by increasing dose of post-ruminal urea infusion despite reduced DMI suggests that mobilization of labile AA stores could have contributed to circulating AA supply as the infused urea dose increased.Further, rapid and substantial transfer of N from ammonia to NEAA can occur within the liver (Geissler et al., 1992;Luo et al., 1995).Synthesis of NEAA from the elevated post-absorptive ammonia levels as urea infusion dose increased could have spared EAA from catabolism for NEAA synthesis, thus supporting the maintained circulating concentration of total AA despite the decrease in DMI.As urea dose increased, the decrease in plasma EAA concentration likely reflects increased efficiency of extraction of EAA by the mammary gland in response to changes in absorbed AA supply as DMI changed quadratically in response to increasing urea dose.Indeed, as total digestible AA supply decreases, mammary gland affinity for EAA uptake remains greater relative to that of NEAA (Cantalapiedra-Hijar et al., 2015;Omphalius et al., 2019).

CONCLUSIONS
The regression analysis based on this does response study reveals that when supplemented on top of a low-CP diet, 179 g/d of post-ruminal urea would result in a maximum DMI of 23.4 kg/d.At this level of DMI, this urea dose corresponds to a dietary urea inclusion level of 0.8% of DMI which is within the current recommendations for urea inclusion in dairy cattle diets.These results indicate that post-ruminal delivery of urea does not mitigate DMI depression as urea dose increases above the currently accepted threshold for urea inclusion in dairy diets.However, this study demonstrated that post-ruminally supplied urea could affect the rumen environment, possibly via urea recycling.Regression analysis revealed that ruminal ammonia-N concentration displayed a quadratic response that peaked at 395 g urea/d.Factors such as ammonia concentration at the rumen epithelium or decreased DMI may have inhibited urea flux at post-ruminal urea doses beyond this level.The efficiency of utilization of infused urea-N and milk N efficiency decreased with increasing urea dose, and the proportion of N intake excreted in urine increased linearly as urea infusion Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA
Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA dose increased.How marginal utilization efficiency of post-ruminal urea compares with that of post-ruminal true protein sources or RDP in lactating dairy cattle warrants further research.

Table 1 .
Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA Ingredient composition of the diet and analyzed and calculated chemical composition of ingredients (corn silage, grass hay, wheat straw, and concentrate) and complete TMR (g/kg of DM, unless otherwise noted)

Table 2 .
Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA Dry matter intake, milk production, and milk composition of lactating dairy cows receiving continuous abomasal infusions of urea 2 Diet only.3 ECM (kg/d)/DMI (kg/d).

Table 4 .
Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA Ruminal pH, ammonia-N concentration, total VFA concentration, and VFA molar proportions of lactating dairy cows receiving continuous abomasal infusions of urea 1 Standard error of the difference.n = 4 for all treatments except 325 g/d where n = 3. 2 Ruminal pH measured continuously during the final 6 d of each period.

Table 5 .
Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA Nitrogen balance and urinary excretion of nitrogenous metabolites of lactating dairy cows receiving continuous abomasal infusions of urea

Table 6 .
al., Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA Arterial plasma hormone and metabolite concentrations of lactating dairy cows receiving continuous abomasal infusions of urea 2Peaks for Glu and Cit could not be separated in the integration.
Nichols et al.: DOSE RESPONSE TO POST-RUMINAL UREA