nutri-Increasing the prepartum dose of rumen-protected choline: Effects on milk production and metabolism in high-producing Holstein dairy cows

Peripartum rumen-protected choline (RPC) supplementation is beneficial for cow health and production, yet the optimal dose is unknown. In vivo and in vitro supplementation of choline modulates hepatic lipid, glucose, and methyl donor metabolism. The objective of this experiment was to determine the effects of increasing the dose of prepartum RPC supplementation on milk production and blood biomarkers. Pregnant multiparous Holstein cows (n = 116) were randomly assigned to one of 4 prepartum choline treatments that were fed from −21 d relative to calving (DRTC) until calving. From calving until +21 DRTC, cows were fed diets targeting 0 g/d choline ion (control, CTL) or the recommended dose (15 g/d choline ion; RD) of the same RPC product that they were fed prepartum. The resulting treatments targeted: (1) 0 g/d pre-and postpartum [0.0 ± 0.000 choline ion, percent of dry matter (%DM); CTL]; (2) 15 g/d pre-and postpartum of choline ion from an established product (prepartum: 0.10 ± 0.004 choline ion, %DM; postpartum: 0.05 ± 0.004 choline ion, %DM; ReaShure, Balchem Corp.; RPC1 RD ‣ RD ); (3) 15 g/d pre-and postpartum of choline ion from a concentrated RPC prototype (prepartum: 0.09 ± 0.004 choline ion, %DM; postpartum: 0.05 ± 0.003 choline ion, %DM; RPC2, Balchem Corp.; RPC2 RD ‣ RD ); or (4) 22 g/d prepartum and 15 g/d postpartum from RPC2 [prepartum: 0.13 ± 0.005 choline ion, %DM; postpartum: 0.05 ± 0.003 choline ion, %DM; high prepartum dose (HD), RPC2 HD ‣ RD ]. Treatments were mixed into a total mixed ration, and cows had ad libitum access via a roughage intake control system (Hokofarm Group). From calving to +21 DRTC, all cows were fed a common base diet and treatments were mixed into the total mixed ration (supplementation period, SP). Thereafter, all cows were fed a common diet (0 g/d choline ion) until +100 DRTC (postsupplementation period, postSP). Milk yield was recorded daily and composition analyzed weekly. Blood samples were obtained via tail vessel upon enrollment, approximately every other day from −7 to +21 DRTC, and at +56 and +100 DRTC. Feeding any RPC treatment reduced prepartum dry matter intake compared with CTL. During the SP, no evidence for a treatment effect on energy-corrected milk (ECM) yield was found, but during the postSP, RPC1 RD ‣ RD and RPC2 RD ‣ RD treatments tended to increase ECM, protein, and fat yields. During the postSP, the RPC1 RD ‣ RD and RPC2 RD ‣ RD treatments tended to increase, and RPC2 HD ‣ RD increased, the de novo proportion of total milk fatty acids. During the early lactation SP, RPC2 HD ‣ RD tended to increase plasma fatty acids and β-hydroxybutyrate concentrations, and RPC1 RD ‣ RD and RPC2 RD ‣ RD reduced blood urea nitrogen concentrations compared with CTL. The RPC2 HD ‣ RD treatment reduced early lactation serum lipopolysaccharide binding protein compared with CTL. Overall, peripartum RPC supplementation at the recommended dose tended to increase ECM yield postSP, but no evidence was seen of an additional benefit on milk production with an increased prepartum dose of choline ion. The effects of RPC on metabolic and inflammatory biomarkers support the potential for RPC supplementation to affect transition cow metabolism and health and may support the production gains observed.


INTRODUCTION
Feeding rumen-protected choline (RPC) during the peripartum period improves milk production and may reduce the risk of some metabolic health disorders (Zenobi et al., 2018a;Arshad et al., 2020). Peripartum RPC supplementation not only improves milk yield but also increases milk fat and protein yields, which results in greater yields of ECM (Arshad et al., 2020). Improved postpartum DMI with RPC supplementation partially explains the greater milk energy output, but RPC may also improve pre-or postabsorptive nutri-ent use efficiency, as feed efficiency was also improved across multiple studies in a recent meta-analysis (Arshad et al., 2020).
It has been suggested in a previous review that the effects of RPC on milk production could be modulated through altered hepatic energy and lipid metabolism (McFadden et al., 2020). Although RPC supplementation has reduced storage of hepatic triglyceride (TG) in nonlactating, feed-restricted cows (Cooke et al., 2007;Zenobi et al., 2018b), effects of peripartum RPC supplementation on hepatic TG are not consistent across all studies (Arshad et al., 2020). Supplementation of choline chloride in vitro (Chandler and White, 2017) or RPC in vivo (Sun et al., 2016) supports verylow-density lipoprotein production and TG export from hepatocytes, which could contribute to milk fat production. Choline may also increase hepatic gluconeogenesis, as greater glycogen concentrations have been observed with in vitro choline chloride supplementation (Chandler and White, 2019). Furthermore, greater postpartum blood glucose concentrations (Xu et al., 2006;Sun et al., 2016;Zhou et al., 2016b;Arshad et al., 2020) and hepatic glycogen concentrations (Piepenbrink and Overton, 2003;Zenobi et al., 2018b) have been observed with RPC feeding in vivo. Research has also demonstrated that RPC alters the immune response to LPS ex vivo (Garcia et al., 2018) and that RPC may mitigate inflammation and improve immune function during the transition to lactation period . These findings have raised interest in the potential role of choline as a nutritional immunomodulator, which may also contribute to production responses, nutrient partitioning, and feed efficiency.
A meta-analysis of RPC supplementation studies determined that increasing the supplemented dose of RPC linearly increased milk production in treatments that ranged up to 25.2 g/d of choline ion; however, it is important to note that only 4 of 21 experiments fed greater than 15 g/d of choline ion (Arshad et al., 2020). Based on previous literature, we hypothesized that feeding increased doses of RPC prepartum would result in greater milk and ECM yield postpartum. Nearly all previous RPC research has involved top-dressing the RPC treatment (Arshad et al., 2020;Holdorf and White, 2021), which results in the same absolute amount of RPC consumed by each cow. On-farm, RPC is mixed evenly throughout the diet, resulting in choline ion intake that varies with DMI. To directly address our hypothesis and to more fully recapitulate on-farm feeding practices, the current study was designed to determine the effects of RPC supplementation when fed in a TMR. The objective of this experiment was to determine the effects of increasing the dose of pre-partum RPC supplementation on milk production and blood biomarkers.

Animal Use, Treatments, and Handling
All protocols for animal use and handling were approved by the University of Wisconsin-Madison College of Agricultural and Life Science Animal Care and Use Committee. This experiment was conducted at the University of Wisconsin-Madison Emmons Blaine Arlington Dairy Research Center in Arlington, Wisconsin. Multiparous,pregnant,nonlactating Holstein cows (n = 116) were enrolled weekly at −21 d before expected calving from December 2020 through March 2021. Cows remained on the study until +100 d relative to calving (DRTC) with the study ending in July 2021. Cows were milked twice daily (0500 and 1700 h) via parlor and fed a TMR once daily (lactating, 0700 h; prepartum, 1100 h).
Pregnant multiparous Holstein cows (n = 116) were randomly assigned to 1 of 4 prepartum choline treatments, which were fed from −21 DRTC until calving. From calving until +21 DRTC, cows were fed diets targeting 0 g/d of choline ion (control, CTL) or the recommended dose (15 g/d of choline ion; RD) of the same RPC product they were fed prepartum. The resulting treatments were: (1) 0 g/d pre-and postpartum (0.0 ± 0.000 choline ion, % DM; CTL); (2) targeting 15 g/d (prepartum: 0.10 ± 0.004 choline ion, % DM; postpartum: 0.05 ± 0.004 choline ion, % DM; RD) pre-and postpartum of choline ion from an established product (ReaShure, Balchem Corp.; 26% choline ion content; RPC1 RD•RD ); (3) targeting 15 g/d pre-and postpartum of choline ion from a concentrated RPC prototype (prepartum: 0.09 ± 0.004 choline ion, % DM; postpartum: 0.05 ± 0.003 choline ion, % DM; RPC2, Balchem Corp.; 48% choline ion content; RPC2 RD•RD ); or (4) targeting 22 g/d prepartum and 15 g/d postpartum from RPC2 [prepartum: 0.13 ± 0.005 choline ion, % DM; postpartum: 0.05 ± 0.003 choline ion, % DM; high prepartum dose (HD), RPC2 HD•RD ]. Ruminal protection of RPC1 and RPC2 was 73% and 63%, respectively, as determined by a third-party commercial laboratory (Cumberland Valley Analytical Services, Waynesboro, PA) using an in situ procedure over 8 h of incubation. For pre-and postpartum diets, a common base TMR was mixed, and treatment concentrates were added to the mixer wagon to create the individual treatment diets (Table 1). Treatment concentrates were mixed at the Arlington Agricultural Research Station feed mill (Arlington, WI) where the appropriate amount of RPC additive for each treatment mix was combined with a soybean meal carrier and appropriate amounts of soy hulls and soybean oil. This approach balanced noncholine nutrients provided by the RPC additives and increased the volume of the treatment to aid in proper distribution within the TMR and thus resulted in the same amount of treatment concentrate added to each treatment diet. A preexperimental audit of the TMR mixing and feeding process, along with periodical observation throughout the experiment, were conducted to verify feed mixing procedures were followed, as proper distribution of treatment throughout the TMR was imperative. Both the pre-and postpartum diets included a rumen-protected source of methionine to allow for testing the direct effects of RPC supplementation in a methionine-sufficient diet [prepartum: 2.23 ± 0.01 methionine, % of MP (mean ± SD); postpartum: 2.04 ± 0.01 methionine, % of MP (mean ± SD); NRC, 2001]. The composition of each treatment concentrate is described in Table 1.
Prepartum cows were housed together on a bedded pack and offered ad libitum access to assigned treatment diets via a roughage intake control system (RIC, Hokofarm Group; 4 feeders/treatment), which allowed for individual cow intakes to be quantified. Upon enrollment, cows entered the prepartum pen and were trained to identify feeders with their assigned treatment diet and were monitored daily. The actual number of days in the prepartum pen was 20.9 ± 4.2 (mean ± SD). Cows were housed in a postpartum pen for 3 d after calving, while being fed assigned treatment diets, so milk could be discarded until it was salable. Afterward, cows were grouped by prepartum treatment into postpartum, sandbedded freestall pens with a maximum of 8 cows/pen and were fed a common base lactating diet with no RPC (CTL) or with the recommended dose from the RPC product consistent with their prepartum treatment. The amount of feed offered and refused from each pen was weighed and recorded daily and divided by the number of cows in the respective pen to determine pen-level DMI. A schematic of pre-and postpartum treatments and resulting treatment groups is provided in Table 2. This experiment targeted 21 d on postpartum treatment diets, but cows were moved once per week from the postpartum treatment pens, resulting in a postpartum supplementation period (SP) of 23.3 ± 2.6 d (mean ± SD). Cows were moved from the postpartum treatment pens to pens of 16 cows, where cows from all treatments were commingled and fed a common lactating diet (0 g/d choline ion) until +100 DRTC (postSP).
Health monitoring and medical treatments were administered according to the farm's standard operating procedure by veterinarians and farm staff. In addition, intensive monitoring for subclinical hyperketonemia detection and treatment was performed by quantifying blood BHB via a cow-side BHBCheck meter (Porta-Check) approximately every other day from +3 to +18 DRTC. The treatment for subclinical hyperketonemia (BHB ≥1.2 mM) was 3 to 5 consecutive days of a propylene glycol oral drench (300 mL), and if clinical hyperketonemia was detected (BHB ≥3.0 mM), a 50% dextrose solution (250 mL) was administered intravenously on the day of detection, in addition to the propylene glycol treatment.

Sample Collection and Analysis
Prepartum as-fed daily intake was recorded electronically for each cow using RIC feeders and then multiplied by the weekly DM content of the diet to determine daily DMI. The meal criterion, defined as 20.98 min in this experiment, was calculated as previously described (Tolkamp et al., 1998;DeVries et al., 2003;Brown et al., 2022) and used to segment bin visits into distinct meals. Electronic recording of bin visit duration and feed consumption allowed for quantification of meal size and length as measures of feeding behavior. Prepartum DMI and feeding behavior were truncated at −18 DRTC to include all cows in the analysis based on actual days in the prepartum pen. Limited stealing of unassigned treatment diets occurred (<10% of total DMI), but any intake of unassigned treatments was detected and recorded for accurate DM and RPC intake. If more than 10% of a cow's DMI was from an unassigned treatment, that cow was removed from the experiment.
Samples of individual TMR ingredients were collected weekly and dried by forced-air oven at 55°C for 48 h to determine DM content for prepartum DMI calculations. Dried samples were ground to pass a 1-mm screen (Wiley Mill, Arthur H. Thomas Company), and composites were sent to a commercial laboratory (Dairyland Labs, Arcadia, WI) for subsequent analysis (Table 1). Feed samples were composited by month, except when a new bunker of forage was introduced mid-month, in which case a composited sample from each bunker was sent for analysis. The analysis included CP (AOAC International, 2012; method 990.03), ADF (AOAC International, 1996;method 973.18), NDF (AOAC International, 2005;method 2002.04), lignin (AOAC International, 1996;method 973.18), ether extract (AOAC International, 2012; method 920.39), ash (AOAC International, 2012; method 942.05), water-soluble carbohydrates (Deriaz, 1961), and starch (AOAC International, 2019; method 2014.10).
The yield of colostrum was recorded for each cow, and a sample was collected for quantification of Brix percent via a handheld digital Brix refractometer (MIS-CO). Milk yield was recorded daily at both milkings and summed to represent daily milk yield. Individual cow milk samples from 4 consecutive milkings each week were collected and preserved with 2-bromo-2-nitropropane-1,3-diol (Advanced Instruments Inc.) for analysis at a commercial laboratory (AgSource, Menominee, WI) for milk composition of fat, protein, lactose, and MUN by Fourier transform infrared spectrometry using the MilkoScan FT+ (FOSS Analytical) and for analysis of milk SCC.
Variables found to have nonnormally distributed residuals were transformed for analysis with either a square root, log 10 , log 10 (x + 1), natural log, natural log(x + 1), x 2 , or cube root transformation. Residuals for analysis of SCC data were nonnormal; thus, for final analysis, the data were transformed to SCS with the formula SCS = log 2 (SCC/100) + 3 (Ali and Shook, 1980). For models analyzing the binary outcome of individual diseases (i.e., hyperketonemia, displaced abomasum, and clinical mastitis), the GLIMMIX procedure in SAS 9.4 (SAS Institute Inc.) was used with the binary distribution option. All other response variables were analyzed using the MIXED procedure in SAS. Covariates (e.g., PTA, parity group, and baseline values from enrollment day) were offered to mixed models to account for variation that could not be randomized within treatment and included in the model if P ≤ 0.10. The linear predictor for all models included the fixed effect of treatment. Models with repeated measures also included fixed effects of time and the interaction of treatment and time. Random effects of cow nested within treatment and month of enrollment were included in all models, and the random effect of postSP pen was added for models analyzing postSP variables. The covariance of repeated measures within each cow was modeled using the repeated measures statement and first-order autoregressive structure type.
Studentized residuals were analyzed for normality using the univariate procedure Shapiro-Wilk and Kolmogorov-Smirnov tests in SAS and by visual interrogation of a Q-Q plot to ensure the assumptions of mixed models were reasonably met. Transformation of the response variable was employed when nonnormality of residuals was detected. Homogeneity of residual vari-ances was analyzed when data transformation could not meet model assumptions using fitted values versus studentized residuals plot and plots of different model effects versus studentized residuals. If heterogeneity was suspected, covariance structures were modeled separately by model effects and remained in the final model when assumptions were met. If multiple models allowed for model assumptions to be met, the model with the lowest Bayesian information criterion was selected. When P > 0.05 for the Kolmogorov-Smirnov test but P ≤ 0.05 for the Shapiro-Wilk, visually determined potential outliers were interrogated and excluded from data analysis if the values were determined to be erroneous. Data points were determined to be erroneous if they were <0.5× or >2× the next closest comparable data point (e.g., a potential erroneous milk fat percent value would be compared with the cow's other 3 milk samples from that week).
Preplanned contrasts were used to analyze differences between treatments, including CTL versus both RPC products at the recommended dose and CTL versus RPC2 HD•RD . Means were compared when P ≤ 0.10 for main effects using the SLICE statement in SAS and Tukey-Kramer studentized adjustment. Evidence in mean comparison tests was considered significant if P ≤ 0.05 and a tendency when 0.05 < P ≤ 0.10. Data are presented as least squares means, back-transformed when applicable, with 95% confidence interval limits [least squares means (lower confidence limit, upper confidence limit)].

Prepartum DMI, BW, BCS, and Feeding Behavior
Prepartum DMI, BW, BCS, and feeding behavior results are presented in Table 3. During the prepartum period, feeding any RPC treatment reduced (P ≤ 0.01) DMI, DMI as percent of BW, DMI as percent of mBW, and NE L intake as a percent of requirement compared with CTL. Feeding any RPC treatment reduced (P ≤ 0.04) the average meal size compared with CTL. No evidence was seen for an effect of treatment (P ≥ 0.11) on total eating time or meal length. No evidence was seen for an effect of treatment (P ≥ 0.15) on prepartum BW or BCS.

Milk Production and Composition, BW, and BCS
Supplementation Period. Milk production-related variables, BW, and BCS are presented in Table 4. Both RPC1 RD•RD and RPC2 RD•RD increased (P = 0.04) the yield of colostrum compared with CTL. The RPC2 HD•RD treatment tended to increase (P = 0.10) Brix percent compared with RPC1 RD•RD . During the SP, no evidence was found for an effect of treatment (P ≥ 0.54) on milk, ECM, or FCM yields. For yields or concentrations of milk fat, protein, lactose, and FA groups, the only evidence for a treatment effect was that RPC2 HD•RD tended to reduce (P = 0.10) lactose percent compared with CTL. The RPC1 RD•RD and RPC2 RD•RD treatments tended to reduce (P = 0.07) MUN concentration compared with CTL. During the first week of lactation, RPC2 RD•RD increased (P < 0.01) and RPC2 HD•RD tended to increase (P = 0.08) SCC compared with CTL, and RPC1 RD•RD tended to increase (P = 0.07; Figure 1) SCC compared with CTL during the third week of lactation. No evidence was found for an effect of treatment (P ≥ 0.21) on BW or BCS.
Postsupplementation Period. Milk productionrelated variables, BW, and BCS are presented in Table 4. During the postSP, no evidence was seen for an effect of treatment (P ≥ 0.42) on milk yield, but the RPC1 RD•RD and RPC2 RD•RD treatments tended to increase (P ≤ 0.08) ECM and FCM yields compared with CTL. The RPC1 RD•RD and RPC2 RD•RD treatments also tended to increase (P ≤ 0.10) fat yield and protein percent and yield compared with CTL. The RPC2 RD•RD treatment tended to increase (P = 0.10) lactose yield compared with RPC2 HD•RD . No evidence was found of a treatment effect (P ≥ 0.15) on percent of milk fat or lactose. The RPC1 RD•RD and RPC2 RD•RD treatments tended to increase (P ≤ 0.08) the de novo and mixed proportions of total milk FA and reduce (P = 0.05) the preformed proportion compared with CTL. Supplementing RPC2 HD•RD increased (P = 0.02) the de novo and tended to increase (P = 0.06) the mixed proportions  Means with different letters within the same row are significantly different (P ≤ 0.05). of total milk FA but reduce (P = 0.03) the preformed proportion compared with CTL. The RPC1 RD•RD and RPC2 RD•RD treatments increased (P ≤ 0.04) de novo and mixed-milk FA yields compared with CTL, and the RPC2 HD•RD treatment increased (P = 0.04) de novo and tended to increase (P = 0.07) mixed-milk FA yields compared with CTL. No evidence was found for an effect of treatment (P ≥ 0.70) on preformed milk FA yields, and no evidence for an effect of treatment (P ≥ 0.18) on MUN. Feeding RPC1 RD•RD increased (P ≤ 0.05) SCC compared with all other treatments. No evidence was found for an effect of treatment on BW or BCS (P ≥ 0.11).

Prepartum Blood Biomarkers
Prepartum blood biomarker concentrations quantified repeatedly are presented in Table 5. No evidence was found for an effect of treatment (P ≥ 0.22) on blood FA, BUN, albumin, AST, ALT, or AST: ALT. The RPC2 HD•RD treatment tended to reduce (P = 0.08) plasma glucose compared with CTL and increased (P = 0.05) plasma BHB compared with RPC1 RD•RD and CTL. The effect of treatment on plasma TG interacted with time (P < 0.01), as shown in Figure 2.
Prepartum blood biomarker concentrations quantified at a single time point (−6.6 ± 0.6 DRTC, mean ± SD) are presented in Table 6. No evidence was found for an effect of treatment (P ≥ 0.11) on blood insulin, choline, methionine, betaine, or glucose: insulin. All RPC treatments reduced (P ≤ 0.02) prepartum blood RQUICKI compared with CTL. The RPC2 RD•RD treatment increased (P = 0.03) serum DMG compared with CTL and tended to increase (P = 0.06) DMG compared with RPC1 RD•RD . The RPC2 HD•RD treatment increased (P = 0.04) serum TMAO compared with CTL.

Postpartum Blood Biomarkers
No evidence was seen for an effect of treatment (P ≥ 0.48; Table 5) on maximum plasma FA concentration; however, the RPC2 HD•RD treatment tended to increase (P = 0.07; Table 5) maximum plasma BHB concentration compared with CTL.
Supplementation Period. Postpartum blood biomarker concentrations quantified repeatedly throughout the SP are presented in Table 5. No evidence was found for an effect of treatment (P ≥ 0.15) on blood glucose, TG, albumin, AST, ALT, or AST: ALT. The RPC2 HD•RD treatment tended to increase (P = 0.09) plasma FA and BHB compared with CTL. The RPC1 RD•RD and RPC2 RD•RD treatments reduced (P = 0.02) plasma BUN compared with CTL.
Postpartum blood biomarkers concentrations, quantified at limited time points during the SP, are presented in Table 6. No evidence was found for an effect of treatment (P ≥ 0.15) on blood insulin, choline, methionine, DMG, TMAO, or glucose: insulin. All RPC treatments reduced (P ≤ 0.03) blood RQUICKI. The RPC2 HD•RD treatment reduced (P ≤ 0.02) serum LBP compared with CTL and RPC2 RD•RD ; in addition, RPC1 RD•RD reduced (P = 0.05) serum LBP compared with RPC2 RD•RD . The RPC2 HD•RD treatment increased (P = 0.02) serum betaine compared with RPC1 RD•RD .
Postsupplementation Period. Blood biomarker concentrations postSP are presented in Table 5. No evidence was found of a treatment effect (P ≥ 0.21) on plasma FA. The RPC2 RD•RD treatment increased (P ≤ 0.03) plasma glucose compared with all other treatments. The RPC2 HD•RD treatment increased (P = 0.03) plasma BHB compared with CTL.

Disease Incidences
The incidences of subclinical and clinical diseases are presented in Supplemental Table S1 (https: / / doi .org/ 10 .17632/ rfg6txg9mh .1). No evidence was seen of a treatment effect (P ≥ 0.20) on subclinical hyperketonemia, displaced abomasum, or clinical mastitis. No evidence was found of a treatment effect (P ≥ 0.70) on subclinical mastitis during the postpartum SP, but the RPC1 RD•RD treatment tended to increase (P = 0.09; 18.5% vs. 52.1% for CTL vs. RPC1 RD•RD , respectively) the incidence of subclinical mastitis compared with CTL.

DISCUSSION
The objective of this experiment was to determine the effects of increasing the dose of prepartum RPC supplementation on milk production and blood biomarkers. Based on a previous literature review (Arshad et al., 2020), we hypothesized that increasing the dose of RPC during the prepartum period would increase postpartum milk production.

Effects of RPC Supplementation on Milk Production
Cows supplemented with the recommended dose of RPC1 or RPC2 during the prepartum period had improved yields of colostrum compared with CTL, without any apparent decline in quality as assessed by Brix percentage. Positive effects of prepartum RPC supplementation have been previously reported on colostrum yield (Swartz et al., 2022) and quality (Zenobi et al., 2018a). Within the current study, no evidence was found Means with different letters within the same row are significantly different (P ≤ 0.05).

a,b
Different italic letters in the same row indicate that means tended to differ (0.05 < P ≤ 0.10). that treatments altered milk or component-corrected milk yields during the SP; however, the RPC1 RD•RD and RPC2 RD•RD treatments tended to increase ECM by 2 kg/d compared with CTL during the postSP. Although not occurring during the SP as previous studies report, the ECM response observed in this experiment is similar to the effect of supplementing 12.9 g of choline ion, estimated by Arshad et al. (2020) in a meta-analysis. This meta-analysis also suggested that increasing the RPC supplementation rate could further increase milk production (Arshad et al., 2020), but that was not observed in this study. It is noteworthy that cows in the current experiment, on average, produced approximately 30% more milk yield than the average in the most recent RPC meta-analysis (Arshad et al., 2020). For further context, milk yield in the first 100 d of lactation was greater than the 80th percentile of upper Midwest US Holstein cows (49 kg/d 80th percentile vs. 53 kg/d current study average; AgSource, 2022). It is possible that the delayed treatment response on ECM observed in this study, compared with past studies, could be due to the greater basal production levels observed. Nevertheless, the observation of a 2 kg/d ECM advantage with RPC supplementation is consistent across studies ranging in basal production levels. Milk fat and protein yields, along with milk protein content, tended to increase with RPC supplementation at the recommended dose, driving the tendencies for increased FCM and ECM yields, which is consistent with a summary of previous studies (Arshad et al., 2020), although in the current study the effect was not observed until the postSP. Feeding any RPC treatment at least tended to increase the de novo and mixed proportion of total milk FA and reduce the preformed proportion during the postSP. Effects of RPC on monoand polyunsaturated FA have been previously observed (Goselink et al., 2013), but the evidence obtained in this study of an effect of RPC supplementation on the proportional contribution of different FA groups to overall milk fat is novel. The effects of RPC on milk fat are thought to be through improved packaging and transport of mobilized or ingested FA (McFadden et al., 2020), but differences in de novo FA groups here suggest that RPC may also influence milk fat synthesis, potentially through increased supply of nutrients such as acetate or BHB to the mammary gland or through direct regulation of molecular pathways related to milk FA synthesis. As RPC supplementation often increases postpartum DMI, increased nutrient supply of one or more of these metabolites to the mammary gland is possible (Urrutia and Harvatine, 2017;Nichols et al., 2020) and could help explain the tendency for improved milk component production (i.e., fat and protein). Although mean blood BHB concentrations were well below the hyperketonemic threshold across the study, it is interesting that the RPC2 HD•RD treatment increased plasma BHB concentrations at +56 and +100 DRTC. The increased BHB with RPC supplementation (+0.06 mM; RPC2 HD•RD compared with CTL) may represent increased precursor availability for de novo milk FA synthesis, which is consistent with the proportional shift to greater de novo milk FA, without a decrease in preformed FA yield. However, we cannot determine whether the effects on de novo milk FA synthesis are related to altered DMI, as postpartum intakes are not available in the current experiment. The effects of choline supplementation on de novo milk fat production should be verified directly via gas chromatography analysis of milk fat and further explored to understand the potential role of choline to influence mammary de novo milk fat synthesis.
Unexpectedly, the RPC treatments at least tended to increase early-lactation SCC compared with CTL, although average SCC across treatment groups remained below 100,000 cells/mL. All cows in this study were supplemented with rumen-protected methionine. In a recent meta-analysis, an interaction was estimated between the dose of RPC and prepartum methionine (Arshad et al., 2020) where RPC increased SCC at higher levels of prepartum methionine as percent of MP. In primiparous cows, feeding both RPC and rumenprotected methionine has been found to increase SCC compared with methionine alone (Potts et al., 2020), but the same effect was not observed in multiparous cows. Although not significant, a similar pattern was observed when peripartum supplementation of either rumen-protected methionine or RPC numerically reduced SCC, whereas feeding them together numerically increased SCC compared with control (Zhou et al., 2016b). Given the general benefits of supplementing rumen-protected methionine and RPC, it is surprising to observe the interaction of the 2 increasing SCC. Although no evidence was seen of increased clinical disease, it is an observation that deserves further exploration.

Effects of RPC Supplementation on Prepartum Feed Intake and Feeding Behavior
The postpartum lactational effects of RPC are usually accompanied by an increase in prepartum and postpartum DMI (Arshad et al., 2020). Surprisingly, supplementation of any RPC treatment in the current study reduced prepartum DMI compared with CTL, which is inconsistent with the meta-analysis that previously estimated a small positive effect of RPC on prepartum DMI across past top-dress studies (Arshad et al., 2020). Cows fed moderate-to-high energy prefresh diets will consume >140% of NE L requirement if not limited by the physical effects of the diet (Drackley and Guretzky, 2007;Janovick et al., 2011). Cows in the CTL group, on average, consumed 122% of their calculated NE L requirement, suggesting that the physical effects of the diet were unable to sufficiently control feed intake. Despite being fed a diet with the same physical aspects of the CTL diet, RPC-supplemented cows had decreased DMI, which resulted in average energy intake of 105% to 109% of the calculated NE L requirement. The lack of postpartum individual cow DMI data precludes full analysis of feed efficiency; however, the decreased prepartum DMI with subsequent tendencies for greater ECM is intriguing and warrants continued interrogation of nutrient use efficiency with RPC supplementation.
The use of RIC feeders allowed for tracking of prepartum meal length and size, which may help explain the effects of RPC on feed intake. We did not observe evidence that treatments altered total daily eating time or individual meal length, but cows supplemented with any RPC treatment had reduced meal size compared with CTL. Reduced DMI could be due to greater sensitivity to the physical effects of the diet or a response to metabolic feedback signals associated with satiety in RPC-supplemented cows. Given that all diets possessed the same physical characteristics, the latter is a more plausible explanation, especially given that the RPC was consumed over the course of the day in the current study, rather than as a bolus top-dress. Competition at the feed bunk could also affect feeding behavior, but no evidence was found that average daily prepartum stocking density differed by treatment (P = 0.38; 1.65, 1.75, 1.78, 1.71 ± 0.06 cows/RIC feeder for CTL, RPC1 RD•RD , RPC2 RD•RD , and RPC2 HD•RD , respectively). It is also worth noting that each feeder was used for 3.6 ± 1.3 h/d on average, suggesting that cows had adequate opportunity to access feed. The novel effect of RPC on prepartum feeding behavior deserves further exploration and could contribute to a better understanding of how RPC influences DMI.

Effects of RPC Supplementation on Potential Biomarkers of Choline Absorption and Metabolism
Assessing the bioavailability of choline has proven challenging given the many different choline derivatives that can be found in the body. Concentrations of choline derivatives in blood and milk have been suggested to be potential markers of RPC bioavailability, specifically betaine (de Veth et al., 2016). We did not observe evidence of treatment effects on serum choline or betaine  Means with different letters within the same row are significantly different (P ≤ 0.05). 2 FA = fatty acids; AST = aspartate aminotransferase; ALT = alanine aminotransferase. SP = postcalving, during RPC supplementation period [calving to +23 ± 2.6 d relative to calving (DRTC)]; all variables were quantified at +3, +7, +14, and +21 DRTC, except BUN, which was not quantified at +21 DRTC; glucose and FA were also quantified at +5 DRTC; and BHB was also quantified at +5, +11, and +18 DRTC. PostSP = postcalving, after RPC supplementation period when cows are fed a common lactating diet; variables quantified at +56 and +100 DRTC. 3 All prepartum variables were quantified at −14, −7, and −1 DRTC, and glucose, FA, and BHB were also quantified at −5 and −3 DRTC.
concentrations during prepartum or the SP. Prepartum, TMAO concentration was increased with RPC2 HD•RD supplementation compared with CTL. Concentrations of TMAO have been suggested as a potential marker for choline bioavailability (France et al., 2022), and building a literature base of concentrations will aid in determining robustness and sensitivity of this marker. The variation in responsiveness of different potential biomarkers further validates the need for future work to interrogate choline derivatives as biomarkers of choline bioavailability. It has been hypothesized that increasing choline supply in vivo could benefit methionine status as its methyl donation ability allows choline to contribute to regeneration of methionine from homocysteine (McFadden et al., 2020), which is supported by in vitro work in primary bovine hepatocytes (Chandler and White, 2017). During the prepartum period, the RPC2 RD•RD treatment increased the concentration of DMG in blood, the residual choline derivative after methyl group donation through betaine in the homocysteine-methionine regeneration pathway. Although we did not observe evidence for treatment effects on serum methionine concentration, circulating DMG concentrations are an intriguing potential biomarker that could reflect methionine regeneration. Previously, in vivo RPC supplementation did not affect expression of enzymes associated with methionine regeneration (Zhou et al., 2017), but together with prior in vitro work (Chandler and White, 2017), these data support further interrogation of the potential for choline to affect methionine regeneration.

Effects of RPC Supplementation on Metabolism
Energy-related blood metabolite concentrations serve as biomarkers of energy status and can provide insight regarding the match between dietary energy and energy requirements. The reduction in circulating TG concentration at −7 DRTC with RPC2 RD•RD treatment compared with CTL may have reflected the reduced DMI observed with RPC2 RD•RD , although the difference in TG was diminished by −1 DRTC. In addition, RQUICKI was reduced by any prepartum RPC treatment that could represent reduced insulin sensitivity (Holtenius and Holtenius, 2007); however, considering the RQUICKI equation, the change could primarily reflect increased insulin concentrations. Regardless, changes in the relative balance of energetic hormones and metabolites could contribute to or reflect differences in DMI and feeding behavior and should be further explored in future studies.
Cows in this experiment exhibited the hallmarks of peripartum metabolite changes in response to negative energy balance, namely elevated plasma FA and BHB concentrations paired with reduced glucose, immediately postpartum. The incidence of subclinical hyperketonemia (blood BHB ≥1.2 mM) was low (23.6%; Supplemental Table S1) compared with previous reports of Midwest dairy herds (McArt et al., 2011), which together with blood FA concentrations supports that, overall, cows in this experiment were not under excessive negative energy balance. Despite this, the RPC2 HD•RD treatment tended to increase early-lactation plasma FA and BHB concentrations. The tendency for increased biomarkers of negative energy balance when comparing RPC2 HD•RD and CTL is worth noting but may not be biologically impactful given that both remained below established thresholds for subclinical disease.
Both RPC1 RD•RD and RPC2 RD•RD reduced BUN and tended to reduce MUN during the early-lactation SP compared with CTL. These responses potentially represent a reduced need for nitrogen waste excretion (Lavery and Ferris, 2021). Previously, peripartum RPC supplementation reduced 3-methylhistidine, a marker of muscle protein mobilization, in early-lactation cows (Zhou et al., 2016a). Reduced muscle catabolism could represent the potential for improved postpartum AA balance with RPC supplementation. These results may represent the potential for peripartum RPC supplementation to help spare limited nutrients, such as AA, from catabolism during the early lactation period.  Means with different letters within the same row are significantly different (P ≤ 0.05).

a,b
Different italic letters in the same row indicate that means tended to differ (0.05 < P ≤ 0.10).

Effects of RPC Supplementation on Hepatic and Gut Function
During the early lactation period, the liver is under metabolic stress to support increased glucose demands, which may be exacerbated by systemic inflammation and its obligatory glucose use (Kvidera et al., 2017). One cause of systemic inflammation during early lactation is likely from endotoxemia as stressors induced by psychological and diet changes may negatively affect gastrointestinal integrity (Meddings and Swain, 2000;Bradford et al., 2015;Gott et al., 2015). Elevated circulating LBP, an LPS-specific acute-phase protein (Lu et al., 2008), is often observed with endotoxemia and is associated with reduced milk yield (Abuajamieh et al., 2016;Kvidera et al., 2017). Across treatments, we observed elevated (P < 0.01) circulating LBP concentrations at +3 DRTC [3.05 (2.85, 3.27) versus 3.73 (3.47, 4.00) mg/L, −21 versus +3 DRTC] but not at +7 DRTC [3.05 (2.85, 3.27) versus 3.00 (2.80, 3.22) mg/L, −21 versus +7 DRTC; (P = 0.92)] compared with −21 DRTC. The RPC2 HD•RD treatment tended to reduce postpartum circulating LBP concentration compared with CTL; however, no evidence was found of any notable benefits to liver function, as measured by circulating albumin, AST, and ALT, nor was there evidence of an effect on milk production during early lactation.

CONCLUSIONS
Consistent with past research, peripartum RPC supplementation (~12 g/d choline ion) with either RPC product at 0.10 choline ion, as a percent of prepartum diet DM, tended to improve ECM even in higher-producing dairy cows, although the response was observed during the postSP, rather than in the 3 weeks immediately following calving as observed in previous studies. No evidence was found for an additional benefit on milk production of increasing the prepartum dose of choline ion from 0.10 to 0.13 choline ion, as a percent of diet DM. Surprisingly, RPC supplementation decreased prepartum DMI compared with CTL, despite postSP tendencies for increased ECM. The effects of RPC supplementation on metabolic (e.g., BUN and RQUICKI) and inflammatory (e.g., LBP) biomarkers support the potential for RPC supplementation to affect transition cow metabolism and health and may support the production benefits observed. Additional research is still needed to fully understand the effect of RPC on prepartum and postpartum DMI when mixed into the TMR and the potential effects on nutrient partitioning and feed efficiency.